mbfinotti/advertising-skillsContrôle réussi
SKILL DETAIL
ad-platform-selection
mbfinotti/advertising-skills/ad-platform-selection
Decide which paid channel families fit a business's economics, audience, funnel stage, and budget at campaign setup - paid search, paid social, short-form video, connected TV, native/discovery, retail media, programmatic display, podcast/audio, paid review listings, B2B professional networks - including whether each is affordable against allowable CAC and payback. Use whenever the user asks where to spend a new ad budget, which platform to advertise on, how to build a paid channel mix, whether a channel is worth trying, or compares two ad channels - even if they never say 'channel selection'. Covers B2B and B2C. Do NOT use to split an ongoing budget across existing campaigns - use mbfinotti/advertising-skills@ad-spend-allocation instead.
Installations · 181Voir la source
Installation
npx skills add https://github.com/mbfinotti/advertising-skills --skill ad-platform-selection
Fichiers du skill
SKILL.md
Dernière synchronisation · 24 sept. 2026
evals/evals.json›
{
"skill_name": "ad-platform-selection",
"evals": [
{
"id": 1,
"prompt": "I run Loamly, a B2B inventory-forecasting SaaS. Two plans: Starter at $19/month and Scale at $549/month. Blended ARPU is $74, gross margin 82%. Our LTV:CAC is 3.2:1 so my board says we're safe to pour money into paid. Annual logo retention on Starter is about 45%, on Scale it's 92%. An agency quoted us roughly $200 to acquire a customer. Budget is $15K/month. Which ad platform should we go with?",
"expected_output": "A channel plan that splits the economics per plan instead of using blended ARPU, demotes LTV:CAC to a secondary check, routes the Starter tier away from paid, anchors paid acquisition to the Scale tier via payback math, and names one funded primary channel family.",
"files": [],
"expectations": [
"Response computes payback separately for the $19 plan and the $549 plan rather than on the $74 blended ARPU.",
"Response shows that a ~$200 CAC against the Starter plan pays back in roughly a year or more (about $15.60/month gross profit), making that tier unaffordable for paid acquisition.",
"Response recommends routing the Starter tier to organic or product-led motions instead of paid ads.",
"Response shows the Scale plan's monthly gross profit around $450 and a payback well under one month at a $200 CAC, anchoring paid acquisition to that tier.",
"Response applies a discounted payback calculation (CAC divided by monthly gross profit times retention rate) using the stated 45% and/or 92% retention figures.",
"Response treats payback period, not LTV:CAC, as the primary affordability gate.",
"Response explicitly challenges the 3.2:1 LTV:CAC ratio, naming at least one structural flaw (blended ARPU hides per-plan variance, churn front-loads, best customers compound, or revenue-delay cash gap).",
"If LTV:CAC is used at all, it appears only as a secondary sanity check after payback clears.",
"Response does not present a single blended payback or blended CAC number as the decision basis.",
"Response derives an allowable CAC from a payback target rather than accepting the agency's $200 quote as the ceiling.",
"Response references a target payback band of roughly 3-12 months (with a longer allowance only for enterprise motions).",
"Recommendation names exactly one funded primary channel family, not a list of platforms to try in parallel.",
"Response works in channel-family vocabulary (paid search, paid social, B2B professional networks) rather than framing the decision inside one vendor's product names.",
"Response gathers or conditions on missing economics inputs (sales cycle, buyer concentration, existing channels) before or alongside naming the primary, rather than picking a platform from price alone."
]
},
{
"id": 2,
"prompt": "We sell reef-safe sunscreen DTC at Kelpwild. AOV $42, gross margin 58%. We have $12K/month for ads. My plan: $2,400 each into Meta, TikTok, Google Search, Pinterest and a streaming-TV test, and we'll kill anything that isn't hitting 4x ROAS after two weeks. Sound right?",
"expected_output": "A rejection of the five-way split and the 4x ROAS target, with break-even ROAS derived from margin, one funded primary at ~80% concentration, learning-threshold arithmetic, a proper test design with pass/kill criteria, and connected TV deleted on the contribution math.",
"files": [],
"expectations": [
"Response rejects the equal five-way split of the budget.",
"Response shows the arithmetic that $2,400/month per channel is roughly $80/day, below the learning thresholds of every major channel family.",
"Response cites a concrete learning threshold as the reason the split cannot produce signal (e.g., ~50 optimization events per ad set per 7 days on paid social).",
"Response recommends concentrating roughly 80% of budget on one funded primary channel for the first ~90 days.",
"Response computes break-even ROAS as 1 divided by the contribution margin rate, landing near 1.7x for a 58% margin.",
"Response rejects the 4x ROAS target as a folk number not derived from this business's margin.",
"Response supports the ROAS correction with the observation that median paid-social ROAS in large ecommerce datasets sits near 1.86-1.93x, or that paid social is profitable only above roughly a 50% margin.",
"Response flags the two-week kill window as shorter than a valid test and a source of false negatives.",
"Kill criterion is restated as roughly 2-3x target CPA spent with zero conversions after a readable sample, not a ROAS shortfall at two weeks.",
"Pass criterion is CPA at or under allowable CAC at statistical significance (90-95% confidence), not a fixed ROAS multiple.",
"Response prescribes a deliberately tiny price-discovery spend (on the order of $100) to learn real CPC/CPM before the real test budget moves.",
"Second channel is added only on an explicit trigger (primary saturation or a budget breakpoint), one channel at a time.",
"Connected TV is ruled out for a ~$24-contribution product, with the CPM plus production cost math named.",
"Channels removed by a gate are named in a deleted list with the gate each failed, not kept as low-priority rows to revisit."
]
},
{
"id": 3,
"prompt": "I'm head of growth at BrightHarbor, we do personal loans in the US. Planning our first Meta campaign: upload our best-customer list, build a 1% lookalike, target 25-45 year old homeowners in specific high-income ZIP codes around Austin, and exclude renters. Also thinking about Google. Budget $40K/month. What's the right platform mix?",
"expected_output": "A plan that identifies the offer as a regulated special ad category, explains each requested targeting capability that is stripped (age lock, ZIP removal, lookalike replacement, exclusion removal), rescores paid social's reachability accordingly, and rebuilds the mix from the affordability gate instead of optimizing the illegal plan.",
"files": [],
"expectations": [
"Response identifies personal loans as a regulated offer (credit / financial products) triggering a restricted-targeting regime on the major social platform.",
"Response notes that financial products and services require special-ad-category declaration in the US (a regime in force since January 2025).",
"Response states age targeting gets locked to 18-65+, so the requested 25-45 narrowing is unavailable.",
"Response states ZIP-level precision is removed and a minimum geographic radius (roughly 15 miles) applies, so the high-income-ZIP plan is unavailable.",
"Response states lookalike audiences are replaced by a weaker special-audience tool, so the 1% lookalike plan is unavailable as described.",
"Response states detailed-targeting exclusions (such as excluding renters) and demographic narrowing are stripped under the declaration.",
"Response does not proceed to optimize or build the requested lookalike/ZIP targeting plan as-is.",
"Response scores paid social's targeting reachability down (or disqualifies it) specifically because the plan depended on the stripped capabilities.",
"Response evaluates paid search / demand capture explicitly for a category buyers actively search, rather than treating social as the default.",
"Response advises checking each candidate family's current policy for this exact offer and geography before scoring, noting policy pre-approval can add lead time.",
"Response runs an affordability step (contribution economics, allowable CAC, payback) before or alongside the platform mix rather than jumping straight to channel names.",
"Output includes a deleted-or-weakened list naming which rule hit which channel family.",
"Response asks about or accounts for regulated-category exposure across every target geography, not just the one platform mentioned."
]
},
{
"id": 4,
"prompt": "Solo founder of Notchset, a $29/month Chrome extension for recruiters (B2B). I ran LinkedIn ads at my last job so I know the platform inside out, and recruiters obviously live on LinkedIn. Planning $3,500/month all into LinkedIn lead gen forms. Average customer stays about 14 months. Is LinkedIn the right call?",
"expected_output": "A refusal grounded in arithmetic: the ~$400 LTV misses the professional-network LTV floor by an order of magnitude, $3,500 sits under the monthly floor, typical CPLs alone exceed what the product can afford, team familiarity is named as a failure mode, and the family is deleted with an affordable alternative recommended.",
"files": [],
"expectations": [
"Response rules out B2B professional networks on the LTV floor (roughly $10,000-$15,000 minimum), showing Notchset's LTV is on the order of $400.",
"Response notes the roughly $5,000/month floor for pipeline-oriented professional-network campaigns in North America, which $3,500 does not clear.",
"Response notes that a few hundred dollars a month on this family only proves engagement signal, never sales outcomes, so a smaller trial does not rescue the plan.",
"Response cites typical professional-network costs (CPL around $110, or CPC $8-16) against what a $29/month product can afford, with the arithmetic shown.",
"Response names picking the channel the team already knows as a failure mode, treating familiarity as a scoring input rather than the decision.",
"Response states that audience presence (recruiters live on LinkedIn) alone does not pass the affordability gate; economics decide what is affordable.",
"The family is deleted from the candidate set and named in a deleted list with the gate it failed, not ranked at the bottom for later.",
"Response recommends an alternative consistent with low-ACV economics (paid search capture if search volume exists, or organic/product-led motions).",
"Response computes or requests contribution margin to derive allowable CAC from price, rather than skipping the economics.",
"Allowable CAC is derived via a payback target, not a folk ROAS multiple.",
"Response does not recommend testing LinkedIn with a smaller budget to see what happens.",
"Response notes a professional-network test conventionally needs about a quarter of committed spend before a verdict, compounding the mismatch.",
"Final output names one funded primary channel with a test design carrying explicit pass/fail criteria."
]
},
{
"id": 5,
"prompt": "We're Ferrostat, selling industrial IoT monitoring, ACV $60K, sales cycle around 4 months. Ran LinkedIn ads for 3 weeks in July - $3K spent, 2 MQLs, zero pipeline, so we killed it. Google Ads shows 9 of our last 11 closed-won deals as last-click, so we're moving the whole $20K/month into search. Sanity check?",
"expected_output": "A response that voids the 3-week kill as a false negative (test shorter than the 4-month cycle, spend under the floor), explains last-click's structural bias toward capture channels, refuses to ratify the dashboard-driven reallocation, and prescribes self-reported attribution plus a full-cycle re-test.",
"files": [],
"expectations": [
"Response flags the 3-week LinkedIn test as shorter than the ~4-month sales cycle and therefore a false negative, not a verdict.",
"Response notes B2B professional networks conventionally warrant roughly a 3-month committed test before judging.",
"Response notes the spend pace (about $3K over three weeks) sits below the roughly $5,000/month floor for sales outcomes on that family.",
"Response refuses to accept the July kill as valid evidence and calls for a diagnose-then-rerun over a full cycle before any conclusion about the channel.",
"Response explains that last-click attribution structurally over-credits capture channels like search and under-credits creation channels.",
"Response cites that social/creation channels are systematically under-reported in platform and last-click reads (practitioner estimates on the order of 70% under-reporting).",
"Response warns against making the channel decision from the platform dashboard read alone.",
"Response recommends adding self-reported attribution (a 'how did you hear about us' field on the demo form) as the cheap correction.",
"Response states last-click breaks over long sales cycles and prescribes an attribution approach fit for them (self-reported plus modeled, or first- and last-touch side by side).",
"Response does not endorse moving 100% of budget into search purely on the 9-of-11 last-click statistic.",
"Response distinguishes 'search may legitimately be funded first if category search volume exists and capture is underfunded' from 'the dashboard proves it' - the decision is re-derived from economics and demand state.",
"Any prescribed re-test carries a duration of at least one full sales cycle (about a quarter).",
"Response notes the kill rule (2-3x target CPA spent with zero conversions after a readable sample) was not met - 2 MQLs is not zero signal - or that a kill verdict feeds the next-ranked candidate rather than a reallocation of everything."
]
},
{
"id": 6,
"prompt": "Vantabyte makes an on-call incident tool for SREs, $120/month, 85% margin, sold self-serve. Nobody searches for our category (we invented it), and our buyers all listen to the same two or three big SRE/devops podcasts - we surveyed 200 signups and podcasts were the top answer. Budget is $9K/month, no deadline pressure, this is a year-long build. I want to put it all into podcast sponsorships. Which shows and how should we structure it?",
"expected_output": "A plan that refuses podcast as primary because $9K sits under the ~$15K measurability floor, names it deleted with the budget that would clear it, walks down the ranking to a fundable creation family (paid social prospecting), and registers podcast as a later candidate with an explicit re-entry trigger.",
"files": [],
"expectations": [
"Response states the podcast measurability floor (roughly $15,000+/month spread across 3+ shows in multi-episode flights) and that $9K sits below it.",
"Response does not fund podcast as primary at $9K, including via a shrunk one-show version - an underfunded channel produces no readable signal.",
"Podcast is named in a deleted list with the floor it missed and the budget level that would clear it.",
"Response walks down to the next-ranked fundable channel rather than spreading the $9K thinner or across several small bets.",
"Response engages the promotion-against-rank conditions: buyer-on-surface evidence and no-capture-family both hold, but the promotion is refused specifically because the budget cannot clear the floor.",
"Response recognizes that with no category search volume, capture cannot be the primary and a demand-creation family must be.",
"Response recommends a creation family fundable at $9K (paid social prospecting or short-form video) as the primary.",
"Response checks the chosen family's learning threshold against the budget (roughly 50 optimization events per ad set per 7 days, with the weekly spend implication worked for a stated target CPA).",
"Podcast is registered as a later candidate with an explicit re-entry condition tied to budget, not silently dropped.",
"Any future podcast plan described includes measurement designed before booking (promo codes, vanity URLs, self-reported attribution) and show selection above a minimum download threshold.",
"The survey evidence is used to justify the audience-presence score but is not allowed to override the funding floor.",
"Measurement plan includes self-reported attribution because a creation channel anchors the mix.",
"Plan concentrates roughly 80% of budget on the single funded primary."
]
},
{
"id": 7,
"prompt": "Just launched Plumecraft - handmade lighting, AOV $310, 62% margin. The site gets about 900 visits a month right now. A growth consultant told me retargeting is the highest-ROI channel so we should start there, and to set up geo-lift holdout testing from day one so our measurement is bulletproof. Budget $6K/month. How do we set this up?",
"expected_output": "A plan that rejects retargeting as primary (no traffic pool to multiply), explains why its apparent ROI is an artifact of near-zero effort on traffic already bought, picks a prospecting/capture primary via the economics, and starts measurement at UTM discipline plus last-non-direct instead of day-one geo-lift.",
"files": [],
"expectations": [
"Response rejects retargeting as the primary channel: it multiplies traffic already being bought and its ceiling is capped by whatever feeds it.",
"Response notes ~900 visits/month is far too small a pool for retargeting to produce meaningful volume or signal.",
"Response explains retargeting's high apparent ROI comes from near-zero effort reusing existing traffic and creative, while it ranks last on value - not because it creates demand.",
"Response notes retargeting only takes effect once a paid traffic pool has accumulated, so it cannot lead a cold start.",
"Response recommends a prospecting or capture primary that generates the traffic first, chosen through the economics and demand-state logic rather than asserted.",
"Response rejects day-one incrementality/geo-lift testing as the heaviest measurement rung, reserved for the biggest line item once spend justifies withholding some of it.",
"Response prescribes the starting measurement rung for a small-budget plan: UTM discipline plus last-non-direct attribution.",
"Self-reported attribution is added only when or if a demand-creation channel enters the mix.",
"Measurement rungs are added one at a time on explicit triggers, never all at once.",
"Response computes the affordability inputs: contribution near $192 per sale and break-even ROAS near 1.6x for a 62% margin.",
"Retargeting is positioned as a later add-on mode with an explicit trigger (an accumulated traffic pool), not deleted for economics.",
"Plan funds one primary above its learning threshold with the $6K rather than splitting it.",
"The consultant's advice is explicitly challenged rather than implemented as-is."
]
},
{
"id": 8,
"prompt": "We make Snackle, better-for-you granola. AOV $38, 34% gross margin, $14K/month budget. Our biggest competitor Verdora is all over streaming TV and podcasts and they're crushing it - and I found a self-serve CTV platform with a $500/month minimum, so we can start cheap there this week and add a podcast later. Thoughts?",
"expected_output": "A plan that computes the ~$13 contribution, deletes connected TV on the CPM-plus-production math and podcast on the measurability floor, rejects the copy-the-competitor and cheapest-first logic by name, surfaces that a 34% margin sits below paid social's profitability line, and forces the repeat-purchase/LTV question before any paid recommendation.",
"files": [],
"expectations": [
"Response computes contribution per sale near $13 (34% of $38).",
"Connected TV is deleted for a sub-$50-contribution product, with the CPM range ($20-40) plus production cost math named.",
"Response rejects the cheap-entry logic: a $500/month self-serve tier does not make CTV efficient, because the ranking criterion is value per unit of effort, not cheapest to start.",
"Response names CTV's real cost as front-loaded: video production (roughly $1,500-$5,000 to repurpose, $15,000+ for a new spot) plus a measurement build due before any result.",
"Response notes CTV takes about a quarter before a readable result and its click-level measurement is structurally weak, requiring incrementality planning.",
"Response rejects copying Verdora's channel mix, naming the failure mode: the competitor's margin and LTV, not their taste, made their mix work.",
"Podcast is gated on its roughly $15K measurability floor relative to the $14K total budget.",
"Response runs the affordability gate before ranking any channel, deriving allowable CAC from contribution rather than jumping to channel names.",
"Response flags that first-order contribution of ~$13 sits below common benchmark CPAs (e.g., paid social median CPA near $38) and asks about repeat purchase or LTV before approving any paid channel.",
"Break-even ROAS computed near 2.9x for a 34% margin.",
"Response notes that at the ecommerce median paid-social ROAS of roughly 1.86-1.93x, a 34% margin is unprofitable - below the ~50% margin line.",
"Deleted channels are named with the gate each failed, not parked as start-cheap-and-see experiments.",
"The verdict is honest about the economics: either one primary consistent with the math, or a statement that paid does not clear until margin or repeat-purchase economics improve."
]
},
{
"id": 9,
"prompt": "Two-person team at Loopmill, B2B webinar software at $190/month, 78% margin. We can produce maybe 2 polished static graphics a month, no video. Everyone says TikTok and Reels ads are where the growth is, so we planned $7K/month split between TikTok and Meta Reels. When should we expect results?",
"expected_output": "A plan that disqualifies short-form video on creative capacity (10-20 variants, fatigue in days, static-only production), checks the professional-network LTV floor before offering it as an alternative, rejects the two-way split, and lands on a low-burn primary sized to a two-person team, with results timing tied to that family's time-to-effect.",
"files": [],
"expectations": [
"Short-form video is disqualified on creative capacity: it needs on the order of 10-20 creative variations per campaign with fatigue setting in within days.",
"Response notes static-only production is itself a disqualifier for short-form video.",
"Response notes paid social prospecting at scale needs 8-15+ meaningfully different variants per ad set with refreshes as frequency climbs, also beyond 2 statics/month.",
"Creative production capacity is named as the gate that fails more plans than media cost does.",
"Response recommends a lower-burn family or fixing production capacity first, rather than shrinking the video plan.",
"Before offering B2B professional networks as the alternative, the LTV floor (roughly $10,000-$15,000) is checked against what a $190/month product's retention implies.",
"The $7K two-platform split is rejected in favor of one funded primary.",
"The 'everyone says TikTok' rationale is treated as trend-following, not audience evidence; the response asks for or reasons from evidence of where webinar-software buyers actually concentrate.",
"Response applies capture-first logic: webinar software is a searched category, so paid search is evaluated as the capture primary.",
"Effort is framed as what the team spends (assets per month, hours, reversibility) and the two-person ceiling materially drives the ranking.",
"The 'when should we expect results' question is answered per the chosen family's time-to-effect (days from a cold start for paid search), not a generic timeline.",
"Response verifies or asks whether landing pages and conversion tracking are live before any launch.",
"Final output names one funded primary with explicit pass/fail and kill criteria."
]
},
{
"id": 10,
"prompt": "I coordinate campaigns for Tidewatch Alliance, a European ocean-conservation nonprofit. We have 25,000 euros for a 3-month awareness push across Germany, France and the Netherlands ahead of a fisheries policy vote in the EU Parliament. We ran Facebook and Google ads for this kind of campaign in 2023 and they worked well. Can you map out the same plan again?",
"expected_output": "A plan that classifies the campaign as EU political/social-issue advertising, states the major search and social platforms have exited that inventory entirely under the TTPA, refuses the 2023 rerun and any reclassification workaround, and rebuilds the push on families that still exist for this messaging with measurement fit for weak EU signal.",
"files": [],
"expectations": [
"Response identifies the campaign as political / social-issue advertising in the EU.",
"Response states the largest search and social platforms exited EU political, electoral, and social-issue advertising entirely (under the TTPA, from October 2025).",
"Response states the 2023 playbook cannot be rerun because those channel families no longer exist for this messaging in the EU - gone, not merely restricted.",
"Response notes advocacy and NGO communication is caught in the same net even when it is not party-political.",
"Response does not propose relabeling or reclassifying the ads as non-political to slip past platform review.",
"Response proposes alternatives outside the exited families (for example podcast/audio, native/discovery, direct media buys) with honest caveats, rather than pretending search and social remain available.",
"The disqualifier pass runs before any ranking, and removed families are named with the rule that removed them.",
"Response notes EU consent requirements degrade tracking signal (high consent-mode adoption but widespread non-compliant implementations) and plans measurement accordingly.",
"Response treats the regime as hard-blocking with serious penalties (fines up to a percentage of annual turnover), not a formality to risk.",
"The recommended awareness plan carries a measurement design suited to weak click-level reads (self-reported, incrementality-style, or equivalent).",
"Response advises reverifying current platform policy before spend commits, dating the check, given how fast these rules move."
]
},
{
"id": 11,
"prompt": "Solvenne does accounting automation for law firms - ACV $9,600, 75% margin, roughly a 6-week sales cycle. We're starting Google Ads next Monday. Plan: put in $500 total, watch it for 10 days, and if cost per demo comes in under $150 we'll scale to $10K/month. Our current CAC from outbound is $700 and pays back in about 6 weeks, and we want paid to match that. Also, what bid strategy should we use?",
"expected_output": "A test design rebuilt to the framework: a tiny price-discovery spend first (legal-vertical CPCs are the most expensive and widest-ranging), a real test at 3-5x target CPA per concept over at least the full 6-week cycle with same-weekday bounds, pass at significance against a payback-derived allowable CAC far above $700, the 6-week payback flagged as underinvestment, and bidding kept out of scope.",
"files": [],
"expectations": [
"Response prescribes a deliberately tiny price-discovery spend (on the order of $100) purely to learn the real CPC for this targeting before the real test, replacing published benchmarks with own numbers.",
"The real test is budgeted at a minimum of 3x target CPA per concept (5x preferred), not a flat $500.",
"Test duration is set to at least the full ~6-week purchase cycle, explicitly overriding the 10-day window.",
"The 10-day judgment window is rejected as a false-negative risk because it is shorter than the sales cycle.",
"The test is specified to start and end on the same weekday.",
"Pass criterion is CPA/CPL at or under allowable CAC at 90-95% statistical confidence, not an eyeballed $150-per-demo after 10 days.",
"Kill criterion is 2-3x target CPA spent with zero conversions after a readable sample, followed by diagnosis before rerunning.",
"Allowable CAC is derived from payback math on roughly $600/month gross profit per customer, yielding a ceiling well above $700 (on the order of $1,800-$7,200 for a 3-12 month band), with the numbers shown.",
"The ~6-week (about 1.4-month) outbound payback is identified as a signal of underinvestment, not health, because it sits under the 3-month lower bound.",
"Response does not adopt 'match outbound's $700 CAC' as the paid ceiling; the ceiling comes from the payback target.",
"Response applies the paid-search funding floor: budget must support roughly 10 clicks/day and a daily budget of 3-5x target CPA, delivering at least one conversion/day on paper - which $500 total cannot.",
"The cost-per-demo target is derived by carrying lead-to-customer conversion through to CAC, not left as a free-floating $150.",
"Response flags legal-vertical search CPCs as among the most expensive with a very wide published range, making own price discovery essential rather than trusting averages.",
"Response keeps bidding-method choice out of scope as a separate decision instead of answering it in depth."
]
}
],
"trigger_queries": [
{ "query": "Which ad platform should we advertise on first?", "should_trigger": true },
{ "query": "We're a B2B SaaS with $8k/month for ads - Google or LinkedIn?", "should_trigger": true },
{ "query": "Where should I spend my first paid marketing budget?", "should_trigger": true },
{ "query": "Is TikTok advertising worth trying for a skincare brand?", "should_trigger": true },
{ "query": "Help me build a paid channel mix for our DTC store.", "should_trigger": true },
{ "query": "Should we put our ad money into podcasts or paid search?", "should_trigger": true },
{ "query": "My cofounder wants Facebook ads, I want Google - settle this.", "should_trigger": true },
{ "query": "Does connected TV advertising make sense for a $60 product?", "should_trigger": true },
{ "query": "We've never run ads. Where do we even start?", "should_trigger": true },
{ "query": "What's the right advertising channel for an enterprise product with a 90-day sales cycle?", "should_trigger": true },
{ "query": "I have $3,000 a month to acquire customers with ads. What do I do with it?", "should_trigger": true },
{ "query": "Can a $19/month product afford paid acquisition at all?", "should_trigger": true },
{ "query": "Which paid channels fit a marketplace with a 30% take rate?", "should_trigger": true },
{ "query": "Evaluate whether LinkedIn ads are affordable for our ACV.", "should_trigger": true },
{ "query": "Is retail media worth it for our CPG brand versus staying organic on the marketplace?", "should_trigger": true },
{ "query": "Pinterest ads vs Meta ads for home decor - which one?", "should_trigger": true },
{ "query": "We want to start advertising in Japan. Which platforms?", "should_trigger": true },
{ "query": "The board approved a paid media budget. Recommend where it goes.", "should_trigger": true },
{ "query": "Should a plumbing company do Google ads or Facebook ads?", "should_trigger": true },
{ "query": "What ad channels can a fintech lending app actually use?", "should_trigger": true },
{ "query": "Our competitor runs YouTube ads everywhere. Should we copy them?", "should_trigger": true },
{ "query": "Is $5k/month enough to advertise on more than one platform?", "should_trigger": true },
{ "query": "Which advertising channel has the fastest payback for ecommerce?", "should_trigger": true },
{ "query": "Help me decide between native ads and programmatic display for our launch.", "should_trigger": true },
{ "query": "We sell $40k software to hospitals - where do we advertise?", "should_trigger": true },
{ "query": "ok so like, meta or tiktok for my candle shop lol", "should_trigger": true },
{ "query": "First paid campaign ever - pick the channel and tell me why.", "should_trigger": true },
{ "query": "Are podcast sponsorships a good first ad channel for a dev tool?", "should_trigger": true },
{ "query": "What channel should our Series A startup bet its growth budget on?", "should_trigger": true },
{ "query": "We're launching in Germany and France - which ad platforms should we use there?", "should_trigger": true },
{ "query": "Should we pay for sponsored placement on software review sites?", "should_trigger": true },
{ "query": "Do we need to be on CTV now that our competitors are?", "should_trigger": true },
{ "query": "My agency proposed spreading $15k over six ad networks. Good idea?", "should_trigger": true },
{ "query": "Which ad channels can produce results before Black Friday?", "should_trigger": true },
{ "query": "I want a compounding ad channel, not a quick hit. What fits?", "should_trigger": true },
{ "query": "We only have capacity for 2 creatives a month. Which paid channel fits that?", "should_trigger": true },
{ "query": "Which paid channel fits a product nobody searches for yet?", "should_trigger": true },
{ "query": "What's the most efficient way to buy our way in front of enterprise IT buyers?", "should_trigger": true },
{ "query": "Ads for a local dental clinic - search, social, or something else?", "should_trigger": true },
{ "query": "Where do B2B companies with a tiny TAM advertise effectively?", "should_trigger": true },
{ "query": "Should we try advertising at all, or is our margin too thin for it?", "should_trigger": true },
{ "query": "How do we get customers faster? We were thinking of putting money behind some ads but no idea where.", "should_trigger": true },
{ "query": "Rank the ad channels we should try for a subscription meal kit.", "should_trigger": true },
{ "query": "Which advertising platform gives readable results on $2k/month?", "should_trigger": true },
{ "query": "Should we advertise our HR software on LinkedIn or just do Google Ads?", "should_trigger": true },
{ "query": "Is programmatic display a good starting channel?", "should_trigger": true },
{ "query": "Choosing between short-form video ads and paid search for our app launch.", "should_trigger": true },
{ "query": "What's the best ad channel for reaching CFOs?", "should_trigger": true },
{ "query": "We got a $50k marketing budget for next quarter - where does the paid part go?", "should_trigger": true },
{ "query": "Are streaming TV ads only for big brands or can startups use them?", "should_trigger": true },
{ "query": "Which paid channel should a nonprofit use for donor acquisition?", "should_trigger": true },
{ "query": "Talk me out of (or into) LinkedIn ads for my $99/month tool.", "should_trigger": true },
{ "query": "What paid channels actually work for ecommerce under $30 AOV?", "should_trigger": true },
{ "query": "New CMO here. Building the paid media plan from zero - channel recommendations?", "should_trigger": true },
{ "query": "Is audio advertising worth it compared to social ads?", "should_trigger": true },
{ "query": "Which platforms should we advertise on given we can't make video?", "should_trigger": true },
{ "query": "Should we advertise where our buyers hang out even if it's expensive?", "should_trigger": true },
{ "query": "Help me pick one ad channel to focus on this year.", "should_trigger": true },
{ "query": "We're post-launch with zero paid spend and $12k/month approved. Where does it go?", "should_trigger": true },
{ "query": "Google Ads vs Bing Ads vs Meta for lead gen - how do I choose?", "should_trigger": true },
{ "query": "Do paid social ads make sense for a 6-month enterprise sales cycle?", "should_trigger": true },
{ "query": "What ad channels can a cannabis brand even use?", "should_trigger": true },
{ "query": "Which channel should we advertise a mobile game on?", "should_trigger": true },
{ "query": "Would advertising on Spotify work for us?", "should_trigger": true },
{ "query": "Our growth is all word of mouth and it's stalling. Where should we put money to acquire customers?", "should_trigger": true },
{ "query": "My mentor says pick one channel and go deep. Which one for B2B services?", "should_trigger": true },
{ "query": "Where should a brand-new ad budget go: Meta prospecting or Google Search?", "should_trigger": true },
{ "query": "What's the cheapest ad platform to start with?", "should_trigger": true },
{ "query": "Which ad networks fit an audience of retired homeowners?", "should_trigger": true },
{ "query": "Starting paid acquisition for our newsletter business - which channel?", "should_trigger": true },
{ "query": "Is it dumb to run ads on five platforms at once?", "should_trigger": true },
{ "query": "We make industrial sensors. Does online advertising even have a channel for us?", "should_trigger": true },
{ "query": "How should I split this month's budget between my running Meta and Google campaigns?", "should_trigger": false },
{ "query": "We're reallocating Q4 spend across our existing campaigns - help.", "should_trigger": false },
{ "query": "Should prospecting or retargeting get more of our current budget this month?", "should_trigger": false },
{ "query": "My campaign is pacing to overspend the monthly budget by 20% - what do I do?", "should_trigger": false },
{ "query": "Track our daily ad spend against the $30k monthly cap.", "should_trigger": false },
{ "query": "Target CPA or maximize conversions for this campaign?", "should_trigger": false },
{ "query": "When should I switch from manual CPC to automated bidding?", "should_trigger": false },
{ "query": "Set our maximum allowable CAC and ROAS floor policy.", "should_trigger": false },
{ "query": "What CAC can we afford given our margins and cash runway?", "should_trigger": false },
{ "query": "Is our $180 CAC healthy for a B2C subscription business?", "should_trigger": false },
{ "query": "Benchmark our ROAS against industry standards.", "should_trigger": false },
{ "query": "Our Google campaign is profitable - how fast can we raise the budget?", "should_trigger": false },
{ "query": "Plan the budget ramp for scaling our winning Meta campaign.", "should_trigger": false },
{ "query": "What audiences should we layer on Meta for our ICP?", "should_trigger": false },
{ "query": "Build our retargeting pools and lookalike tiers with exclusions.", "should_trigger": false },
{ "query": "Carousel or single image for this campaign?", "should_trigger": false },
{ "query": "Which ad format works best for Stories placements?", "should_trigger": false },
{ "query": "Design an A/B test for our new ad creatives.", "should_trigger": false },
{ "query": "How much budget should each creative test cell get?", "should_trigger": false },
{ "query": "Verify our conversion tracking is firing correctly before launch.", "should_trigger": false },
{ "query": "Purchases aren't showing up in Ads Manager - help me debug.", "should_trigger": false },
{ "query": "Meta reports 240 conversions, Shopify shows 95 - reconcile this.", "should_trigger": false },
{ "query": "Why does GA4 disagree with Google Ads on conversion counts?", "should_trigger": false },
{ "query": "Our CPA doubled in three weeks - diagnose the account.", "should_trigger": false },
{ "query": "ROAS dropped from 3.1 to 1.4 - what's wrong with the account?", "should_trigger": false },
{ "query": "Is my ad creative fatiguing or is it audience saturation?", "should_trigger": false },
{ "query": "Frequency hit 3.8 and CTR is falling - is the creative worn out?", "should_trigger": false },
{ "query": "Write 10 headline variants for our spring sale ad.", "should_trigger": false },
{ "query": "Give me ad copy angles for cost-conscious buyers.", "should_trigger": false },
{ "query": "Draft a creative brief for our UGC video shoot.", "should_trigger": false },
{ "query": "Script a 30-second UGC ad for our supplement brand.", "should_trigger": false },
{ "query": "Score these five video hooks and tell me which deserve test budget.", "should_trigger": false },
{ "query": "How long should our cart-abandonment retargeting window be?", "should_trigger": false },
{ "query": "Design the retargeting sequence for our checkout drop-offs.", "should_trigger": false },
{ "query": "How many customers do I need in a lookalike seed list?", "should_trigger": false },
{ "query": "Should I seed the lookalike with all purchasers or the top 25% by LTV?", "should_trigger": false },
{ "query": "Build negative keyword lists from this search terms report.", "should_trigger": false },
{ "query": "We have too many campaigns - which should we merge without resetting learning?", "should_trigger": false },
{ "query": "Should we consolidate our 40 ad sets into 5?", "should_trigger": false },
{ "query": "Plan a campaign promoting our CEO's LinkedIn posts as ads.", "should_trigger": false },
{ "query": "Map the buying committee for our $80k security platform.", "should_trigger": false },
{ "query": "Our ads get clicks but the landing page doesn't convert - audit it.", "should_trigger": false },
{ "query": "Why is our landing page conversion rate stuck at 0.4%?", "should_trigger": false },
{ "query": "Build a swipe file of our competitors' ads.", "should_trigger": false },
{ "query": "What are our competitors running in the Meta Ad Library right now?", "should_trigger": false },
{ "query": "Adapt our ad copy for AI chat assistant ad slots.", "should_trigger": false },
{ "query": "How do I become a media buyer?", "should_trigger": false },
{ "query": "Review my performance marketing resume.", "should_trigger": false },
{ "query": "Write a job description for a paid media manager.", "should_trigger": false },
{ "query": "Which PPC newsletters and podcasts should I follow to stay current?", "should_trigger": false },
{ "query": "What should my ad agency's monthly performance report include?", "should_trigger": false },
{ "query": "Should we focus our organic content on LinkedIn or Instagram?", "should_trigger": false },
{ "query": "SEO or content marketing for long-term growth?", "should_trigger": false },
{ "query": "Which podcasts should I pitch myself on as a guest?", "should_trigger": false },
{ "query": "Pick the best influencers for our product seeding campaign.", "should_trigger": false },
{ "query": "Where should we distribute our press release?", "should_trigger": false },
{ "query": "Which marketplaces should we list our product on?", "should_trigger": false },
{ "query": "Choose an email marketing platform for our newsletter.", "should_trigger": false },
{ "query": "Which CRM platform should our sales team pick?", "should_trigger": false },
{ "query": "Help me choose an analytics platform for the website.", "should_trigger": false },
{ "query": "What social media platform should our brand post on daily?", "should_trigger": false },
{ "query": "Should our B2B startup do webinars or trade shows this quarter?", "should_trigger": false },
{ "query": "Best affiliate networks to join for our SaaS?", "should_trigger": false },
{ "query": "How do we improve our Google Ads quality score?", "should_trigger": false },
{ "query": "What keywords should we bid on for the new campaign?", "should_trigger": false },
{ "query": "Set the daily budgets per campaign in our Google Ads account.", "should_trigger": false },
{ "query": "My boss wants a report on last month's paid performance across channels.", "should_trigger": false },
{ "query": "When is the best time of day to run our ads?", "should_trigger": false },
{ "query": "How do I get access to TikTok Spark Ads for creator posts?", "should_trigger": false },
{ "query": "Compare HubSpot vs Marketo for marketing automation.", "should_trigger": false },
{ "query": "Which DSP vendor should we sign the programmatic contract with?", "should_trigger": false },
{ "query": "Audit whether our ad spend last quarter was wasted.", "should_trigger": false }
]
}
references/channel-family-profiles.md›
# Channel family profiles
Directional profiles for the affordability gate (step 1), ranking (step 4), and funding floor check (step 5). Named platforms below appear only as the source of a specific figure or policy, not as the operating vocabulary.
All figures are 2025-2026 reporting-period numbers from vendor-published datasets and practitioner guidance; vendors have a commercial interest in more spend, and medians hide very wide variance (published search CPCs in the legal vertical alone span roughly $1.50 to $50+). Use as ranges, never targets; reverify against current platform documentation before spend commits.
Both the table and the sections below run in SKILL.md step 4's default efficiency order - highest value returned per unit of effort first. That order is a default, not a law: re-rank it against the business's own advantages, in-house skills and owned assets, and against the interview's deadline, one-off-versus-compounding and effort-ceiling answers, before using it.
## Table of Contents
- [Summary table](#summary-table)
- [Paid search](#paid-search)
- [Retail media](#retail-media)
- [Paid social](#paid-social)
- [B2B professional networks](#b2b-professional-networks)
- [Short-form video](#short-form-video)
- [Podcast / audio](#podcast-audio)
- [Connected TV](#connected-tv)
- [Native / discovery - price-discovery bucket](#native-discovery-price-discovery-bucket)
- [Programmatic display - price-discovery bucket](#programmatic-display-price-discovery-bucket)
- [Paid review listings - price-discovery bucket](#paid-review-listings-price-discovery-bucket)
## Summary table
Effort is what the team spends, not what the media costs: setup hours, assets produced per month, coordination with third parties, and how reversible the commitment is.
| Family | Demand state | Value ceiling | Effort load | Practical monthly minimum | Signal threshold |
| ------------------------------ | -------------------------------- | ------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------- |
| Paid search | Capture | Bounded by existing search volume; the highest-yield revenue inside that bound | A week of setup, near-zero ongoing (text + landing page) | Budget for ~10 clicks/day; daily budget 3-5x target CPA | ≥1 conversion/day on paper |
| Automated cross-surface search | Capture + creation blend | Same demand pool, wider surfaces, less control | A week of setup, near-zero ongoing; opaque to diagnose | ~$2,000-$10,000 (practitioner-cited) | ~30 conversions/month; ~50/month for stability |
| Retail media | Capture | Modest, bounded by the host marketplace's shelf traffic | A week (listing setup plus marketplace terms), low ongoing | $2,000-$3,000 (sponsored listings); $100/day on the second-largest network | No technical minimum on sponsored listings |
| Paid social prospecting | Creation (+ retargeting capture) | Highest of any family - unbounded by existing demand | A standing job: 8-15+ variants per ad set, refresh on frequency | ~$1,000/week per ad set at a $20 CPA | ~50 optimization events per ad set per 7 days |
| B2B professional networks | Capture/creation hybrid | High value per lead, capped by a small addressable audience | A week per asset cycle (~4-6 week asset life); a quarter of commitment before a verdict | $3,000-$5,000; ~$5,000 (North America) for sales outcomes | Audience 300 technical floor; ~5,000-50,000 working range; 15-25 conversions/month realistic |
| Short-form video | Creation | High ceiling on broad consumer categories | A standing job, heavier than paid social: 10-20 variants, fatigue in days | Similar per-campaign logic to paid social | 10-20 creative variations per campaign |
| Podcast / audio | Creation | Strong endorsement effect, hard to read | A quarter: show booking lead times, multi-episode flights, a measurement build before launch | ~$15,000+ across 3+ shows to be measurable | Multi-episode flights; shows with 300-1,000+ downloads/episode |
| Connected TV | Creation | Broad reach, structurally weak click-level read | A quarter: video production plus an incrementality build, both due before the first result | $500 self-serve; $5,000+ programmatic DSP; $50,000+ direct | ~$165/day self-serve guidance; run ≥1 month |
Retargeting is a mode, not a family - it runs on the paid social and programmatic display surfaces. It sits second on step 4's efficiency ordering because it reuses creative and audience already paid for, so its effort is near-zero. Its value ceiling is capped by whatever traffic feeds it, which is why it is never the primary.
### What this order starves
Podcast/audio and connected TV sit last because a ratio demotes anything high on value and high on effort, and the affordability gate and the funding floor push the same two families down again. Promote one to primary against its rank only when the buyer is demonstrably on that surface, no capture family reaches them, the budget clears that family's floor by concentrating rather than spreading, and the deadline sits at least a quarter out - the full condition is in SKILL.md step 4. Short-form video is the same case one row up wherever video production is already in-house.
A family the affordability gate, the disqualifier pass or the funding floor rules out is deleted from the account's candidate set and named as deleted - it does not come back into this table as a low-scoring row.
### Price-discovery bucket - deliberately unranked
Native/discovery, programmatic display and paid review listings sit outside the ordering above. Ranking them against benchmarked families would be invented precision because cost and value vary widely. Gate each with the step 7 price-discovery test and treat them as second-channel candidates once that test returns a real number - never as the funded primary on a first plan.
## Paid search
- Capture. Fits any category buyers actively search, at any price point where CPC clears the math. Low creative burden (text + landing page).
- 2026 US benchmark (13,474 campaigns across two search networks, Apr 2025-Mar 2026): all-industry average CPC $5.42, average CPL $66.69 - the first CPL decline in five years. Cheapest verticals ~$1.63-$2.05 CPC; most expensive (legal) $9.87 average, with the underlying range ~$1.50-$50+.
- Floor logic: fund ~10 clicks/day and a daily budget of 3-5x target CPA; if the budget cannot deliver at least one conversion/day on paper, it cannot produce signal.
- Automated cross-surface search products: the only official floor is "3x CPA daily budget"; ~30 conversions/month minimum and ~$2,000-$10,000/month are practitioner extrapolations sometimes wrongly attributed to the platform.
- Effort: a week to build the account, then near-zero ongoing - no creative treadmill, and the whole thing pauses in minutes. That combination, not its media price, is why it leads the default ordering wherever the category has search volume.
- Disqualified by: no category search volume; restricted categories requiring certification.
## Retail media
- Capture at the shelf: shoppers already in a buying session. Fits products actually sold on the host marketplaces; disqualified outright if you don't sell there.
- Heavily US-concentrated: the largest network holds ~79.7% of US retail media (2025), and the top two take ~88-89% of US incremental retail media spend.
- Entry: $2,000-$3,000/month working floor for sponsored product listings on the largest network (no technical minimum); the second-largest runs self-serve from $100/day (~$3,000/month test), cut bid floors to $0.50 in late 2025, and shows ~55% lower CPCs on matched CPG terms. Managed DSP tiers start around $50,000.
- Effort: a week to set up listings and clear marketplace terms; near-zero creative burn afterwards. Value is capped by the host's own traffic, which is what keeps it below the creation families on value while staying above them on efficiency.
## Paid social
- Creation, plus the highest-volume retargeting surface. B2C default; also carries B2B creation.
- The most-cited learning threshold in paid media: ~50 optimization events per ad set per 7-day window (largest platform's official guidance) - at a $20 cost per conversion that implies roughly $1,000/week per ad set. Below it the ad set never exits learning; consolidate ad sets or optimize to a higher-frequency proxy event.
- 2025 ecommerce dataset (~35,000 brands): median CPM ~$13.48-$15.06, median CPA ~$38.19-$38.99, median ROAS 1.86-1.93x, median CTR 2.19%. CPM rose ~20% YoY; CPA held nearly flat because conversion rates rose 8.3%. Highest-CPM vertical: health & wellness at ~$20.70-$21.80.
- Small-budget practitioner tiers: "small" is under ~$3,000/month, "tiny" under ~$600/month; at those tiers a 2-3x ROAS is the reasonable goal, not 10x, and a low-volume account (~5 purchases/week) can take about a month to exit learning.
- Small-budget tactics:
- One offer.
- Skip awareness objectives.
- Lean on the platform's automated campaign type.
- Avoid edits that reset learning.
- Profitability check: at the median 1.86x ROAS, paid social is profitable only above roughly a 50% contribution margin.
- Creative burden: high. This is a standing job, not a setup task, and it is the reason the family's top-of-table value ceiling still lands mid-table on efficiency.
- 8-15+ meaningfully different variations per ad set.
- Refresh when frequency reaches ~2.5-3.
- Refresh cadence tightens to every 3-5 days above ~$1,000/day.
- Its retargeting mode is the exception: reusing existing creative against an existing audience is near-zero effort, which is why retargeting outranks prospecting on efficiency while ranking last on value.
- Disqualified/weakened by: special-ad-category regimes for credit, employment, housing, and financial products (see disqualifier checklist); static-only creative capacity at scale.
## B2B professional networks
- Hybrid: precise firmographic reach to buyers in a work mindset. The defining B2B creation-and-capture family.
- Practitioner gate (LinkedIn-focused agency guidance, published 2024-2026):
- Don't run it for pipeline below ~$5,000/month (North America); ~$500/month only proves engagement signal, not sales.
- Require LTV of roughly $10,000-$15,000 minimum to recoup the front-end premium.
- Require a clearly defined title/seniority/industry/company-size audience.
- Commit to a ~3-month test.
"Users aren't ready to buy your $10K – $100K products or services after seeing one ad."
- Costs:
- North America CPC ~$8-$16 (3-5x the ~$3-$5 typical of cheaper social).
- CPL ~$110 typical, with qualified leads at $20-$60 in favorable cases.
- Cost per sales-qualified lead often runs less than half that of cheaper social despite the higher CPC.
- Minimum daily budget per campaign ~$25-$50; the platform can spend up to 1.5x the daily campaign budget on a given day.
- Signal: audience floor of 300 is technical only - the working range is ~5,000-50,000; 15-25 conversions/month is realistic, so the 50-event learning bar of paid social rarely applies here.
- Creative burden: low volume, long life (~4-6 weeks per asset) - a week of work per asset cycle, well inside most teams' ceilings. The binding effort cost is the quarter of committed spend before a verdict, not the assets.
- Disqualified by: LTV below the floor; ICP not expressible in firmographic terms.
## Short-form video
- Creation. Fits broad B2C, younger-skewing audiences, and brands with real video capacity.
- Creative volume is the bottleneck, not media budget:
- 10-20 creative variations per campaign.
- Fatigue is algorithmic and can begin within 3-5 days at scale, before frequency reaches 2.
- The fastest brands produce 200+ assets/month.
- Production anchor: user-generated-style video runs ~$150-$300 per asset; growth-stage brands typically invest ~$1,500-$5,000/month in creative.
- Effort: the heaviest standing creative job of any family here. A team that already shoots video in-house has an unfair advantage that moves this family several rows up the default ordering; a team that does not should read its position as generous.
- Disqualified by: static-only or low-volume creative capacity - this gate fails more plans than media cost does.
## Podcast / audio
- Creation via endorsement. US podcast ad revenue $2.862B in 2025 (+17.6% YoY).
- Measurability floor: ~$15,000+ spread across 3+ shows in multi-episode flights; pick shows above roughly 300-1,000 downloads/episode.
- CPM: host-read mid-roll $25-$50, pre-roll $15-$30, programmatic ~$15.
- Measurement runs on promo codes, vanity URLs, and self-reported attribution - plan them before booking.
- Effort: a quarter, most of it before any impression - show research, booking lead times, host-read script and claim review, and a measurement scheme that has to exist at launch. Booked flights are hard to pull back mid-run, so this family also carries the least reversibility.
## Connected TV
- Creation / brand. US-centric maturity: ~117M US connected-TV households in 2025; thin in many other markets.
- Entry:
- $500/month self-serve tiers exist.
- $5,000+ via programmatic DSPs.
- $50,000+ for direct deals.
One self-serve vendor recommends ~$165/day; run at least one month.
- CPM $20-$40 managed (settling near $25); ad-supported streaming tiers $15-$30.
- Production on top of media: $1,500-$5,000 to repurpose existing video; $15,000+ for a new 30-second spot.
- Measurement is structurally weak at the click level - plan incrementality/geo-lift from day one.
- Effort: a quarter, and uniquely front-loaded - the video production and the incrementality build both come due before the first readable result, which is why this family sits last in the default ordering despite a broad reach ceiling. An already-owned finished spot removes most of that and moves it up.
- Disqualified by: low-contribution products (the CPM plus production math rarely clears a sub-$50 contribution); non-US target markets with thin CTV penetration.
## Native / discovery - price-discovery bucket
- Creation, mid-funnel: paid placement inside content feeds and recommendation surfaces.
- Taboola sets no hard floor: its own budget guidance recommends 10x the target CPA or at least $50/day, enough for the algorithm to gather signal, and separately frames $300/day as the threshold to exit the learning phase and scale.
- Outbrain enforces a hard minimum of $20/day and recommends $100-150/day to gather enough data to optimize.
- Neither platform publishes an official CPC rate card. Third-party CPC estimates for both span roughly $0.03-$2.00+ and disagree substantially across sources - treat any cited CPC figure as directional only, not as a benchmark.
- Gate with the step 7 price-discovery test and the general test-duration guidance (display-class: 4-8 weeks) before committing real budget beyond the platform-published minimums above.
## Programmatic display - price-discovery bucket
- Creation and retargeting at low CPMs across open-web inventory.
- Amazon DSP is the only major platform with a published minimum: $50,000/month for managed service, with no published minimum for self-service since Amazon removed its self-service floor in late 2025.
- The Trade Desk and Google DV360 publish no minimum spend on their own sites. Practitioner-reported practical floors run $50,000-$1,000,000+/month depending on account structure and buying model - unverified against platform documentation, treat as directional only.
- CPM ranges commonly cited in industry sources: $1-$4 open exchange/broad targeting, $4-$12 private marketplace/targeted, $10-$25 for B2B firmographic targeting - all third-party aggregated, not platform rate cards.
- Post-tracking-changes reality: multi-touch attribution coverage industry-wide fell from 90%+ to ~30-60%, and display is among the hardest-hit - plan measurement accordingly.
- Judge over 4-8 week tests; avoid folklore numbers and run price discovery to establish real costs for this category and your ICP.
## Paid review listings - price-discovery bucket
- Capture, B2B software: sponsored placement on software review and comparison directories, reaching buyers mid-evaluation. "Often skipped, high intent" per B2B practitioner playbooks.
- G2's published entry tier: Starter at $299/month, or $2,999 for the first year rising to $5,999/year from year two. Professional and Enterprise tiers are quote-only.
- Capterra publishes a pay-per-click floor directly in its PPC service terms: $2/click minimum bid against a $500/month budget floor, second-price auction, bids move in $0.25 increments.
- TrustRadius lists its vendor "Customer Voice" package starting at $30,000/year - the widest spread of the three sites, and a sign that category and tier drive cost far more than any single benchmark can capture.
- Fits: B2B SaaS in categories buyers actively compare. Disqualified by: categories with no active review-site traffic.
references/channel-plan-template.md›
# Channel plan template and worked examples
The output of this skill is one plan document in the structure below. Scenario inputs in the examples (prices, margins, budgets) are hypothetical; every benchmark figure they are checked against comes from the profiles file and carries its caveats - directional ranges, not targets.
## Template
```
# Channel plan - {business}, {date}
## 1. Economics
AOV/ACV: - | Gross margin: - | Contribution per sale: -
Break-even ROAS (1 / contribution margin rate): -
Monthly gross profit per customer: - | LTV / retention (if known): -
Allowable CAC: - (derivation: payback target - months × monthly gross profit; or org guardrail, source)
Payback at allowable CAC: - months (discounted: - months at -% retention)
Results due by: - | One-off win or compounding asset: - | Effort ceiling: - assets/month, - hours/week, reversibility required: -
## 2. Deleted channels
| Family | Gate failed (affordability / disqualifier / funding floor) | Specifics, dated |
(Deleted means out of the candidate set for this account - none of these reappear in section 3.)
## 3. Candidates ranked by efficiency - survivors only
| Rank | Family | Value subtotal (presence, reachability, reach, measurability) | Effort subtotal (creative burn, skill/coordination, reversibility) | Value ÷ effort | Headroom vs. allowable CAC (tiebreak only) |
(Evidence note next to every 4-5 value score and every 1-2 effort score.)
Departures from the step 4 default ordering, and what moved each row: - (unfair advantage / in-house skill / owned asset / answer to Q15-17).
Starved family promoted against its rank: - (which of podcast/audio, connected TV or short-form video, and which of the four promotion conditions held). None, if the ratio order stands.
Price-discovery bucket, unranked by design: native/discovery, programmatic display, paid review listings - second-channel candidates only, after a tiny test returns a real number.
## 4. Primary recommendation
Family: - . Rationale: demand state, value ÷ effort ratio, benchmark CPA/CPL vs. allowable CAC.
## 5. Funding plan
Budget: - /month. Primary's signal threshold: - . Share to primary: - (≥ threshold; ~80% for first 90 days is the default pattern).
## 6. Test design
Price discovery: ~$ - to learn real CPC/CPM for this targeting.
Real test: - (3-5x target CPA per concept) over - weeks (≥ full purchase cycle; same weekday start/end).
Pass: CPA/CPL ≤ allowable CAC at 90-95% confidence. Kill: 2-3x target CPA spent, zero conversions, after a readable sample.
## 7. Second-channel triggers
Add one channel only when: primary CPA rises - % on saturation signals, or budget crosses ~$ - /month with payback still ≤ - months. Next-ranked candidate: - .
## 8. Measurement approach
Attribution model, self-reported question (if creation channel), MER vs. platform ROAS (ecommerce), incrementality cadence. Known signal gaps (EU consent, mobile-OS tracking loss).
```
## Worked example - B2B
Workflow-automation SaaS. ACV $18,000, 80% gross margin, LTV ~$45,000, ~90-day sales cycle, budget $12,000/month, US + UK, demo-led (lead-gen), sales follow-up same-day, 4 net-new static assets + 1 video per month, tracking verified. Results needed by the end of next quarter. Wants a compounding asset but will take the fast win first. Effort ceiling is roughly 5 assets/month, with no committed flights.
1. **Economics**: monthly gross profit per customer = ($18,000 / 12) × 0.80 = $1,200. Allowable CAC at a 5-month payback target = $6,000 (well inside the 3-12 month band; discounted variant not binding at ~90% logo retention). Break-even ROAS not the operative metric - pipeline and payback are (B2B column).
2. **Deleted**:
- Retail media: product not sold on marketplaces.
- Short-form video and connected TV: creative capacity (one video/month cannot feed families that fatigue in days), and CTV production at $1,500-$15,000+ per asset is out of budget.
- Podcast: budget $12K < ~$15K measurability floor across 3+ shows.
All four are out of the candidate set for this account - they do not appear in the ranking below, and adding one back means re-running the gate that removed it.
3. **Ranked by efficiency**:
- Paid search: value 17 / effort 5 → 3.4 (buyers search the category; CPL benchmark ~$67 average, category-checked; a week of setup then near-zero ongoing, pausable in minutes).
- B2B professional networks: value 16 / effort 9 → 1.8 (LTV $45K clears the ~$10-15K floor; budget clears the ~$5K/month floor; CPL ~$110 → at a 10% lead-to-customer rate, CAC ~$1,100 ≪ $6,000, but a quarter of committed spend before a verdict, against a deadline this quarter).
No departure from the default ordering: nothing in this business's assets or skills moves a row, and the Q15 deadline reinforces the capture-first order rather than changing it. Paid review listings sit in the unranked price-discovery bucket (high intent, no cost benchmark).
4. **Primary**: paid search - capture first: the category has search volume, capture is unfunded, and it returns the most per hour spent before any creation budget moves.
5. **Funding**: $9,600/month (80%) to paid search - supports ~$300+/day, comfortably above the ~10 clicks/day and 3-5x target-CPA floor at the expected CPC.
6. **Test**: ~$100 price discovery on exact-match category terms; then $3,000-$5,000 per concept (3-5x a $1,000 blended target CPA at expected lead→customer rates) over 12 weeks - one full sales cycle, honoring the ~3-month B2B commitment norm. Pass: cost per customer ≤ $6,000 at 90% confidence. Kill: $2,000-$3,000 spent on a concept with zero SQLs after a readable sample.
7. **Second channel**: professional networks, added only when search CPA rises on impression-share saturation or budget crosses ~$50K/month with payback ≤ 12 months.
8. **Measurement**:
- CRM as source of truth.
- Self-reported attribution on the demo form (long cycles break last-click).
- First-touch and last-touch read side by side.
## Worked example - B2C
DTC fitness-accessories brand. AOV $90, 60% gross margin → $54 contribution; no reliable repeat-purchase data yet; impulse-to-days cycle; budget $9,000/month; US only; ~10 UGC-style videos/month (at the ~$150-$300/asset norm) plus statics; tracking verified. No hard external date; wants a compounding creative library, not a one-quarter spike; effort ceiling ~10 videos/month with one person owning production.
1. **Economics**: break-even ROAS = 1 / 0.60 ≈ 1.67x. With no proven LTV, allowable CAC is first-order: $38 (≈70% of contribution, leaving margin for variable costs). Payback: immediate at first order or never - the discounted-payback machinery is not binding.
2. **Deleted**:
- B2B professional networks: no LTV near the $10-15K floor, instant fail.
- Connected TV: $54 contribution cannot clear $20-40 CPMs plus $1,500+ production.
- Podcast: budget < ~$15K floor.
- Retail media: not yet sold on marketplaces.
No starved-family promotion applies: podcast and connected TV are the two the ratio normally starves, but the promotion condition requires that no other family reaches the buyer, and paid social demonstrably does.
3. **Ranked by efficiency**:
- Paid search: value 11 / effort 5 → 2.2 (reach scores 1, category search volume is thin and branded-skewed, but measurability and setup cost are unbeatable).
- Paid social prospecting: value 17 / effort 11 → 1.55 (audience demonstrably there; median CPA ~$38 sits exactly at the $38 ceiling, so it passes only with mandatory price discovery; the 8-15-variant burn is a standing job this team can just carry).
- Short-form video: value 16 / effort 13 → 1.23 (10 videos/month meets the 10-20-variant floor with no refresh headroom at scale).
The axes disagree and the plan says so: paid search wins the ratio on a ceiling that exhausts within weeks, so it earns a small always-on slice, not the funded primary. Q16's compounding answer promotes paid social and short-form video, both of which leave a reusable creative library behind; Q17's ceiling keeps short-form video third.
4. **Primary**: paid social - creation channel for a category nobody searches yet, with the capacity to feed it and the only value ceiling large enough to absorb the budget. Margin note in the plan: at the dataset's median 1.86x ROAS, paid social is profitable above ~50% margin - this brand's 60% clears it, thinly.
5. **Funding**: $7,600/month (~84%) to one consolidated prospecting ad set: the ~50-events/week threshold at a $38 CPA implies ~$1,900/week. Remainder to retargeting and the branded-search capture slice. No second prospecting ad set - it would split events below threshold.
6. **Test**: ~$100 price discovery against the exact audience; then $114-$190 per concept (3-5x the $38 target CPA) across the 8-12 concepts capacity allows, over 4-6 weeks (social norm, covers the short cycle, same-weekday start/end). Pass: CPA ≤ $38 at 95% confidence on the account's real numbers, not the platform dashboard alone. Kill: ~$76-$114 spent on a concept with zero purchases.
7. **Second channel**: short-form video, only after paid social saturates (frequency-driven CPA rise) - not before, and never by splitting the $9K.
8. **Measurement**:
- MER (total revenue / total ad spend) as the deciding metric - platform ROAS is inflated post-tracking-changes.
- "How did you hear about us?" at checkout.
- Quarterly geo-lift once spend is material.
## Negative example - what a failing plan looks like
Bootstrapped B2B tool, plans at $9/month and $199/month, blended "ARPU $58", 85% margin, budget $10,000/month, team of two, two static assets a month, tracking not yet verified. The draft plan: "$2,000 each to paid search, paid social, professional networks, podcast, and native - diversify and see what sticks."
Every gate fails:
- **Affordability**: blended ARPU hides the split. At a plausible $300 CAC, the $9 plan pays back in ~39 months (unaffordable - route to organic/product-led); only the $199 plan (~$169/month gross profit, ~1.8-month payback) can carry paid at all. The plan targets a blended customer who does not exist.
- **Gates ignored**: professional networks fail twice ($2K < ~$5K/month floor; no LTV near $10-15K on a $199/month plan without proven multi-year retention). Two statics/month cannot feed paid social's 8-15-variant burn. Tracking is unverified, so every learning phase is wasted by design.
- **Funding floors**: $2,000/month ≈ $66/day per channel - the textbook anti-pattern ($10K split five ways lands at $60-80/day, below every family's learning threshold). Podcast at $2K is 13% of its ~$15K measurability floor. Five experiments, all below minimum sample size; total spend $10K/month, total learning zero.
- **No test design**: no price discovery, no pass/fail, no kill criteria, no duration tied to the sales cycle - "see what sticks" is unfalsifiable.
- **No ranking at all**: an equal split is what a menu with no ordering produces. Even the instinct behind it is backwards - it treats the five families as interchangeable when their effort loads span a week of setup to a standing creative job, and the team is two people producing two statics a month.
The correct rewrite:
1. Verify tracking first.
2. Gate paid to the $199 tier only.
3. Delete professional networks and podcast outright - they failed the floor and the LTV gate, so they are named as deleted, not ranked last where the next budget conversation revives them.
4. Rank the remaining survivors by value ÷ effort and take the top one.
5. Fund it above its threshold with ~80% of budget.
6. Run one designed test.
7. Park only the still-eligible families as ranked candidates with explicit add-triggers.
references/disqualifier-checklist.md›
# Disqualifier checklist
Run before any ranking (SKILL.md step 2). A channel the business cannot legally, geographically, or operationally run is deleted from the candidate set for this account and named as deleted, with the rule that removed it - never demoted to the bottom of the ranked table, where it silently returns as scope.
Policies and platform rules change frequently - everything below reflects the 2025-2026 reporting period; if you can browse the web, reverify against current platform policy pages before finalizing the plan, and date every check in the plan document. Named platforms appear only where the rule itself is platform-specific.
## 1. Regulated offer categories
- **Special ad categories (largest social platform)**: credit, employment, housing, and - since January 21, 2025 in the US - financial products and services must be declared. This does not disqualify the family outright, but it materially weakens it for lenders, recruiters, realtors, and insurers - score targeting reachability accordingly, or disqualify if the plan depended on the stripped capabilities.
- Declaration locks age targeting to 18-65+.
- Removes gender narrowing.
- Imposes a 15-mile minimum geographic radius.
- Removes ZIP-level precision and detailed-targeting exclusions.
- Replaces lookalike audiences with a weaker special-audience tool.
- **Political, electoral, and social-issue ads in the EU**: under the TTPA (applicable October 10, 2025), the largest search platform stopped EU political ads (its video platform from September 23, 2025) and the largest social platform stopped EU political, electoral, and social-issue ads from early October 2025 - fines run up to 6% of annual turnover. For that messaging in the EU, paid search and paid social are disqualified, and much advocacy/NGO communication is caught in the same net.
- **Restricted categories generally**: alcohol, gambling, crypto, CBD, firearms, adult, and pharma/health each face platform-specific bans, certification requirements, or geo-limits. Check each candidate family's policy for the exact offer and every target country before scoring - pre-approval processes can add weeks.
## 2. Geography and language
- **Markets with a local stack**: several major markets run on channels outside the global search/social platforms. A "global" plan that ignores the local stack simply doesn't reach these markets - the blocked family has no inventory there, so delete it for that market and name it as deleted rather than ranking it low.
- China: the dominant local search engine holds ~60-85% search share; the major global search/social platforms are blocked.
- South Korea: a local portal (~39-40M MAU) dominates search alongside a national messaging app.
- Japan, Thailand, Taiwan: a messaging platform with ~94M users in Japan and ~91% penetration in Taiwan.
- Russia: dominated by local search and social incumbents.
- Vietnam: dominated by a local messaging app.
- **Channel maturity by market**: connected TV is deepest in the US (~117M connected-TV households, 2025) and thin elsewhere; retail media is US-concentrated (top two networks take ~88-89% of US incremental retail media spend). A family that works at home may be marginal in the expansion market.
- **Localization cost**: language and creative reshoots are a real budget line - a family that needs per-market video is more expensive than its CPM suggests.
- **EU signal quality**: consent requirements degrade tracking - consent-mode adoption in the EEA is 90%+, but ~67% of implementations are non-compliant, so expect materially weaker measurable signal on EU traffic and plan measurement (step "Measurement coupling") accordingly.
## 3. Capacity gates
These are operational disqualifiers - the channel works, the business can't run it yet.
1. **Creative production volume**: compare the family's burn rate (profiles file) against interview Q11. Short-form video needs 10-20 variations per campaign with fatigue in days; high-volume paid social needs 8-15+ variations per ad set with refresh at frequency ~2.5-3. A team producing two static assets a month cannot run either as primary - pick a lower-burn family or fix production first.
2. **Landing page readiness**: channels sending cold traffic (paid social, short-form video, connected TV, native) require dedicated, tested landing pages; without them the entire learning phase is wasted spend. Route a review through mbfinotti/advertising-skills@paid-landing-page-audit before launch, not after the first bad week.
3. **Speed-to-lead**: lead-gen channels require fast sales follow-up; slow response collapses the effective conversion rate and makes the channel look unprofitable when the constraint is operational. If interview Q13 reveals follow-up measured in days, fix the process before buying leads - or the test will fail for reasons that have nothing to do with the channel.
## 4. Measurement readiness
Launching before conversion tracking is verified wastes the learning phase and voids the channel test. Verify events end-to-end via mbfinotti/advertising-skills@ad-conversion-tracking before any spend. Post-tracking-changes context: mobile-OS privacy prompts cut 30-45% of iOS conversion tracking, and multi-touch attribution coverage fell from 90%+ to ~30-60% - factor the measurable share into pass/fail math rather than assuming the dashboard sees everything.
SKILL.md›
---
name: ad-platform-selection
description: "Decide which paid channel families fit a business's economics, audience, funnel stage, and budget at campaign setup - paid search, paid social, short-form video, connected TV, native/discovery, retail media, programmatic display, podcast/audio, paid review listings, B2B professional networks - including whether each is affordable against allowable CAC and payback. Use whenever the user asks where to spend a new ad budget, which platform to advertise on, how to build a paid channel mix, whether a channel is worth trying, or compares two ad channels - even if they never say 'channel selection'. Covers B2B and B2C. Do NOT use to split an ongoing budget across existing campaigns - use mbfinotti/advertising-skills@ad-spend-allocation instead."
license: MIT
metadata:
author: Maya-Beth Finotti
version: "1.2.6"
---
# Platform Selection
You are a paid-media strategist. Recommend which paid channel family (or small set of families) a business should run, given its unit economics, audience, funnel stage, and budget. This is a one-time, campaign-setup decision framework: it ends with a channel plan naming one funded primary channel, the reasons every other channel was rejected, and a test with pass/fail criteria.
Channel choice is arithmetic, not preference. Work in channel categories, never in one vendor's product vocabulary - every family here has several competing vendors that share the same mechanics.
What decides the shortlist:
- Economics decide what is affordable.
- Learning thresholds decide what can produce a readable signal.
- Regulation, geography, and operational capacity eliminate the rest.
Out of scope - hand off to the sibling skill instead of covering it here:
- Recurring budget split across running channels: mbfinotti/advertising-skills@ad-spend-allocation
- Day-to-day pacing: mbfinotti/advertising-skills@ad-budget-pacing
- Bidding method: mbfinotti/advertising-skills@ad-bidding-strategy
- Setting the CAC ceiling / ROAS floor policy: mbfinotti/advertising-skills@ad-spend-guardrails
- Benchmarking current CAC/ROAS health: mbfinotti/advertising-skills@cac-roas-benchmark
- Scaling a proven channel: mbfinotti/advertising-skills@paid-media-scaling
- Audience tiering: mbfinotti/advertising-skills@ad-audience-targeting
- Ad format choice: mbfinotti/advertising-skills@ad-format-fit
- Creative variant tests: mbfinotti/advertising-skills@ad-creative-test-plan
- Tracking verification: mbfinotti/advertising-skills@ad-conversion-tracking
- Landing page review: mbfinotti/advertising-skills@paid-landing-page-audit
This skill uses allowable CAC and payback as gates: it takes those targets as input, rather than setting them.
A note on every number in this skill and its references: the largest public benchmark datasets are published by vendors with a commercial interest in more ad spend, and their medians hide very wide variance (published search CPCs in one vertical span roughly $1.50 to $50+). Treat all figures as directional ranges, never as targets. If you can browse the web, reverify any figure against current platform documentation before spend commits; if you cannot, say the figure is unverified-current and date it.
## Interview
Ask one question at a time, multiple-choice where possible. Skip anything already answered. Do not recommend anything until the economics questions (2-6) are answered or explicitly unknown.
1. B2B, B2C, or both?
2. What do you sell, and at what price - average order value (B2C) or annual contract value (B2B)?
3. Gross margin: under 30% / 30-50% / 50-70% / over 70%?
4. LTV and retention, if known. (Unknown is fine - the gate falls back to first-order contribution.)
5. Sales cycle: impulse (hours-days) / weeks / about a quarter / multi-quarter?
6. Total monthly budget available for paid media?
7. Target geographies and languages?
8. Is the category regulated or restricted anywhere you sell (credit, employment, housing, financial products, health, alcohol, gambling, crypto, weapons, adult, political/advocacy)?
9. Who is the buyer, and where do they already concentrate - what do they search, watch, listen to, or scroll?
10. Which paid channels are already running, and with what results?
11. Creative production capacity: how many net-new assets per month, and video or static only?
12. Are landing pages and conversion tracking live and verified?
13. If lead-gen: how fast does sales follow up on a new lead?
14. Which channel families does the team already know how to run?
15. By what date must results have landed - a specific date, not "soon"?
16. Do you want a one-off win this quarter, or a compounding asset that keeps returning?
17. What is your effort ceiling: assets per month, hours per week, headcount, and how reversible the commitment has to be?
Question 14 is a scoring input, not a shortcut - "the channel the team knows" is a named failure mode when it overrides the economics.
Questions 15-17 re-rank step 4's default ordering, so ask them before recommending anything.
- Q15 (near date): promotes the fast-acting capture rungs (paid search, retargeting, retail media); demotes every creation family whose verdict takes a quarter.
- Q16 (compounding mandate): promotes short-form video, podcast/audio and paid social prospecting, whose creative library and audience keep working after the flight ends.
- Q17 (tight effort ceiling): promotes low-burn families (paid search, retargeting, B2B professional networks at ~4-6 weeks per asset); rules out the standing creative jobs outright, whatever they score on value.
## Step 1 - Affordability gate
Run this before looking at any channel. It is identical for B2B and B2C; only the input names change (AOV vs. ACV, ROAS vs. pipeline).
1. Contribution margin per sale = price × gross margin rate (net of variable costs).
2. Break-even ROAS = 1 / contribution margin rate. A 60% margin breaks even at ~1.67x; the folk target of "4x ROAS" is wrong for most businesses. For reference, the median paid-social ROAS across a ~35,000-brand ecommerce benchmark dataset was 1.86-1.93x - profitable only above roughly a 50% margin.
3. Allowable CAC = the maximum contribution you will spend to acquire a customer. If the org has set one (mbfinotti/advertising-skills@ad-spend-guardrails), take it as given; otherwise derive it from the payback target below.
4. Payback period = CAC / monthly gross profit per customer. Target band: 3-12 months; up to 18 months for enterprise sales motions. Under 3 months usually signals underinvestment, not health.
5. Discounted payback = CAC / (monthly gross profit × annual retention rate). Worked check: $300 CAC, $99/month gross profit, 70% retention → 3.0 months raw but 4.3 months discounted. When retention is weak, discounted payback blows past 12 months even when raw payback looks fine - that gap is the early warning.
6. Delete every channel family whose realistic benchmark CPA/CPL exceeds allowable CAC (directional ranges: [references/channel-family-profiles.md](references/channel-family-profiles.md)). Deleted means gone from the candidate set for this account and named in the plan's deleted list with the gate it failed - never carried into step 4's table as a low-scoring row. A ruled-out family left at the bottom of a score table reappears as scope for whoever reads the table without the gate.
Compute payback per plan or per cohort, never on blended averages. The same $300 CAC is ~33 months on a $9/month plan (unaffordable - route that tier to organic or product-led motions), 3.0 months at $99/month, and days at $999/month. One blended number describes none of them.
Worked gate: an $80-AOV product at 40% margin leaves ~$32 contribution - that excludes B2B professional networks (CPLs commonly $100+) and most connected TV before any scoring happens. A B2B product near the ~$32,000 average B2B SaaS LTV can absorb both.
### Why payback is the gate, not LTV:CAC
The popular 3:1 LTV:CAC test has four flaws:
1. It assumes every customer churns - the best customers compound instead.
2. It assumes churn is evenly timed - in practice most churn front-loads into the first three months.
3. Blended ARPU hides per-plan variance - a $9/month and a $999/month plan average into a number that describes neither.
4. It ignores revenue delay - free trials and long sales cycles mean cash arrives after CAC is spent, and that gap is where companies run out of cash.
Use LTV:CAC only as a secondary sanity check, and only when payback is already under 12 months.
## Step 2 - Disqualifier pass
Delete channels the business cannot legally, geographically, or operationally run, including any target market where the family has no inventory. A deleted channel leaves the candidate set entirely and is named in the plan's deleted list with its reason; it never appears in step 4's table, whatever it would have scored. Full checklist with sourced specifics: [references/disqualifier-checklist.md](references/disqualifier-checklist.md).
- **Regulated category rules**: credit, employment, housing, and financial-product offers trigger restricted-targeting regimes on major social platforms (demographic and geographic narrowing stripped, similar-audience tools removed); other restricted categories (alcohol, gambling, crypto, health, weapons, adult) face outright bans, certification requirements, or geo-limits per channel.
- **Political and issue ads**: the largest search and social platforms exited EU political, electoral, and social-issue advertising entirely in October 2025 under the TTPA - for that messaging in the EU, those families simply do not exist.
- **Geography and language**: several major markets block the global search/social platforms or are dominated by local ones (search, social, and messaging each have local leaders in China, South Korea, Japan, Russia, Vietnam, and others); connected TV and retail media are heavily US-concentrated.
- **Channel maturity by market**: a family that is mainstream in one market can be marginal in the target market - check before scoring, not after launch.
- **Capacity gates**:
- Creative production volume: short-form video and paid social burn creative faster than most teams can produce it.
- Landing pages: cold-traffic channels waste their entire learning phase without dedicated, tested pages.
- Speed-to-lead: slow sales follow-up collapses lead-gen conversion and makes the channel look unprofitable when the constraint is operational.
## Step 3 - Demand state and funnel stage
Classify what the budget must do. This distinction is identical for B2B and B2C.
- **Demand capture** harvests intent that already exists: paid search, retargeting, retail media, paid review listings. Fastest payback; ceiling limited by existing demand.
- **Demand creation** generates intent that doesn't exist yet: paid social, short-form video, connected TV, podcast/audio, native/discovery. Slower payback; the only way to grow past the captured ceiling.
Decision logic: if the category has search volume and capture is underfunded, capture comes first - it returns the most per hour spent, because the intent already exists and the creative burden is a text ad plus a landing page. If nobody searches for the category, capture cannot be the primary and a creation channel must be, with the longer payback and the standing creative job priced in. Multi-quarter sales cycles eventually need both, but the sequencing still starts at the higher ratio, not the larger ceiling.
Warning that must precede any data read: platform-reported attribution systematically over-credits capture channels and undercounts creation channels - last-click favors the channels people pass through when already ready to buy; one practitioner analysis estimates social is under-reported by roughly 70%. A naive read of existing dashboards therefore biases the channel decision toward the bottom of the funnel before you have decided anything.
Named frameworks that fit this step when the user wants a fuller exploration:
- Bullseye method: brainstorm all channels, cheaply test 3-4, concentrate on the single best.
- Four Fits framework (channel-model fit): "lower ARPU models depend on lower CAC channels, while higher ARPU models can afford higher CAC channels" - products mold to channels, not the reverse.
## Step 4 - Rank the survivors by efficiency
Rank every family that survived steps 1-2 by value returned per unit of effort - never by whichever is cheapest to start. Cheap and efficient are different orderings, and only the second answers "which one first". Policy eligibility is already handled by the disqualifier pass and is not re-scored.
Effort here is what the team spends:
- Assets produced per month.
- Setup hours.
- Coordination with third parties.
- Skill already in-house.
- How reversible the commitment is.
Default ordering, before anything specific to this business is known:
- efficiency: paid search > retargeting > retail media > paid social prospecting > B2B professional networks > short-form video > podcast/audio > connected TV
- value (ceiling the family can reach): paid social prospecting > short-form video > paid search > connected TV > B2B professional networks > podcast/audio > retail media > retargeting
- effort: connected TV > podcast/audio > short-form video > paid social prospecting > B2B professional networks > retail media > paid search > retargeting
- time-to-effect: paid search (days from a cold start) > retargeting (days, but only once a traffic pool has accumulated) > retail media (weeks) > paid social prospecting (exits learning at ~50 events per ad set per 7 days) > B2B professional networks (a quarter of committed spend is the practitioner norm) > short-form video (serves the day assets land, but needs several fatigue cycles to read) > podcast/audio (show booking lead times delay the first impression) > connected TV (production, then an incrementality build, before any readable result)
- compliance cost: podcast/audio (host improvises the claim on air, so review covers wording you never see) > connected TV (booked flights, creative clearance, no clean way to pull back mid-flight) > retargeting (consent for audience lists, data-residency review) > paid social prospecting == B2B professional networks (special-ad-category declaration strips targeting for regulated offers) > short-form video (talent and music rights on creator-made assets, with usage windows that expire mid-flight) > retail media (marketplace listing terms) > paid search (ad-copy policy review only)
The one tie is real: paid social prospecting and B2B professional networks face the same self-declaration for the same four regulated categories, and both answer it by stripping targeting rather than by review - identical in kind, in cost, and in timing.
**What this order starves.** A ratio demotes whatever is high on value and high on effort, and here that is podcast/audio and connected TV: the highest monthly minimums and the longest wait for a readable verdict. Step 1's affordability gate and step 5's funding floor push the same two down a second and third time, so a plan built on efficiency alone never funds them, even when they are the only families that reach the buyer at all. Short-form video is the same failure one row up, and gets the same treatment when video production is already in-house.
Promote one to primary against its rank only when all four hold:
- Interview Q9 evidence puts the buyer on that surface.
- No capture family reaches them (thin category search volume, not sold on a host marketplace).
- The budget clears that family's floor by concentrating rather than spreading.
- The Q15 deadline sits at least a full quarter out.
Where the axes disagree: paid social prospecting and short-form video hold the highest value ceilings and the heaviest effort - a standing creative job, not a setup task - which is why they sit mid-table on efficiency while topping value. Retargeting is near-zero effort and tops nothing on value; it multiplies traffic already being bought and is never a primary.
Rows that move with business type:
- Professional networks usually leave at step 1 for consumer offers.
- Paid social prospecting rises to second behind paid search for broad consumer categories - the same call step 3 already makes on capture versus creation.
Native/discovery, programmatic display and paid review listings are deliberately left out of the ordering. Placing them against benchmarked families would be invented precision because cost and value vary too widely to rank. Treat them as a price-discovery bucket: eligible as a second channel once a tiny test returns a real number, never as the funded primary on a first plan.
This ordering is a default, not a law - it shifts with context and with who executes it. Re-rank it against what you already know about this business before scoring anything:
- An unfair advantage moves its family up several rows: an in-house video team, an existing audience on one surface, a founder already known in a podcast niche, an owned list large enough to seed matching.
- A hard skill already in-house cuts that family's effort score, which is the whole denominator.
- An asset already owned - finished video, a live storefront on a host marketplace, a review listing with traffic - removes setup effort the default order assumes nobody has.
- The interview's deadline, one-off-versus-compounding and effort-ceiling answers (Q15-17) override the default outright; see the note under the interview.
Then score the survivors for this business and order them by the ratio. Value axes score 5 = strongest; effort axes score 5 = heaviest lift.
| Side | Axis | Score 5 | Score 1 |
| ------ | -------------------------- | -------------------------------------------------------------------- | ----------------------------------------------------------- |
| Value | Audience presence | Buyers demonstrably concentrate here (interview Q9 evidence) | No evidence buyers are here |
| Value | Targeting reachability | The family's targeting can isolate the actual ICP | ICP not expressible in this family's targeting |
| Value | Reach at that targeting | Enough volume at that targeting to matter | Audience exhausts in weeks or pays a small-audience premium |
| Value | Measurability | Conversions attributable with the tracking you have | Mostly unmeasurable without incrementality infrastructure |
| Effort | Creative burn vs. capacity | Burn rate far exceeds current production capacity | Burn rate sits well inside it |
| Effort | Skill and coordination | No experience; launch needs third parties, contracts, or a new build | Team already runs this family unaided |
| Effort | Reversibility | Committed flights or contracts hold the budget for a quarter | Can be paused or redirected within days |
- Sum the value side, sum the effort side, and rank by value ÷ effort. That ratio, not either sum, is the ordering.
- Score from interview evidence, not enthusiasm; note the evidence next to any 4-5 on the value side and any 1-2 on the effort side.
- Break ties with headroom against allowable CAC: the family whose benchmark CPA/CPL sits furthest under the ceiling wins. Headroom never re-enters the ratio - step 1 already removed everything over the ceiling, and the surviving benchmarks are too wide to rank on.
- The ratio ranks survivors only. It never overrides the affordability gate or the disqualifier pass - a high-ratio unaffordable channel was deleted at step 1, stays out of this table, and is named in the deleted list rather than ranked last.
## Step 5 - Funding floor check
The top-ranked channel must be fundable above its learning/signal threshold with the stated budget - thresholds are the true floor, not platform billing minimums. Directional floors and signal thresholds per family: [references/channel-family-profiles.md](references/channel-family-profiles.md).
If the budget cannot fund the winner above its floor, the correct answer is the next-ranked channel whose floor the budget does clear - never a thinner spread of the same money. Walk down the step 4 ratio ordering, not down a price list; an underfunded channel produces no readable signal, which is worse than a readable one from a lower-ranked family.
Delete the unfundable family from this plan's candidate set and name it in the deleted list with the floor it missed and the budget that would clear it. Leaving it in the ranked table as a top row nobody funds is how it returns as scope at the next budget conversation, with its failed floor forgotten.
## Step 6 - Concentration and sequencing
- One primary channel for a startup; two to three for a scaled team. "Don't diversify yourself in too many channels; it's like a death sentence" is the practitioner consensus, and most businesses find only one channel that truly works.
- Fund the primary above its threshold before spending anywhere else; a common working pattern is ~80% of budget on the primary for the first 90 days.
- The anti-pattern, in arithmetic: $10K/month split five ways is $60-80/day per channel - below every major family's learning threshold - producing five separate experiments, each running below minimum sample size. Nothing is learned; everything is spent.
- Add exactly one channel at a time, and only when the primary saturates (rising CPA, audience exhaustion) or the budget crosses a breakpoint (practitioner anchor: ~$50K/month) with payback still inside the target band. Scaling the proven primary itself belongs to mbfinotti/advertising-skills@paid-media-scaling.
## Step 7 - Channel test design
Design the test for the chosen channel before full commitment. This is channel-level validation - creative variant testing belongs to mbfinotti/advertising-skills@ad-creative-test-plan.
1. **Price discovery first**: run a deliberately tiny spend (on the order of $100) purely to learn the real CPC/CPM for this specific targeting. Published benchmarks are consistently wrong for a specific ICP; this step replaces them with your own number before real money moves.
2. **The real test**: budget 3x target CPA minimum (5x preferred) per concept. Run it for the full purchase cycle - directional durations: search 2-4 weeks, social 3-6 weeks, display/programmatic 4-8 weeks, brand/awareness 6-12 weeks; B2B professional networks warrant a ~3-month commitment. Start and end on the same weekday.
3. **Pass**: CPA/CPL at or under allowable CAC at statistical significance (90-95% confidence) - significance, not arbitrary click counts.
4. **Kill**: 2-3x target CPA spent with zero conversions after a readable sample, then diagnose before rerunning.
5. **The false-negative trap**: a test shorter than the sales cycle reads as "zero lift" even when the channel works - a two-week test cannot validate a 90-day cycle. Never judge before one full cycle has elapsed.
## Measurement coupling
Channel choice and attribution model are coupled; the plan must state its measurement approach, because the channel mix determines which readings can be trusted. Add rungs in efficiency order as the mix and the spend justify them, never all four at once.
- efficiency: UTM discipline + last-non-direct > self-reported attribution at signup/checkout > MER (total revenue / total ad spend) > incrementality or geo-lift tests
- effort: incrementality/geo-lift (a standing job - holdout design, clean geographies, weeks of deliberately withheld spend) > MER (a reporting rebuild spanning finance and ad accounts) > self-reported (one form field plus somewhere to store the answer) > UTM discipline (a naming convention, an hour to set, then enforcement forever)
- value: incrementality/geo-lift (the only reading that survives a creation-heavy mix) > MER > self-reported > UTM discipline
The axes invert, which is the point: the most valuable reading is the least efficient one to reach for first. Start at rung 1 and climb only on a trigger.
- Default rung: capture-heavy plan at small budget stops at UTM discipline plus last-non-direct.
- Add self-reported the moment any creation channel enters the mix - it costs one form field and is the cheapest correction to the step 3 bias.
- Add MER once ecommerce spend is material enough that platform-reported ROAS and actual revenue visibly diverge.
- Add incrementality/geo-lift only on the biggest line item, and only once its spend justifies withholding some of it.
- Never judge a creation channel on its platform dashboard alone - per the bias in step 3, that reading is structurally rigged against it.
- Verify tracking before launch via mbfinotti/advertising-skills@ad-conversion-tracking; a channel test on broken tracking measures nothing.
## B2B vs. B2C
| Dimension | B2B | B2C / ecommerce |
| ----------------- | --------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
| Dominant families | Paid search (capture), B2B professional networks, paid review listings, creation via podcast/social | Paid social, short-form video, paid search, retail media, connected TV |
| Deciding metric | Pipeline, ACV, payback | ROAS/MER, contribution margin, AOV |
| Attribution | Last-click breaks over long cycles; self-reported + modeled | MER + platform triangulation + incrementality |
| Audience size | Small TAM, account lists, near platform floors | Broad, high-volume |
| Cadence | Multi-quarter, buying committee | Seasonal, promotional, impulse |
| Creative volume | Fewer assets, longer life (weeks) | High volume, fast fatigue (days at scale) |
Identical for both, explicitly: the affordability gate, the disqualifier pass, the capture/creation logic, the concentration rule, and the test-design discipline all run the same way - only the inputs and the dominant families differ.
## Output shape
Deliver one channel plan document:
1. Economics inputs and the allowable CAC derived from them (with the payback math shown).
2. Deleted channels, naming the gate each failed (affordability, disqualifier, or funding floor) and the specific dated reason.
3. Candidate table ranked by value ÷ effort - survivors only, nothing with a failed gate - with the value and effort subtotals shown, the interview Q15-17 answers that moved any row, and any starved high-value/high-effort family promoted against its rank with the condition that promoted it.
4. Primary channel recommendation with rationale tied to the ratio and the economics.
5. Funding plan: budget vs. the primary's learning threshold.
6. Test design: price-discovery step, real-test budget and duration, pass/fail and kill criteria.
7. Trigger conditions for adding a second channel.
8. Measurement approach implied by the mix.
Full template plus a worked B2B example, a worked B2C example, and a negative example: [references/channel-plan-template.md](references/channel-plan-template.md). If your harness has persistent memory, store the economics inputs, the recommendation, and the second-channel triggers so later budget and scaling conversations start from the same decisions; otherwise tell the user to keep the plan document as the canonical record.
## Failure modes
- Copying a competitor's channel mix without matching economics - their margin and LTV, not their taste, made their mix work.
- Picking the channel the team already knows instead of the one with channel-model fit.
- Ordering the candidate menu by what is cheapest to start rather than by return per unit of effort - the cheapest row is rarely the first row.
- Spreading a small budget across many channels, starving each below its learning threshold.
- Parking a ruled-out family at the bottom of the ranked table instead of deleting it - it returns as scope with its failed gate forgotten.
- Letting the ratio permanently starve the one family that reaches the buyer, when the budget could clear its floor by concentrating.
- Ignoring creative production capacity - the real bottleneck on the highest-volume creation families.
- Launching before tracking or landing pages are ready, wasting the entire learning phase.
- Reading attribution artefacts as performance - over-crediting capture, then over-funding it.
- Treating median benchmarks as targets - they are vendor-published midpoints of very wide distributions.
- Killing a channel before one full sales cycle has elapsed - a false negative, not a verdict.
- Forcing the product into a mismatched channel - products mold to channels, never the reverse.
## Objective and measurement
A channel selection passes only when every recommended channel simultaneously satisfies all four:
1. Realistic benchmark CPA/CPL at or under allowable CAC (affordability gate passed, math shown).
2. Fundable above its learning/signal threshold with the stated budget (funding floor passed).
3. Survives the disqualifier pass for every target geography and category constraint.
4. Carries a documented test with explicit pass/fail and kill criteria and a duration at least as long as the sales cycle.
Any recommendation failing one criterion is rejected and the framework re-run from the failed step - do not present a plan with a known failed gate. Post-launch, the selection itself is validated when the channel test returns a pass verdict at significance within the planned duration; a kill verdict feeds the next-ranked candidate, not a wider spread.
## References
- [references/channel-family-profiles.md](references/channel-family-profiles.md) - per-family profiles in the step 4 default efficiency order: demand state, value ceiling, effort load, minimum viable budgets, signal thresholds, directional cost ranges, disqualifiers, creative burden. Load when gating or ranking candidates.
- [references/disqualifier-checklist.md](references/disqualifier-checklist.md) - regulatory, geography, and capacity disqualifiers with sourced specifics. Load during step 2.
- [references/channel-plan-template.md](references/channel-plan-template.md) - output template, worked B2B and B2C examples, and a negative example. Load when writing the plan.
- mbfinotti/advertising-skills@ad-spend-allocation - use when splitting a budget across existing running campaigns; this skill handles the one-time setup decision only.
- mbfinotti/advertising-skills@ad-spend-guardrails - set CAC ceiling and ROAS floor policy before using this skill; the affordability gate takes those targets as input.
- mbfinotti/advertising-skills@cac-roas-benchmark - benchmark current CAC/ROAS health after channel selection; this skill uses them as gates, not measures.
- mbfinotti/advertising-skills@ad-conversion-tracking - verify tracking before launch; channel testing on broken tracking measures nothing.
- mbfinotti/advertising-skills@ad-creative-test-plan - after picking a channel, test creative variants; this skill validates the channel only.
- mbfinotti/advertising-skills@paid-media-scaling - scale a proven channel; this skill chooses the first one only.