mbfinotti/partnerships-skills已通过检查
SKILL DETAIL
partner-economics
mbfinotti/partnerships-skills/partner-economics
Model the unit economics of one partner relationship - partner P&L, margin modeling, cost-to-serve, referral-fee and lifetime-value split, channel partner ROI, partner CAC vs direct CAC, ramp and payback - to decide whether to sign, scale, renegotiate, or exit that partner. Covers B2B channel and alliance partners plus B2C affiliate and creator economics. Use whenever the user mentions partnership unit economics, a single-partner business case, reseller, VAR, MSP, OEM or marketplace deal economics, revenue-share and margin-stack analysis, or asks whether a partner is worth pursuing, even if they never say economics. Do NOT use for ranking a portfolio of candidates - use mbfinotti/partnerships-skills@alliance-prioritization instead.
安装量 · 507查看来源
Installation
npx skills add https://github.com/mbfinotti/partnerships-skills --skill partner-economics
技能文件
SKILL.md
最近同步 · 2026年10月23日
evals/evals.json›
{
"skill_name": "partner-economics",
"evals": [
{
"id": 1,
"prompt": "Sanity-check me before I take this to our CFO on Thursday. Brightpath Analytics, we sell a $38,000 ACV analytics platform, 74% gross margin, and our direct CAC is $9,500. We have a reseller called Calderon Systems on the table at a 30% discount off list, and their plan says 12 closed deals in year two. To acquire those 12 we would spend $24,000 MDF, about 0.3 of a partner manager which is $39,000 loaded, $15,000 of enablement and certification delivery, $9,000 of extra support load, and $2,000 of portal allocation. That is roughly $7,400 a head against our $9,500 direct CAC, so a 22% acquisition win. Am I reading that right?",
"expected_output": "A partner economics analysis that rejects the per-customer CAC comparison, restates both sides as acquisition cost per $1 of net revenue, finds the partner more expensive than direct rather than 22% cheaper, and lands on renegotiate or do-not-sign at 30%.",
"files": [],
"expectations": [
"Rejects the per-customer comparison of roughly $7,400 against $9,500 as invalid, because the 30% discount lowers revenue per partner deal so the two figures price different goods",
"States net revenue per partner deal as $26,600, being $38,000 less the 30% discount",
"States year-two partner net revenue as $319,200 and the discount value conceded as $136,800",
"Expresses direct acquisition cost as $0.25 per $1 of net revenue, from $9,500 on a $38,000 ACV",
"Expresses partner acquisition cost per $1 of net revenue using the $89,000 of non-discount cost over $319,200 of net revenue, landing between $0.27 and $0.29",
"Concludes the partner is more expensive per $1 of net revenue than the direct motion, reversing the user's claimed 22% win",
"Names which costing convention it used - revenue booked gross with the discount charged to acquisition cost, or revenue booked net of discount with the discount excluded - and does not mix the two inside one comparison",
"Does not declare one of the two costing conventions more correct, more conservative, or preferred over the other",
"States the go/no-go threshold the model is judged against before delivering any verdict",
"Addresses incrementality explicitly rather than assuming every Calderon deal is net-new",
"Reports a sensitivity pass naming which assumptions flip the verdict, with effective discount among them",
"Delivers a renegotiate or do-not-sign verdict at 30%, naming the discount level or deal count that would clear the direct benchmark, instead of recommending signature on the user's figures"
]
},
{
"id": 2,
"prompt": "Quick one. Halverson Group is our best reseller - they closed $2.4M of list ACV for us last year at a 22% discount. Our average deal is $60K, gross margin is 68%, and customers stick around about five years, so partner LTV is $300K a head. On top of the discount we pay them $180,000 MDF, and internally they cost us one full-time partner manager at $140,000, $70,000 of certification delivery and $40,000 of extra support. I want to put 'Halverson contributed $2.4M and each customer is worth $300K' in the Q1 partner review. Anything wrong with that?",
"expected_output": "A restatement that strips the headline revenue down to net contribution across the full cost stack, recomputes lifetime value on gross margin instead of revenue, and challenges the asserted five-year lifetime.",
"files": [],
"expectations": [
"Refuses to report $2.4M as Halverson's contribution and restates it net of the full partner cost stack",
"Names every cost bucket in the restatement: the 22% discount, the $180,000 MDF, the $140,000 of partner-manager time, the $70,000 of certification delivery and the $40,000 support uplift",
"States the 22% discount in money as $528,000",
"Reports total partner cost of $958,000 against the $2.4M headline",
"Reports restated revenue net of partner cost below $1.5M rather than $2.4M",
"Rejects the $300,000 lifetime value because it is computed on revenue rather than on gross margin",
"Recomputes lifetime value on the 68% gross margin, giving roughly $204,000 per customer on the stated five-year assumption",
"Applies gross margin once, not twice, in the lifetime-value calculation",
"Flags the five-year lifetime as asserted with no churn basis behind it",
"Asks the user for the churn rate or a measured average customer lifetime for this segment instead of substituting an assumed figure",
"Separates partner-sourced from partner-influenced revenue, or asks which share of the $2.4M is which, instead of accepting one blended attributed number",
"States that the review should carry annual net contribution rather than headline revenue, and lists what remains unverified in the supplied figures"
]
},
{
"id": 3,
"prompt": "I am building the board deck for our channel program review next month. Opening slide I want three stats I pulled off a channel blog: 80% of channel revenue comes from the top 20% of partners, only 11% of partners hit their incentive goals, and most partners never transact at all. Then the ask: we have 140 signed partners, 19 of them closed anything last year, and I want approval to cut the bottom 40 by tenure - anyone signed more than three years ago with under $50K. Help me tighten the argument.",
"expected_output": "A rewrite of the board slide that refuses all three untraceable statistics, substitutes attributable figures with their evidence class, and replaces tenure-based pruning with net contribution.",
"files": [],
"expectations": [
"Flags the 80/20 channel-revenue concentration figure as untraceable and refuses to present it as fact",
"Flags the 11% incentive-attainment figure as untraceable and refuses to present it as fact",
"Flags 'most partners never transact' as untraceable, noting that the circulating variants trace to no original study",
"Offers the Forrester 2016 finding that 14% of channel incentives meet their goals as the attributable substitute, labeled as an analyst survey",
"Offers the CRN Research 2025 Partner Journey Study figure - 70% of partners terminated or dormant within twelve months, split 29% terminated and 41% stopped selling - as the best-attributed adjacent number",
"Attaches an evidence class to every benchmark it allows into the deck",
"States that the direction of heavy partner concentration is credible even though the specific 80/20 figure is not",
"Rejects tenure as the pruning criterion and replaces it with net contribution per partner",
"States that 140 signed partners with 19 active is a 19-partner channel",
"Computes or requests revenue per partner on the active count only, warning that dividing by all 140 flatters the program",
"Requires each partner proposed for cutting to carry a fully-loaded cost-to-serve figure, not just its revenue",
"Does not invent a replacement statistic where no attributable source exists, writing unverified instead",
"Notes that a partner below $50K may still be net-contribution positive if it costs almost nothing to serve"
]
},
{
"id": 4,
"prompt": "Tessura has been our AWS Marketplace co-sell partner for two years. Last year our CRM shows $3.1M of Tessura-attributed closed-won. My CFO does not believe it is real revenue and wants proof before we renew. My plan is to pause the whole Tessura motion for a quarter, watch what happens to bookings, and bring him the number. Our agreement with them runs 12 months more with EMEA exclusivity and a $750,000 annual minimum we owe them. How would you design the test?",
"expected_output": "An incrementality plan that starts from an assumed-rate band rather than the proposed pause, deletes the pause outright on contractual grounds, and ranks the remaining measurement options by evidence bought per unit of effort.",
"files": [],
"expectations": [
"Separates attributed revenue from incremental revenue rather than treating the $3.1M as the quantity in question",
"Cites Blake, Nosko and Tadelis (2015, Econometrica) and the eBay result that 99.5% of brand-keyword clicks were retained after paid search was halted",
"Does not open by endorsing the channel pause the user proposed",
"Recommends the assumed-rate band as the first step, stating the assumed incrementality rate as a named assumption",
"Shows the verdict at 100%, 70% and 40% incremental rather than at a single assumed rate",
"States that buying an experiment is justified only when the verdict flips inside the plausible range",
"Deletes the channel pause from the menu outright rather than ranking it last, because suspending the motion would breach the EMEA exclusivity and the $750,000 annual minimum",
"Names the specific contract terms that killed the pause, recording it as deleted rather than merely unattractive",
"Ranks the remaining options with the assumed-rate band ahead of matched holdout, and matched holdout ahead of geo split, on evidence bought per unit of effort",
"States that matched holdout and assumed-rate band carry no compliance cost because both run inside the vendor's own systems",
"Notes that a geo split amends the rules of engagement and is visible to the partner",
"States that the academic literature on channel incrementality versus cannibalization is unsettled, documenting both synergy and substitution",
"Reports the incrementality estimate with its evidence class rather than as a bare percentage"
]
},
{
"id": 5,
"prompt": "Setting a ceiling on what we can pay a referral partner per closed customer. Lifetime revenue per customer is $2,400, gross margin 65%, and our target CAC is $350. A partner is asking for $900 per closed customer, one time. Separately our affiliate platform rep told us SaaS programs pay 20-30% recurring, so he thinks $720 is the safe number. What is our actual ceiling and do we take the $900?",
"expected_output": "A ceiling of $1,210 computed from lifetime revenue rather than margin-adjusted LTV, approval of the $900 ask, and the platform rep's benchmark demoted to a labeled vendor figure that does not bind the decision.",
"files": [],
"expectations": [
"Computes the maximum affordable payout per customer as lifetime revenue multiplied by gross margin, minus target CAC",
"Uses the $2,400 lifetime revenue figure as the input, not a margin-adjusted lifetime value",
"Reports a ceiling of $1,210",
"Does not apply the 65% gross margin twice, and does not report a ceiling near $664",
"Approves the $900 ask as sitting below the ceiling",
"States the headroom between the $900 ask and the $1,210 ceiling as $310",
"Labels the 20-30% recurring SaaS affiliate range as a vendor benchmark supplied by an affiliate platform, not as a constraint on this decision",
"States that the ceiling, not the platform rep's benchmark, is what binds the payout",
"Explains that any payout above the ceiling buys revenue at a loss",
"Names what would move the ceiling: a change in lifetime revenue, in gross margin, or in the target CAC",
"Asks whether the $2,400 lifetime revenue figure is measured or assumed rather than treating it as verified"
]
},
{
"id": 6,
"prompt": "We are a B2B SaaS company - project management for construction firms, $9,600 average annual contract - running a 25% recurring revenue-share affiliate program with 4,100 signed affiliates. Priya Raghunathan, a construction-tech newsletter writer, drove $310,000 of tracked revenue last year on her own. She is asking for a custom 35% rate. Our next two biggest earners are a SaaS coupon site and a cashback browser extension. Build me the economics on whether to give Priya 35%, and while you are at it I would like a P&L for each of the 4,100 so I can see who else deserves a bump.",
"expected_output": "An individually modeled case for the head affiliate on effective rather than stated rev-share, cohort-level treatment of the 4,100-partner tail, and segment-level incrementality that isolates the coupon and cashback partners.",
"files": [],
"expectations": [
"Models the payout on an effective revenue-share rate rather than on the stated 25% or requested 35% headline rate",
"States that effective revenue share typically runs 20-40% below the stated rate after refunds, chargebacks, fraud reversals and non-qualifying actions, and labels that range as vendor or practitioner class",
"Refuses to build 4,100 individual partner P&Ls and models the long tail at cohort level instead",
"Models Priya individually, treating the head of the distribution the way a B2B reseller would be modeled",
"Uses B2C per-action mechanics even though the product is sold to businesses, on the grounds that the money flow rather than the buyer decides the model",
"Does not apply a 6-12 month ramp or a double-cost enablement period to the affiliate motion",
"Estimates incrementality separately for the coupon site and the cashback extension rather than for the program as a whole",
"Identifies coupon and cashback partners as the textbook case of attributed but non-incremental volume",
"Builds a cost stack that still includes tracking or platform fees, fraud losses, payment costs, creative production and program-management time",
"Computes lifetime value on gross margin, and asks for the gross margin and churn figures that the prompt does not supply",
"Reports what Priya's $310,000 of tracked revenue becomes once the effective rate and validation deductions are applied, instead of carrying the tracked figure into the verdict unadjusted",
"States the threshold the 35% decision is judged against before recommending for or against it",
"Reports the incrementality assumption behind Priya's own traffic with its evidence class rather than assuming her tracked revenue is fully incremental"
]
},
{
"id": 7,
"prompt": "We sell a $120,000 list-price industrial IoT platform direct in North America, 71% gross margin, direct CAC around $26,000. To open LATAM we are being pitched a two-tier structure: regional distributor Caravela takes 12% of whatever the resellers pay them, and the resellers under Caravela buy at 30% off our list. Our VP Sales says it is a 30% discount, the distributor's cut comes out of the reseller's side, so nothing changes for us. Model it.",
"expected_output": "A full end-to-end stack model showing the vendor nets roughly $73,920 rather than $84,000, naming double marginalization as the cause and the two-part tariff as the fix, with the channel-mode question raised before the economics settle it.",
"files": [],
"expectations": [
"Rejects the claim that the distributor's cut comes out of the reseller's side at no cost to the vendor",
"Models the full two-tier stack end to end rather than one tier at a time",
"States the net price the vendor actually receives per unit as roughly $73,920 against the $120,000 list",
"States the effective all-in discount as roughly 38%, not the 30% quoted by the VP Sales",
"Names double marginalization, and Spengler's 1950 work, as the root cause of margin stacking",
"Offers the two-part tariff as the peer-reviewed coordination fix, citing Jeuland and Shugan (1983) or McGuire and Staelin (1983)",
"Labels the 8-15% distributor margin range and the 25-35% VAR gross margin range as vendor or practitioner class, not as settled benchmarks",
"Compares acquisition cost per $1 of net revenue against the direct motion rather than comparing per-customer figures",
"Raises whether a partner channel is the right mode for LATAM entry at all before the unit economics settle it, referencing ally-versus-acquire or the transaction-cost make-versus-buy test",
"Builds a cost-to-serve stack for the LATAM motion that goes beyond the discount alone",
"Asks for the deal volume, ramp expectation and contract terms the model needs rather than assuming them",
"Runs a sensitivity pass naming which assumptions flip the LATAM verdict"
]
},
{
"id": 8,
"prompt": "Board meeting in nine days and I need a recommendation. Our reseller Oakhurst Technologies is up for renewal at their current 35% discount and I want to push them to 28%. Two complications: Oakhurst has told me twice they will not open their own books to us, and I have no finance analyst free this quarter, so it is me alone on this. None of their deals route through a marketplace - everything is a direct paper contract. What is the right way to build this?",
"expected_output": "A recommendation of the vendor-side contribution model, with the dual-sided P&L and the marketplace-mechanics layer deleted and recorded as deleted with their reasons, plus a named limitation about arguing a split from one ledger.",
"files": [],
"expectations": [
"Deletes the dual-sided partner P&L from the options rather than listing and demoting it, because Oakhurst will not disclose its own numbers",
"States the second independent reason the dual-sided model is gone: no finance reviewer is assigned",
"Deletes the marketplace-mechanics layer because no deal routes through a marketplace",
"Records each deleted option as deleted and why, so it does not return as scope at a later review",
"Recommends the vendor-side contribution model as the approach that survives the stated constraints",
"States that the decision at stake is the split rather than the signature, which would ordinarily promote the dual-sided P&L",
"Names the resulting limitation: the split cannot be argued honestly from the vendor's ledger alone",
"Rules out any incrementality experiment on the nine-day timeline and does not propose a matched holdout, a geo split or a channel pause",
"Recommends the assumed-rate band for incrementality as the only option that fits inside nine days",
"Asks its clarifying questions one at a time rather than batching them into a single list",
"Asks for the current direct CAC and the average deal size it buys, without which there is no threshold to test",
"Asks for product gross margin and churn or average customer lifetime before computing any lifetime value",
"Fixes the go/no-go threshold for the 28% ask before computing anything"
]
},
{
"id": 9,
"prompt": "Enterprise prospect Kessler Freight wants to buy our platform through AWS Marketplace so it retires their EDP commit - $1.8M over three years, as a private offer. Our AE's framing: the marketplace costs us 3%, the customer gets their committed spend burned down so it is basically free money for them, and sales wants an 18% discount on top to get it signed this quarter. One wrinkle, the AWS partner rep asked whether we are hosted entirely on AWS. Our data plane is on AWS but our control plane runs on GCP. Model this deal for me.",
"expected_output": "A routed-deal cost stack in which the 18% win discount dwarfs the marketplace fee, the commit-drawdown 'free money' framing is refused, and the split-plane architecture is flagged against the eligibility restriction.",
"files": [],
"expectations": [
"Models the 18% win discount and the marketplace fee inside one cost stack rather than treating the fee as the cost of the marketplace route",
"States the 18% discount as roughly $324,000 against a marketplace fee of roughly $54,000 or less, so the discount dwarfs the fee",
"Uses the 1.5-3% private-offer fee band rather than assuming the roughly 3% public SaaS listing rate applies",
"Labels the marketplace fee figures as platform or vendor documentation, the best-attributed class in this domain, and says to verify the current published schedule before quoting it",
"Rejects the 'free money' framing of commit drawdown",
"Explains that a qualifying marketplace purchase retires commit at full price, while direct consumption of the same value would cost the buyer less after a negotiated discount",
"Flags the eligibility risk created by the split architecture: commit retirement was restricted to SaaS hosted entirely on AWS effective May 2025",
"States that eligibility must be confirmed for this specific product rather than assumed for the SaaS category",
"Identifies the control plane running on GCP as exactly the configuration that strands commit-retirement eligibility",
"States that marketplace mechanics settle only the routed-deal cost stack and are a layer on top of a vendor-side contribution model, never a substitute for one",
"Computes acquisition cost per $1 of net revenue for this routed deal rather than reporting a per-customer figure",
"Names what happens to the deal thesis if commit retirement turns out to be ineligible, since the buyer's entire motive disappears",
"Lists open questions covering the unverified fee schedule and the unconfirmed eligibility ruling"
]
},
{
"id": 10,
"prompt": "Nine months ago we signed Vantree, an MSP, on the back of a business case I built. Modeled partner CAC was $6,200; actuals say $9,800. Direct CAC is $8,400. Vantree-sourced customers are churning at 19% annually against 7% for our direct-sourced base. Our CRO says MSP relationships take 18 months and we should give it more time - we have also sunk about $210,000 into enabling them. What do I do?",
"expected_output": "A diagnostic worked in cheapest-first order on the CAC overrun, a retention-gap read against the practitioner threshold and the survey figure, an assumption audit, and a prune-or-continue call made on net contribution rather than tenure or sunk cost.",
"files": [],
"expectations": [
"Works the partner-CAC-exceeds-direct diagnostic in a stated order, cheapest cause first",
"Checks attribution double-counting of influenced deals first, as a re-run of the query at near-zero effort",
"Checks the commission or discount rate against deal size second, as roughly an hour of arithmetic against a term that can be renegotiated",
"Checks partner lead quality last, as the slowest fix, needing cohort win-rate and retention data",
"Identifies the 12 percentage point retention gap between 19% and 7% as beyond the 10-15pp practitioner threshold that signals ICP misalignment",
"Cites the SaaS Capital survey of August 2019 finding channel-majority companies running roughly 3pp lower retention, showing how far outside survey-typical this gap sits",
"Labels the 10-15pp threshold as a practitioner rule of thumb rather than a measured standard",
"Recomputes lifetime value on the 19% partner-sourced churn rather than on the direct-cohort figure used in the original model",
"Rejects the $210,000 of sunk enablement spend as a reason to continue",
"Rejects tenure and the CRO's 18-month convention as the decision basis, and prunes or continues on net contribution",
"Runs an assumption audit naming which modeled inputs proved wrong and by how much, as the input to the next model",
"Distinguishes judging the original economic decision from running Vantree's ongoing performance scorecard, and points the ongoing scorecard at the partner performance review discipline",
"Compares realized partner CAC of $9,800 against both the $6,200 projection and the $8,400 direct figure on a like-for-like basis rather than per customer alone"
]
}
],
"trigger_queries": [
{ "query": "Model the unit economics of one partner relationship before we sign them", "should_trigger": true },
{ "query": "Is this reseller worth signing at a 30% discount off list?", "should_trigger": true },
{ "query": "Build a partner P&L for our biggest VAR", "should_trigger": true },
{ "query": "What is our partner CAC versus our direct CAC?", "should_trigger": true },
{ "query": "how do i know if this channel partner is actually making us money", "should_trigger": true },
{ "query": "We are two years into a marketplace partnership and I think most of the revenue is not incremental", "should_trigger": true },
{ "query": "Build both sides of the P&L for an MSP alliance so we can renegotiate the split", "should_trigger": true },
{ "query": "Should we keep paying this distributor 12%?", "should_trigger": true },
{ "query": "Work out whether a referral fee of $900 per closed customer is affordable", "should_trigger": true },
{ "query": "What is the maximum we can pay a partner per closed deal without losing money?", "should_trigger": true },
{ "query": "our channel partner brought in $2M last year, is that good", "should_trigger": true },
{ "query": "Cost to serve analysis for one of our resellers", "should_trigger": true },
{ "query": "I need a business case for signing an OEM embed deal", "should_trigger": true },
{ "query": "Model the margin stack across vendor, distributor and reseller", "should_trigger": true },
{ "query": "Is our top affiliate worth a custom rate?", "should_trigger": true },
{ "query": "Partner ROI calculation for the QBR", "should_trigger": true },
{ "query": "how long until this reseller pays for itself", "should_trigger": true },
{ "query": "Ramp and payback model for a new channel partner", "should_trigger": true },
{ "query": "Compare the acquisition cost of our partner channel against direct sales", "should_trigger": true },
{ "query": "We want to cut a partner - help me prove it is the right call", "should_trigger": true },
{ "query": "Our AE says the marketplace only costs us 3%, is that the whole story?", "should_trigger": true },
{ "query": "What does a 25% discount plus MDF plus a partner manager actually cost us per deal?", "should_trigger": true },
{ "query": "figure out if we should renew this partner or walk away", "should_trigger": true },
{ "query": "Build the numbers behind a reseller renegotiation", "should_trigger": true },
{ "query": "Does this partnership pencil out?", "should_trigger": true },
{ "query": "How much of our partner-attributed revenue would have closed anyway?", "should_trigger": true },
{ "query": "Lifetime value of customers our partner brings versus ones we source ourselves", "should_trigger": true },
{ "query": "Justify to my CFO that the channel is cheaper than direct", "should_trigger": true },
{ "query": "i am not sure this partner is worth the headcount we put on them", "should_trigger": true },
{ "query": "Write the business case for scaling our top VAR", "should_trigger": true },
{ "query": "Our partner CAC came out higher than direct - what did I get wrong?", "should_trigger": true },
{ "query": "Economics of a two-tier channel structure in LATAM", "should_trigger": true },
{ "query": "Should we give this creator a 35% revenue share instead of 25%?", "should_trigger": true },
{ "query": "Model an affiliate payout against our gross margin to see what we can afford", "should_trigger": true },
{ "query": "How much commission can we afford on a customer worth $2,400 in lifetime revenue?", "should_trigger": true },
{ "query": "What is the real net contribution from our largest channel partner?", "should_trigger": true },
{ "query": "is the money we make from this alliance worth the effort", "should_trigger": true },
{ "query": "I need a memo my CFO will sign off on about one partner deal", "should_trigger": true },
{ "query": "Break-even month for a partner we onboard in March", "should_trigger": true },
{ "query": "Quantify the double-cost period while we enable a new reseller", "should_trigger": true },
{ "query": "Our reseller wants better margin - can we afford it?", "should_trigger": true },
{ "query": "how do i split value with a partner so they stay motivated to keep selling", "should_trigger": true },
{ "query": "Test whether our channel revenue is incremental or cannibalised", "should_trigger": true },
{ "query": "Should we sign this agency on a referral fee or a reseller discount?", "should_trigger": true },
{ "query": "What is the break-even discount on a channel deal?", "should_trigger": true },
{ "query": "Model a private offer through a cloud marketplace including the win discount", "should_trigger": true },
{ "query": "Evaluate whether a commit-drawdown deal is actually good for us", "should_trigger": true },
{ "query": "we spent 200k enabling this partner and they have closed nothing, now what", "should_trigger": true },
{ "query": "Decide whether to scale, renegotiate, or exit one specific partner", "should_trigger": true },
{ "query": "Partner margin modelling for next year's plan", "should_trigger": true },
{ "query": "I need to know if the top affiliate in our program is profitable", "should_trigger": true },
{ "query": "Put numbers on whether a 6% OEM royalty works for us", "should_trigger": true },
{ "query": "Is a 40% reseller discount sustainable at our gross margin?", "should_trigger": true },
{ "query": "Build the financial case for exiting a channel partner", "should_trigger": true },
{ "query": "What should I put in a single-partner business case?", "should_trigger": true },
{ "query": "how much does this partner actually cost us once you count everything", "should_trigger": true },
{ "query": "Work out the payback period on what we invested in one partner", "should_trigger": true },
{ "query": "Does the split with this partner motivate them to keep contributing?", "should_trigger": true },
{ "query": "Analyse one partner's contribution net of discount, MDF and support load", "should_trigger": true },
{ "query": "Rank these eight candidate technology alliances and tell me which three to pursue", "should_trigger": false },
{ "query": "Which of our twelve candidate partners should we go after first?", "should_trigger": false },
{ "query": "Design the commission tiers for our affiliate program", "should_trigger": false },
{ "query": "What rate model should our affiliate program use - revenue share or flat CPA?", "should_trigger": false },
{ "query": "Set the cookie window and clawback rules for our affiliate program", "should_trigger": false },
{ "query": "Draft the terms and conditions affiliates accept when they join", "should_trigger": false },
{ "query": "Audit this month's affiliate payout run before we disburse it", "should_trigger": false },
{ "query": "Build a dashboard spec for affiliate performance", "should_trigger": false },
{ "query": "Write detection rules for cookie stuffing and fake lead submissions", "should_trigger": false },
{ "query": "Design the onboarding sequence for newly approved affiliates", "should_trigger": false },
{ "query": "Write the recruitment email we send to prospective affiliates", "should_trigger": false },
{ "query": "Check this sponsored video script against FTC disclosure rules", "should_trigger": false },
{ "query": "Design our partner program from scratch - tiers, benefits, motions", "should_trigger": false },
{ "query": "How many partner tiers should we have and what goes in each?", "should_trigger": false },
{ "query": "Write the promotion and demotion rules for our gold tier", "should_trigger": false },
{ "query": "Build the per-partner scorecard and QBR structure for our program", "should_trigger": false },
{ "query": "Our partner missed targets two quarters running - what is the corrective action ladder?", "should_trigger": false },
{ "query": "Write our rules of engagement for contested deals between direct and partners", "should_trigger": false },
{ "query": "Design the deal registration policy and its protection window", "should_trigger": false },
{ "query": "How should partner-sourced and partner-influenced credit be split for sales comp?", "should_trigger": false },
{ "query": "Plan a joint go-to-market campaign with our new partner", "should_trigger": false },
{ "query": "Which marketplaces should we list on next year?", "should_trigger": false },
{ "query": "Sequence the certification tracks our partners need to become self-sufficient", "should_trigger": false },
{ "query": "Map our entire partner ecosystem and find the coverage gaps", "should_trigger": false },
{ "query": "Which partner category should we add next - SIs or MSPs?", "should_trigger": false },
{ "query": "Set reward caps and velocity limits for our refer-a-friend program", "should_trigger": false },
{ "query": "Should we reward the referrer, the friend, or both?", "should_trigger": false },
{ "query": "Prepare me to negotiate rates and usage rights with a creator", "should_trigger": false },
{ "query": "Define the KPIs for our upcoming influencer campaign", "should_trigger": false },
{ "query": "Write the creative brief for a creator we have already signed", "should_trigger": false },
{ "query": "Build the criteria we use to shortlist candidate creators", "should_trigger": false },
{ "query": "Draft the first pitch message to a newsletter writer we want to partner with", "should_trigger": false },
{ "query": "I am moving into partnerships from sales - what does the career path look like?", "should_trigger": false },
{ "query": "Write the job description and interview loop for a channel manager", "should_trigger": false },
{ "query": "I am starting a new partnerships project, where do I begin?", "should_trigger": false },
{ "query": "Configure partner offers on our agent marketplace listing", "should_trigger": false },
{ "query": "Set up a Shopify Partner account and publish an app listing", "should_trigger": false },
{ "query": "Calculate our blended CAC across paid search, outbound and organic", "should_trigger": false },
{ "query": "Model our SaaS unit economics for the next fundraise", "should_trigger": false },
{ "query": "What should our self-serve pricing tiers be?", "should_trigger": false },
{ "query": "Forecast next quarter's pipeline by stage", "should_trigger": false },
{ "query": "Build a cohort retention analysis for our self-serve customers", "should_trigger": false },
{ "query": "Our LTV to CAC ratio is 2:1 - how do we fix the paid acquisition mix?", "should_trigger": false },
{ "query": "Write the press release announcing our new partnership", "should_trigger": false },
{ "query": "Negotiate a lower rate with our cloud hosting vendor", "should_trigger": false },
{ "query": "Model the economics of hiring two more SDRs", "should_trigger": false },
{ "query": "Run a valuation on an acquisition target's ARR multiple", "should_trigger": false },
{ "query": "Model the economics of our field sales team versus inside sales", "should_trigger": false },
{ "query": "Design the compensation plan for our channel account managers", "should_trigger": false },
{ "query": "Draft the legal clauses of our reseller agreement with counsel", "should_trigger": false },
{ "query": "Which CRM fields do we need to track partner deals?", "should_trigger": false },
{ "query": "Set up partner attribution reporting in our CRM", "should_trigger": false },
{ "query": "Write a QBR agenda for our top ten partners", "should_trigger": false },
{ "query": "Train our partner managers on objection handling", "should_trigger": false },
{ "query": "What commission plan should we pay our own account executives?", "should_trigger": false },
{ "query": "Analyse the churn drivers in our direct customer base", "should_trigger": false },
{ "query": "Compare our product gross margin to industry benchmarks", "should_trigger": false },
{ "query": "Plan a co-branded webinar with a technology partner", "should_trigger": false },
{ "query": "Recruit resellers for us in the DACH region", "should_trigger": false }
]
}
references/b2b-vs-b2c-contrast.md›
# B2B vs B2C Partner Economics
## What is identical
The method does not change: attributable revenue minus fully-loaded partner cost, against partner CAC, over a lifetime, adjusted for incrementality. Never rebuild the method per side; only re-parameterize it.
The discipline rules carry over too:
- Sourced/influenced separation.
- Margin-based LTV.
- Incrementality never assumed.
- Benchmarks labeled by evidence class.
One structural note worth knowing: the B2B channel/alliance literature and the B2C affiliate/creator literature are two largely separate practitioner bodies that do not cross-cite each other. Expect no shared vocabulary between a channel chief and an affiliate manager describing the same math.
## What genuinely differs
| Dimension | B2B channel/alliance | B2C affiliate/creator |
| ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------- |
| Unit of analysis | Partner firm; few, large, negotiated | Creator/publisher; thousands, self-serve |
| Money flow | Discount off list, rebates, MDF, marketplace fees, royalties | CPA per action or revenue share on recurring |
| Deal economics floor | Two thresholds, not one: a VAR/SI's own margin starts covering their acquisition and delivery around ~$20K ACV, but the vendor can only fund the full reseller/SI motion from ~$50K. Referral/MSP motions work down to ~$5K ACV. All blog-tier and unaudited; `mbfinotti/partnerships-skills@partner-channel-program` owns the ACV band table - cite it rather than restating a floor here | Works at any price point; per-action payout scales down |
| Ramp and enablement cost | 6-12+ months of double cost, certification, dedicated headcount | Near zero; activation is instant, enablement is a creative kit |
| Attribution | Deal registration, CRM opportunity tagging, fixed window, one partner per deal | Cookie/attribution window per action; last-click disputes; fraud screening |
| Concentration | A handful of partners carry the channel | Power law with a very long tail of near-zero earners |
| Contract shape | Negotiated agreements, exclusivity, clawback, MDF commitments | Standardized terms accepted at signup; testable at scale |
| Lifecycle review | JBP + QBR, relationship-bound | Automated dashboards; cohort-level review, not per-partner meetings |
## The effective-rate trap, per side
- B2B: margin stacking. Each tier's markup compounds (Spengler's double marginalization); a two-tier structure must be modeled as one stack end to end, or the end price silently goes uncompetitive.
- B2C: stated vs effective revenue share. Practitioners warn the effective rev-share rate typically runs **20-40% below the stated rate** after deductions - refunds, chargebacks, fraud reversals, non-qualifying actions (vendor/practitioner source; label it, but the direction is well-attested). Model payouts on the effective rate, never the headline rate.
## Parameter defaults when modeling B2C
- Revenue mechanics: SaaS affiliate 20-30% recurring share; ecommerce 10-15% of first order; lead-gen/fintech $50-200 flat CPA (vendor benchmark class - label in the memo).
- Cost stack shrinks but does not vanish: platform/tracking fees, fraud losses, payment costs, creative production, and program-management time replace MDF and enablement.
- Incrementality bites harder, not softer: coupon and cashback affiliates are the textbook case of attributed-but-not-incremental volume - the Blake/Nosko/Tadelis lesson applies with full force. Model incrementality by affiliate segment, not for the program as a whole.
- Per-partner modeling only pays for the head of the distribution: model a top creator or publisher individually the way B2B models a reseller; handle the long tail at cohort level.
## Hybrid cases
- An agency that both refers B2B deals and runs affiliate links: model the two flows separately - they carry different attribution rules, different effective rates, and different incrementality profiles. One blended number hides which motion works.
- A B2B vendor paying influencers/creators for developer or SMB reach: use the B2C mechanics (per-action payout, effective-rate discounting, fraud screening) even though the product is B2B - the money flow, not the buyer, decides the model.
references/benchmarks-and-figure-grading.md›
# Benchmarks and Figure Grading
Partner-economics benchmarks graded by how much weight each one carries. Almost every widely repeated figure in this domain is vendor marketing or practitioner rule-of-thumb rather than research, so most of them sit low on the evidence ladder.
- Quote nothing from this file without its evidence-class label; prefer the best-attributed rows.
- When a number the user needs is not here and cannot be verified, write "unverified, user-supplied" in the memo. Never invent one.
Evidence-class ladder, strongest first: peer-reviewed → vendor/platform documentation → industry survey (named, disclosed sample) → analyst (beware vendor-commissioned) → vendor benchmark/blog → untraceable.
## Best-attributed numbers: prefer these
| Benchmark | Value | Source | Evidence class |
| ------------------------------------------------------------ | ----------------------------------------------- | -------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Cloud marketplace listing fee | ~3% public SaaS; 1.5-3% private offers | AWS documentation, 2024; Azure/GCP in the same band | Vendor documentation: the most reliable numbers in this domain |
| Channel CAC per $1 new ACV | $0.53 | 2014 Pacific Crest / KeyBanc Private SaaS Survey | Industry survey; the companion "$1.02 field sales" figure does not appear in the published write-up |
| Partners terminated or dormant within 12 months | 70% (29% terminated + 41% stopped selling) | CRN Research 2025 Partner Journey Study | Industry survey; full methodology not public |
| Channel-majority retention gap | ~3pp lower revenue retention vs direct-majority | SaaS Capital survey, Aug 2019, 700+ private B2B SaaS companies | Industry survey: named, disclosed sample and date |
| Channel incentives meeting their goals | 14% | Forrester 2016 survey | Analyst survey |
| eBay brand-keyword clicks retained after halting paid search | 99.5% | Blake, Nosko & Tadelis 2015, Econometrica | Peer-reviewed: the incrementality anchor |
| Alliance failure within 24 months | 48% of 1,592 alliances (1993-1997) | Dyer, Kale & Singh 2004, HBR, citing empirical studies | Academic-adjacent: but one point inside a 20-80% published spread that turns on how failure is defined. Quote it with the spread and the definition, never as the rate; `mbfinotti/partnerships-skills@alliance-prioritization` owns this number's treatment |
## Usable with labels: vendor/practitioner class
| Benchmark | Value | Evidence class and caveat |
| --------------------------------- | ------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------- |
| Partner CAC advantage vs direct | 20-40% lower typical; "up to 50%" claimed | Vendor/practitioner; only the Pacific Crest survey has a disclosed dataset behind any variant |
| VAR gross margin | 25-35% (20-40% outer band) | Convergent across independent vendor blogs; unaudited |
| Distributor margin | 8-15% | Vendor/practitioner |
| Deal-registration margin uplift | +5 to +15pp over base tier discount | Convergent blog-tier; moderately trustworthy rule of thumb |
| OEM royalty | 1-10% of license revenue | Vendor/practitioner, wide variance |
| Affiliate rates | SaaS 20-30% recurring; ecommerce 10-15% first order; fintech $50-200 CPA | Vendor benchmark (affiliate-platform vendors) |
| Partner activation rate | 30-50% typical; <20% unmanaged | Vendor benchmark (PRM vendor) |
| MDF unused | ~60% quarterly | Vendor survey (worldwide channel survey); the "$50B/half unused" figure cites no source |
| Partner ROI / payback health line | >3x ROI, CAC payback <120 days | Practitioner blog benchmark, unaudited |
| Ecosystem services multiplier | $6.40-$7.13 services per $1 platform revenue | Analyst, vendor-commissioned (Canalys/Omdia for AWS et al.): directional and interested |
| MSP managed-services gross margin | 45-60%; top quartile >60% | Vendor benchmark (Service Leadership Index, subscription data) |
| Effective B2C rev-share vs stated | 20-40% below stated rate after deductions | Vendor/practitioner |
## Untraceable: flag or omit, never quote as fact
- **"80% of channel revenue comes from 20% of partners"** and **"only 11% of partners hit incentive goals"**: originate from a media-company blog citing nothing. Widely repeated, never sourced. The direction (heavy concentration) is credible; the specific figures are not.
- **"Most partners never transact"**: circulating variants span "<30% ever generate a deal" to "80% generate zero revenue," none traceable to an original study. The best-attributed adjacent number is the CRN 70% churn/dormancy figure above.
- Partner-sourced win-rate deltas (+46% sourced, +99% influenced, −11% time-to-close): repeated across vendor content with no traceable methodology.
- "Channel sales up to 50% lower CAC than field sales" as a flat claim: traces loosely to Pacific Crest but is quoted far beyond what that survey supports.
## Genuinely unsettled: say so
- Incrementality vs cannibalization: the academic literature documents both synergy and substitution depending on channel and product. There is no default answer; that is why the workflow forces an explicit incrementality step.
- Sourced-vs-influenced reporting: practitioners agree they must be tracked separately, but whether influenced revenue deserves board-level weight is an active industry debate, not a settled rule.
- The partner-CAC-advantage magnitude: real but range-disputed; anchor on the $0.53 survey figure and label everything else.
references/metrics-and-formulas.md›
# Metrics and Formulas
The working metric set for one partner relationship, with worked numbers. No canonical formula sheet exists in the field; these are the forms practitioners converge on. Compute every value on gross margin where the formula says so - revenue-based versions flatter the partner.
## Revenue attribution - the foundation
- Partner-sourced: closed-won revenue from opportunities the partner originated, proven by a timestamp such as deal registration. The deal would not exist without the partner.
- Partner-influenced: revenue on deals your team sourced where the partner materially advanced the deal.
- Partner-attached: any partner touch - the loosest bucket.
- Defensible defaults (practitioner consensus):
- Track all three separately in the CRM.
- Apply a fixed attribution window (~14 days from deal creation).
- Credit one partner per deal.
- Uncapped influence is how attributed revenue exceeds total revenue - the result that destroys finance's trust permanently.
## Net contribution
```
Net revenue after partner cost = list or ACV
− partner discount − rebates/SPIFFs − MDF − marketplace fee − cost-to-serve
```
Cost-to-serve is three distinct buckets: recruit, enable, serve. It covers:
- Account-manager time.
- Training/certification delivery.
- Portal/tooling allocation.
- Support.
- Events/travel.
Worked restatement (practitioner example): a partner generating $2,000,000 in annual revenue, net of $500,000 discounts, $200,000 MDF, $150,000 dedicated account management, and $100,000 training/support, contributes **$1,050,000** - not $2M. Run this restatement on every headline number before it enters a memo.
## Partner CAC
```
Partner CAC = (partner program costs + incentives + MDF + channel overhead attributable to this partner)
/ new customers acquired via this partner
```
- Compare against direct CAC over the same window. The channel thesis exists only if partner CAC is materially below direct CAC.
- Diagnostic when partner CAC comes out HIGHER than direct: check these three causes in order. Each is cheaper to rule out than the next, and the first one invalidates the number outright.
1. Attribution double-counting deals that would have closed anyway - a re-run of the query, near-zero effort.
2. Commission rate too high for the deal size - an hour of arithmetic, and the fix is a term you can negotiate.
3. Poorly qualified partner leads - weeks of cohort win-rate and retention data, and the slowest fix of the three.
- Best-documented anchor: the 2014 Pacific Crest/KeyBanc survey put channel CAC at $0.53 per $1 of new ACV (industry survey - the strongest channel-CAC evidence available; the often-quoted "$1.02 field sales" companion figure is not confirmed in the published write-up).
## ROI, LTV, payback
```
Partner ROI = ((partner-attributable revenue × gross margin) − partner-specific costs) / partner-specific costs
LTV (partner-sourced cohort) = ARPU × gross margin % / churn rate : margin, never revenue
LTV:CAC ≥ 3:1 as the standard health line (SaaS practitioner convention)
CAC payback = partner CAC / (ARPU × gross margin %)
```
Practitioner rules of thumb, all vendor/blog evidence class - label them:
- Partner ROI above 3x with CAC payback under ~120 days reads healthy.
- Partner CAC above ~1.5x direct signals a broken structure.
- Partner-sourced retention lagging direct by more than 10-15pp signals ICP misalignment.
## Ramp and payback
- The first 6-12 months are a double cost: funding the partner's enablement and your own team simultaneously, before the partner sells or services anything alone. Analyst description of the career risk: an 18-month channel build where most people get fired at month 12.
- Model: months to first revenue → months to steady-state productivity → the month cumulative net contribution crosses zero. Present that break-even month explicitly; NPV-discount the ramp if the sign-off owner is finance.
- The upside case once ramped: a partner that can sell and service deals alone pushes incremental CAC toward zero and lifts LTV via lower churn - but only after the ramp is paid for.
## Incrementality
Attributed revenue is not incremental revenue. The canonical demonstration: Blake, Nosko & Tadelis (2015, _Econometrica_) - eBay halted brand-keyword paid search and 99.5% of clicks were retained organically. A channel can carry a real, positive, statistically significant revenue number while contributing almost zero incremental business.
Ranked by evidence bought per unit of effort, strongest ratio first - the axes disagree, so read all four:
- efficiency: assumed-rate band > matched holdout > geo split > channel pause
- value, meaning evidence strength: channel pause > matched holdout == geo split > assumed-rate band
- effort: channel pause > geo split > matched holdout > assumed-rate band
- compliance cost: channel pause > geo split > matched holdout == assumed-rate band
Matched holdout and geo split tie on evidence strength because they are the same quasi-experiment read two ways - a treated group against an untreated one - and each is confounded in its own direction, matching quality against territory heterogeneity, with neither dominating the other. They do not tie on effort, so the holdout wins on efficiency. Matched holdout and assumed-rate band tie on compliance cost at genuine zero: both run inside the vendor's own systems, touch no clause of the partner agreement, and can be abandoned mid-way with nobody outside the company aware they ran.
| Method | What it buys | Effort |
| --------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------- |
| Assumed-rate band - default | Bounds the verdict without measuring anything: state the assumed incrementality rate as a named assumption and show the verdict at 100%, 70% and 40% incremental. A verdict that holds across the band settles the question with no experiment at all | An hour, alone, on the model already built. Fully reversible |
| Matched holdout | Withhold partner-sourced leads from a comparable segment and read the conversion difference. Real measurement, confounded only by matching quality | A week to set up in the CRM, then weeks to a quarter of read-out. Needs the partner manager's agreement and sales-ops help |
| Geo split | Same measurement on a territory carve-out; cleaner separation, coarser matching | A quarter. Needs a rules-of-engagement change and territory sign-off, and it is visible to the partner |
| Channel pause | The eBay design and the strongest evidence available outside a lab - but it spends real pipeline to get it | A quarter of foregone revenue plus sales-leadership sign-off, and the lost pipeline never comes back |
Compliance cost by method, heaviest first:
- Channel pause: suspending a partner motion can breach exclusivity, minimums or rules of engagement written into the agreement, so it needs a contract read and legal sign-off before it starts, and it cannot be unwound afterwards.
- Geo split: a smaller version of the same exposure - a territory carve-out amends the rules of engagement and is visible to the partner, so it needs sign-off and is awkward to reverse.
- Matched holdout and assumed-rate band: neither carries any compliance cost.
Default to the assumed-rate band; buy an experiment only when the verdict flips inside the plausible range.
What that default starves is the channel pause: strongest evidence on the page, heaviest effort on the page, last on efficiency every time, so a program ranking on the ratio never once measures incrementality properly. Promote it when a nine-figure renewal, an exit decision, or a finance owner who rejects assumed rates makes the incrementality number the whole verdict rather than one input to it.
Where the agreement carries exclusivity or minimums the pause would breach, delete it from the menu and record it as deleted with the clause that killed it - a method left at the bottom of the list gets proposed again by whoever reads the memo next.
Re-rank against the situation too: an existing holdout, or a channel already paused for unrelated reasons, makes the measured options nearly free. This ordering is a default, not a law.
Academic literature is genuinely unsettled - documented synergy and documented substitution both exist depending on channel and product.
## Incentive ceiling
```
Max affordable partner payout per customer = (lifetime REVENUE per customer × gross margin %)
− target CAC
```
Worked example: lifetime revenue $1,200, gross margin 70%, target CAC $200 → ceiling = $640. Any referral fee, commission, or discount value above the ceiling buys revenue at a loss.
Note the input here is lifetime _revenue_, not the margin-adjusted LTV defined above - applying gross margin twice halves the ceiling and rejects payouts the business can afford.
## Revenue per partner
```
RPP = partner-sourced revenue / active partners
```
For a single-partner model, use RPP of comparable active partners as the volume sanity check - and count only active partners, or the denominator flatters the program.
references/partner-type-economics.md›
# Partner-Type Economic Models and Money Flows
Pick the model that matches how money actually moves; the margin math is not interchangeable across types. Range figures below are mostly vendor or practitioner estimates - carry the source label into any memo that quotes them.
## Money flow by partner type
| Partner type | How money flows | Typical margin / fee range | Evidence class |
| ----------------------------- | ------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------ | ----------------------------------------------------- |
| Referral / agency | Referral fee as % of first-year ACV, one-time or recurring | 10-15% typical for software; 30-50% outliers exist | Vendor/practitioner |
| Reseller / VAR | Discount off list; front-end margin + back-end rebate + MDF | 25-35% VAR gross margin; ~40% traditional on-prem discount; ~10% for a registered referral | Vendor/practitioner |
| Distributor / two-tier | Margin stacks across vendor → distributor → reseller → customer | 8-15% distributor margin on top of reseller margin | Vendor/practitioner |
| MSP | Margin on wrapped managed services, recurring | 45-60% managed-services gross margin best-in-class; 18-20% EBITDA top quartile | Analyst + vendor benchmark (Service Leadership Index) |
| OEM / embed / white-label | Royalty or revenue share, per unit or per customer | Royalty commonly 1-10% of license revenue; splits vary widely | Vendor/practitioner |
| ISV / tech alliance / co-sell | Influenced revenue; marketplace listing fees; private offers; commit burn-down | Listing fee ~3% public SaaS; 1.5-3% private offers | Platform documentation (primary source) |
| Affiliate / creator | CPA flat per action, or revenue share on recurring | SaaS 20-30% recurring; ecommerce 10-15% first order; fintech $50-200 CPA | Vendor benchmark |
## Margin stacking and its fix
- Root cause is double marginalization (Spengler 1950, _J. Political Economy_): when successive firms with market power each add a markup, the end price exceeds what an integrated channel would charge and total output falls.
- Every added tier compounds the stack until the end price is uncompetitive or the economics stop supporting the value each party adds. Model the full stack for any two-tier structure, never one tier at a time.
- The peer-reviewed coordination fix is the two-part tariff (Jeuland & Shugan 1983; McGuire & Staelin 1983): supply near marginal cost plus a fixed fee, so no tier has an incentive to over-mark-up. Use it as a term-design lens when the stack breaks.
## Cloud marketplace mechanics
- Listing fees are the best-documented numbers in partner economics: ~3% for public SaaS offers, 1.5-3% on private offers (AWS primary source, 2024; Azure and GCP sit in the same ~3% band after both cut fees from far higher rates). Verify current schedules before quoting - they change.
- Committed-spend drawdown is the dominant enterprise buying motive: a marketplace purchase retires spend the buyer already committed to the cloud provider (AWS PPA/EDP, Azure MACC, GCP CUD), which is use-it-or-lose-it money.
- Model the drawdown nuance: a $1M qualifying marketplace purchase retires $1M of commit but costs the buyer $1M - versus roughly $800K for $1M of directly consumed services at a 20% discount. The buyer's calculus is not free money.
- Eligibility rules move: effective May 2025, AWS restricted commit retirement to SaaS hosted entirely on AWS, stranding vendors who relied on it. Check eligibility for the specific product, not the category.
- The win discount usually matters more than the fee: the discount conceded to close a marketplace or channel deal frequently dwarfs the 3% listing fee. Model both in the same stack.
## Services multiplier - handle with care
Vendor-commissioned analyst studies report partners earning $6.40-$7.13 in services per $1 of AWS sold, and up to ~$7 per $1 on other platforms (Canalys/Omdia, commissioned by the platform vendors). Useful for arguing the PARTNER's side of a Bech-style P&L - why the partnership is worth the partner's investment - but label it vendor-commissioned, never treat it as a neutral benchmark.
## Mode check before economics
Before modeling any of the above, confirm a partner channel is even the right mode for the synergy sought: ally-vs-acquire (Dyer, Kale & Singh 2004, HBR) and the transaction-cost make-vs-buy channel test (Anderson & Schmittlein 1984; confirmed by the 2006 AMJ meta-analysis). A wrong-mode relationship fails regardless of how the unit economics pencil out.
references/worked-examples.md›
# Worked Examples
Both examples model the same fictional case: a B2B SaaS vendor ($40K average ACV, 78% gross margin, direct CAC $11,000) evaluating a proposed reseller. Numbers are illustrative, built to be internally consistent - not benchmarks.
## Positive example - a memo that earns a verdict
```
PARTNER ECONOMICS: Meridian Systems (reseller/VAR), 2026-03-10
Decision : sign / don't sign. Threshold agreed with CFO: partner acquisition cost
per $1 of net revenue below the direct motion's $0.275 ($11,000 CAC on
$40K ACV), and cumulative contribution positive by month 20.
Partner type : Reseller. 25% discount off list; +5pp deal-registration uplift;
$30K/yr MDF commitment; no marketplace flow.
Revenue model : Sourced only: Meridian originates; no influenced credit claimed.
Y1: 6 deals × $40K ACV = $240K. Y2: 14 deals = $560K. Y3: 20 = $800K.
Volume source: Meridian's own plan, haircut 30% (unverified, partner-supplied).
Cost stack (Y2) : discount $168K (30% effective incl. registration) | MDF $30K |
0.4 FTE partner manager $52K | enablement/certification $18K |
support uplift $12K | portal allocation $3K → total $283K
Net contribution: Y1 −$74K (ramp) | Y2 +$154K margin-adjusted | Y3 +$248K
Partner CAC : Y2 acquisition cost excluding discount $115K → $8,214/customer; fully
loaded including the $168K discount $283K → $20,214/customer. Neither
figure compares to direct CAC ($11,000) per customer: the partner nets
us $28K per deal against direct's $40K, so per-customer numbers price
different goods. Like-for-like, per $1 of net revenue: partner $0.293
vs direct $0.275. Fails the threshold by 6.7%. (Against the Pacific
Crest $0.53-per-$1-of-new-ACV channel benchmark the partner is normal
at $0.505: normal for the channel, still worse than our own direct.)
Ramp & payback : first revenue month 5; double-cost months 1-8;
cumulative contribution crosses zero month 17. Inside threshold.
Incrementality : Meridian's install base overlaps our outbound territory ~20%.
Assumed 75% incremental; verdict still holds at 50%, flips at 35%.
Partner's P&L : Meridian earns $168K margin + est. 2x in implementation services (their
model, unverified). Passes the Third-Law check: the split funds their climb.
Sensitivity : two levers, both tight. Hold effective discount at or under 25.3% at the
planned 14 Y2 deals, or land 15 Y2 deals at the proposed 30%. Either one
clears $0.275; neither has slack. Incrementality < 35% flips it whatever
the discount. Ramp length ±3 months does not flip it.
Verdict : RENEGOTIATE: do not sign at 30% effective. Ask: 25% all-in, funding the
deal-registration uplift from MDF instead of stacking it on the discount.
Re-run at signature; sign if the term lands at or under 25%, with a
month-12 gate of ≥5 closed deals.
Open questions : Meridian's services attach rate unverified; churn of their sourced
customers unknown: a 10pp retention gap vs direct would flip Y3.
```
Why it works:
- Sourced-only revenue with a haircut and a named source.
- The cost stack is fully loaded, including people time.
- CAC is stated on both conventions and then compared on the only basis that is like-for-like.
- Incrementality carries a number and a flip point.
- The partner's own P&L is checked.
- The verdict names the exact term that would reverse it.
The CAC line is the part worth copying. Divide each side's acquisition cost by the net revenue it actually bought: neither convention produces a per-customer number comparable to direct when the partner nets $28K per deal against direct's $40K.
Two conventions are internally consistent; only one can be used at a time:
- Book revenue gross and charge the discount to acquisition cost.
- Book revenue net of discount and leave it out of cost.
Reporting the $8,214 non-discount figure against the $11,000 direct CAC as a 25% win is the trap: it charges the $168K discount to nobody while still counting the revenue it bought at list, and it makes every reseller look cheap.
A memo that lands on "renegotiate" rather than "sign" is the skill working, not failing.
## Negative example - a memo that flatters the partner
Each line annotated with what is wrong.
```
PARTNER ECONOMICS: Meridian Systems, 2026-03-10
Revenue : $1.2M partner-attributed pipeline over 3 years.
```
- Blends sourced and influenced into one "attributed" number, and quotes pipeline, not closed-won. The single most common inflation move.
```
Costs : 25% reseller discount.
```
- Cost stack missing everything but the discount: no MDF, no partner-manager time, no enablement, no support uplift. This is the $2M-partner-that-is-really-$1.05M failure in miniature.
```
Partner LTV : $40K ACV × 5-year lifetime = $200K per customer.
```
- LTV computed on revenue, not gross margin, and lifetime asserted with no churn basis. At 78% margin the honest figure is ~$156K before churn adjustment.
```
Partner CAC : $0: Meridian brings the customers.
```
- Ignores every acquisition-side cost: MDF, enablement, partner-manager time, the discount itself. Partner CAC is never zero; a zero here means the model never looked.
```
Incrementality : all partner revenue is net-new by definition (they source the deals).
```
- Assumes 100% incrementality. Sourcing proves origination, not incrementality - the deal may have arrived through direct or another partner anyway.
```
Benchmark : industry data shows partners deliver 80% of channel revenue
through top performers like Meridian.
```
- Launders the untraceable 80/20 statistic as fact, and misapplies a concentration claim as if it predicted this partner's performance.
```
Verdict : strong strategic fit: recommend signing.
```
- No threshold, no sensitivity, no ramp, no payback month, no open questions. "Strategic fit" is doing the work the numbers were supposed to do.
The tell across all seven lines: every omission moves the answer in the partner's favor. A model whose every simplification points the same direction was built to justify, not to decide.
SKILL.md›
---
name: partner-economics
description: Model the unit economics of one partner relationship - partner P&L, margin modeling, cost-to-serve, referral-fee and lifetime-value split, channel partner ROI, partner CAC vs direct CAC, ramp and payback - to decide whether to sign, scale, renegotiate, or exit that partner. Covers B2B channel and alliance partners plus B2C affiliate and creator economics. Use whenever the user mentions partnership unit economics, a single-partner business case, reseller, VAR, MSP, OEM or marketplace deal economics, revenue-share and margin-stack analysis, or asks whether a partner is worth pursuing, even if they never say economics. Do NOT use for ranking a portfolio of candidates - use mbfinotti/partnerships-skills@alliance-prioritization instead.
license: MIT
metadata:
author: Maya-Beth Finotti
version: "1.4.1"
---
# Partner Economics
Model one partner relationship as a small P&L - attributable revenue net of every partner cost, against partner CAC, ramp, and incrementality - to decide sign, scale, renegotiate, or prune. Two structural facts frame the whole exercise; state them plainly rather than paper over them:
- **No canonical "partner P&L" template exists.** Practitioners converge on the same revenue and cost components, but vendors publish incompatible definitions, and any claim to _the_ standard (PARTNERNOMICS-style certification included) is a proprietary product, not industry consensus.
- **Most widely quoted partner benchmarks are vendor marketing, not research.** The famous "80% of channel revenue from 20% of partners / only 11% hit incentive goals" figures are untraceable; so are the "most partners never transact" claims. Label every number's evidence class; quote nothing as settled fact without a named source. Read [references/benchmarks-and-figure-grading.md](references/benchmarks-and-figure-grading.md) before citing any figure: the graded benchmark set, plus the untraceable figures to refuse.
What the field genuinely offers:
- **Hans Peter Bech's channel-partner P&L** - model the PARTNER's own 3-5 year economics, because "it is crucial that the channel partner takes ownership of the P&L." The single most articulated practitioner method for one relationship.
- **Spengler's double marginalization (1950)** and the two-part tariff fix (Jeuland & Shugan 1983; McGuire & Staelin 1983) - the peer-reviewed root of margin stacking.
- **Blake, Nosko & Tadelis (2015, Econometrica)** - when eBay halted brand-keyword ads, 99.5% of clicks were retained, the canonical proof a channel can show strong attributed revenue while being nearly non-incremental.
- **Dyer & Singh's relational view (1998, AMR)** - "relational rents", profit neither firm could generate alone - and **Gomes-Casseres' Third Law** (_Remix Strategy_, 2015): the value split must motivate both parties to keep contributing.
- **Dyer, Kale & Singh (2004, HBR)** ally-vs-acquire, plus **Anderson & Schmittlein (1984, RAND J. Econ.)** and the 2006 AMJ TCE meta-analysis - peer-reviewed grounding for whether a partner channel is even the right mode before any economics get modeled.
## Interview
Ask before modeling anything. One question per message; offer multiple-choice options; skip whatever the user already answered.
- Partner type: referral/agency, reseller/VAR, distributor/two-tier, MSP, OEM/embed, ISV tech alliance/co-sell/marketplace, or affiliate/creator?
- Whose economics: your side, the partner's own P&L (Bech's point - the partner must own theirs), or both? The ranked answer is in Brainstorming the model; the three questions below decide which rung applies.
- Which decision must this model serve - sign / don't sign, scale, renegotiate, prune/exit - and what go/no-go threshold applies? Who signs off?
- By what date must the verdict land - a signature deadline, a board or QBR slot? Anything inside a few weeks rules out both the dual-sided P&L and any incrementality experiment; each needs partner disclosure or a read-out window.
- One-off decision, or a model re-run at every QBR? A re-run mandate promotes the dual-sided P&L and a written assumption register; a one-off promotes the vendor-side contribution model alone.
- Effort ceiling: analyst hours available, whether a finance reviewer is assigned, and whether the partner will open its own numbers? No finance reviewer, or a partner that will not disclose, caps the work at the vendor-side model whatever the decision needs.
- What is your current direct CAC, and the average deal size it buys? Step 6 compares against both; without them there is no threshold to test.
- What is product gross margin, and the churn rate or average customer lifetime for this segment? Lifetime value is computed on margin, so guessing either one silently sets the verdict.
- Revenue mechanics: fee as % of first-year ACV or recurring, discount off list, royalty, revenue share, CPA?
- What stacks on top: rebates, SPIFFs, MDF commitments, deal-registration uplift, marketplace listing fee?
- Costs to recruit, enable, and serve: account-manager time, training/certification delivery, portal and tooling, support load?
- Contract shape: exclusivity, term, deal registration, clawback, MDF commitments, minimums?
- Ramp expectation: time to first revenue, and who funds the double-cost period?
- What data actually exists: CRM attribution with sourced/influenced tagged, marketplace disbursement reports, finance systems - or estimates only?
- B2B channel/alliance or B2C affiliate/creator?
## Workflow
1. Run the Interview. Fix the decision the model serves and its go/no-go threshold before computing anything - a model built without a threshold produces a flattering number, not a verdict.
2. Pick the partner-type economic model; money flows and margin math differ sharply by type. See [references/partner-type-economics.md](references/partner-type-economics.md).
3. Separate partner-sourced, partner-influenced, and partner-attached revenue; never blend them. Cap influence credit so attributed revenue can never exceed total revenue - uncapped influence ends the conversation with finance immediately and permanently.
4. Build the fully-loaded cost stack: discount off list, rebates/SPIFFs, MDF, marketplace fee, account-manager time, enablement and certification delivery, support, portal/tooling allocation. A $2M partner net of $500K discounts, $200K MDF, $150K account management, and $100K training/support contributes $1.05M, not $2M.
5. Compute partner-adjusted gross margin and annual net contribution. Compute lifetime value on gross margin, never on revenue. See [references/metrics-and-formulas.md](references/metrics-and-formulas.md).
6. Compute partner CAC fully loaded, then compare it to direct CAC on a like-for-like basis: acquisition cost per $1 of net revenue, never per customer. Per-customer CAC prices different goods whenever a discount lowers revenue per partner deal.
Two conventions are internally consistent; state which one you used and never mix them:
- Book revenue gross and charge the discount to acquisition cost.
- Book revenue net of discount and leave it out.
Do not rank the two conventions: on the per-$1-of-net-revenue basis they produce the same number, so an ordering would be false precision. If partner cost per $1 of net revenue is not below direct's, the channel thesis is broken; say so in the memo.
7. Model ramp and payback. The first 6-12 months are a double cost - funding the partner's enablement and your own team simultaneously. State time to first revenue and the month cumulative contribution crosses zero.
8. Test incrementality explicitly: how much of this revenue would have arrived through direct anyway? Attributed is not incremental (Blake, Nosko & Tadelis).
Start with the assumed-rate band, never with an experiment - efficiency: assumed-rate band > matched holdout > geo split > channel pause. Buy an experiment only when the verdict flips inside the plausible range.
That order starves the channel pause, which buys the strongest evidence available and costs the most. Promote it only when incrementality is the entire verdict; delete it outright where exclusivity or minimums in the agreement forbid suspending the motion.
Evidence strength, effort and compliance exposure are ranked separately in [references/metrics-and-formulas.md](references/metrics-and-formulas.md).
9. Run sensitivity on the 2-3 assumptions that actually move the answer - usually deal volume, effective discount, and ramp length. Report which shifts flip the verdict.
10. When modeling both sides, build the partner's own 3-5 year P&L (Bech) and test the Gomes-Casseres Third Law: does the value split motivate the partner to keep contributing?
11. Present the memo section by section - revenue, costs, contribution, CAC, ramp, incrementality, sensitivity, verdict - and validate each with the user before drafting the next.
12. Stop at the approval gate: finalize nothing without the user's explicit approval of the assembled memo.
13. Check the Pass Threshold below; iterate until every check holds.
14. Set the revisit cadence and triggers: joint-business-plan refresh at period start, a QBR economics review each quarter, a renewal/exit gate - plus event triggers such as a marketplace fee change, a commission renegotiation, or a ramp-milestone slip.
15. If your harness has persistent memory, memorize the approved model, its named assumptions, and the verdict, so QBR-time re-runs start from them instead of from scratch.
16. If you can browse the web, verify current marketplace fee rates and partner-type benchmark ranges before finalizing - fee schedules change (AWS restricted commit-retirement eligibility in May 2025). Otherwise flag every figure as user-supplied and unverified.
## Brainstorming the model
Run the exercise the way channel and alliance teams actually run it - collaboratively, through joint business planning, deal-desk review, and QBR economics reviews - never as a finance-silo calculation:
- Ask one question per message throughout; never batch questions.
- Treat the model as co-created where both sides can be convened: in joint business planning the vendor and partner build the plan together, and the partner owns its own P&L. A model the partner never saw predicts a partnership the partner never agreed to.
- Put the modeling approaches on the table ranked, state your recommendation and why, then wait. Rank by decision quality bought per analyst hour - the axes disagree, so read all four:
- efficiency: vendor-side contribution > dual-sided P&L > marketplace-mechanics-first
- value, meaning decisions it can settle: dual-sided P&L > vendor-side contribution > marketplace-mechanics-first
- effort: dual-sided P&L > vendor-side contribution > marketplace-mechanics-first
- compliance cost: marketplace-mechanics-first > dual-sided P&L > vendor-side contribution
| Approach | Settles | Effort |
| ---------------------------------- | ----------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Vendor-side contribution - default | sign / don't sign, scale, prune, partner CAC against direct | About a week of one analyst on data you already hold - CRM attribution and finance. One finance reviewer, no partner disclosure. Fully reversible. |
| Dual-sided Bech-style P&L | everything above, plus the split itself: renegotiation terms and the Third-Law check | A quarter. Needs the partner to disclose cost and services economics you cannot see, and coordination across both finance teams. Asking is itself a commitment signal - hard to walk back. |
| Marketplace-mechanics-first | only the routed-deal cost stack: listing fee, private-offer rate, commit-drawdown eligibility, win discount | An hour; the rates are published primary sources. Settles nothing alone - a required layer on top of one of the two above whenever deals route through a marketplace, never a substitute for either. |
- Compliance cost by approach, heaviest first:
- Marketplace mechanics: eligibility is the platform's to change (AWS restricted commit retirement in May 2025), so re-verify a routed model's eligibility against current platform documentation rather than against any figure written down here.
- Dual-sided P&L: a smaller version of the same exposure - asking a partner to open its cost base usually means a mutual NDA, puts another company's confidential numbers in your files, and signals a commitment you cannot un-signal.
- Vendor-side model: touches nobody outside the company.
- Default to the vendor-side contribution model. Move up to dual-sided when any of these hold: the decision is the split rather than the signature, the partner is reseller-class and must own its own P&L (Bech), or partner disclosure is already on the table.
- The dual-sided P&L is what this efficiency order starves: it settles strictly more than the default and costs a quarter against a week, so the ratio demotes it in every engagement and renegotiations keep getting run on one side's numbers. Promote it whenever the question is how the value splits rather than whether to sign - the split cannot be argued honestly from the vendor's ledger alone.
- Delete rather than demote what this user's answers rule out, and say in the memo that it was deleted and why:
- A partner that will not open its books removes the dual-sided rung entirely.
- No assigned finance reviewer removes it too.
- A deal that routes through no marketplace removes the marketplace-mechanics layer.
- A rung carried to the bottom of the list comes back as scope at the next review, from someone who never heard the constraint.
- Re-rank against what you already know about this user before recommending: a finance reviewer already assigned collapses the dual-sided coordination cost, and an approved model from a past engagement makes extending it cheaper than any of the three. The ordering is a default, not a law - it shifts with the decision at stake and with who does the work.
- Name the assumptions out loud before computing: which volume number is partner enthusiasm, which benchmark is vendor-published, which cost is a guess.
- Argue the strongest case AGAINST the partnership before presenting the model; relationship investment and logo halo are the documented ways this exercise fools itself.
- Route non-standard terms through a deal-desk lens - written discount bands, exceptions justified in writing. A desk answering to the leader carrying the quarterly number approves what closes the quarter.
- For recurring reviews, insist both sides work off one shared scorecard read BEFORE the meeting. The two documented QBR failure modes: "status theater" - a meeting that ends with nobody owning a dated action - and "no shared data going in" - both sides arrive with different pipeline numbers and spend half the session arguing about whose spreadsheet is right.
- Secure CFO and CRO buy-in one-on-one before presenting the case to wider leadership.
- Present the deliverable one section at a time; validate each before drafting the next; gate finalization on the user's explicit approval.
## The Partner Economics Memo
Deliver the engagement as this artifact - the business case the user takes to their sign-off owner:
```
PARTNER ECONOMICS: <partner>, <date>
Decision : sign / scale / renegotiate / prune, with the agreed go/no-go threshold
Partner type : type + money-flow mechanics (fee %, discount, royalty, rev-share, CPA, marketplace)
Revenue model : sourced and influenced projected separately: never one blended number
Cost stack : discount | rebates/SPIFFs | MDF | marketplace fee | account mgmt | enablement | support | tooling
Net contribution: partner-adjusted gross margin and annual net contribution, by year across the ramp
Partner CAC : fully loaded, vs direct CAC: attribution method stated
Ramp & payback : time to first revenue, double-cost period, month cumulative contribution crosses zero
Incrementality : share of attributed revenue that would have arrived anyway, and its evidence class
Partner's P&L : the partner's own 3-5yr economics and the Third-Law check (when modeled dual-sided)
Sensitivity : the 2-3 assumptions that move the verdict, and where each flips it
Verdict : against the stated threshold: and what result would change it
Open questions : missing data, unverified figures, and which finding would reverse the verdict
```
Three rules govern the memo:
- Every benchmark carries a source-quality label - peer-reviewed, analyst, industry survey, vendor benchmark, or untraceable.
- Every input traces to a named source or is marked "unverified, user-supplied" - never an invented number, and never a figure carried over from a worked example. If direct CAC, gross margin, or churn is still missing at step 5, stop and ask the user rather than reaching for the nearest number on the page.
- The open-questions section is mandatory: a model with no stated uncertainty has stopped asking.
## Pass Threshold
Ship nothing until all of these hold; iterate until they do:
1. **Partner acquisition cost per $1 of net revenue projected below the direct motion's**, fully loaded, on one stated convention. Per-customer CAC is not a valid comparison when a discount lowers revenue per partner deal. Otherwise the channel thesis is broken and the memo says so. Evidence class: practitioner consensus, anchored by the 2014 Pacific Crest/KeyBanc survey ($0.53 channel CAC per $1 of new ACV - the best-documented channel-CAC figure available, and already a ratio for this reason).
2. **Sourced and influenced never blended**, one partner credited per deal, influence capped so attributed revenue can never exceed total revenue.
3. **Cost-to-serve fully loaded.** The model survives the $2M-partner restatement: headline revenue re-expressed net of discount, MDF, account management, and training/support ($2M → $1.05M in the worked example).
4. **Lifetime value computed on gross margin**, never on revenue.
5. **Incrementality addressed explicitly**, never assumed at 100% - the cannibalization estimate and its evidence class appear in the memo.
6. **Sensitivity run and reported**; a verdict that survives only one assumption set is a bet, not a model.
7. **Every benchmark labeled** by evidence class; the untraceable famous figures (80/20 concentration, 11% incentive attainment, "most partners never transact") are flagged as untraceable or omitted.
## KPIs
Judge the economic decision itself, six to twelve months out - not the partner's ongoing performance, which belongs to `mbfinotti/partnerships-skills@partner-performance`:
- Realized partner CAC vs the modeled projection, and vs direct CAC over the same window.
- Actual cumulative net contribution vs the modeled ramp curve.
- Retention/NRR of partner-sourced customers vs direct-sourced - a gap beyond 10-15pp signals ICP misalignment (practitioner threshold; the SaaS Capital Aug 2019 survey found channel-majority companies run ~3pp lower retention).
- Time to first partner revenue vs the modeled ramp.
- Assumption audit: which modeled inputs proved wrong, by how much - the input to the next model.
- Verdict durability: did the sign/scale/prune call survive the first two quarterly reviews without reversing on facts the model should have caught?
## B2B and B2C
The method is identical on both sides - attributable revenue minus fully-loaded partner cost, against partner CAC, over a lifetime, adjusted for incrementality. Say so when asked; never reinvent it per side.
What genuinely differs - design for it:
| Dimension | B2B channel/alliance | B2C affiliate/creator |
| ------------------- | ------------------------------------------------- | ---------------------------------------------------------------------- |
| Unit of analysis | Partner firm - few, large | Creator/publisher - many, long tail |
| Money flow | Discount off list, rebates, MDF, marketplace fees | CPA or revenue share paid per action |
| Ramp & enablement | 6-12+ months double-cost, certification | Near zero; instant activation |
| Attribution | Deal registration, CRM opportunity tagging | Cookie/attribution window per action |
| Concentration | A handful of partners dominate | Power-law long tail of small earners |
| Effective-rate trap | Margin stacking across tiers | Effective rev-share runs 20-40% below the stated rate after deductions |
Full sourced contrast: [references/b2b-vs-b2c-contrast.md](references/b2b-vs-b2c-contrast.md).
## Failure Modes
Work the table in row order - it is ranked by verdict distortion removed per hour spent, not by how often the failure occurs. The first five rows are re-runs of arithmetic and labeling you already hold, each able to reverse a verdict on its own within a working day. The rest are term-design and program-hygiene fixes measured in quarters: they change the next contract, not the number in this week's memo.
| Failure | Fix |
| ------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------- |
| Double-counting influenced revenue | Separate sourced/influenced; one partner per deal; cap influence at total revenue |
| Ignoring cost to serve | Fully-loaded stack; run the $2M → $1.05M restatement on every headline number |
| Counting cannibalized revenue as incremental | Model incrementality explicitly; segmentation and rules of engagement, not hope |
| Misattributed partner CAC | If partner CAC exceeds direct, check in this order - double-counted influence, then commission against deal size, then lead quality |
| Quoting an untraceable benchmark as fact | Label the evidence class; prefer named-source figures; write "unverified" over inventing |
| Margin stacking collapse across tiers | Model the full stack (Spengler); consider a two-part tariff shape for the terms |
| Over-discounting to win the partner | Model the win discount together with the marketplace fee - the discount frequently dwarfs a 3% listing fee |
| Marketplace commit-drawdown surprises | Model commit-retirement mechanics and eligibility rules, not just the listing fee |
| MDF that never converts | Tie MDF to documented plan commitments; model utilization, not allocation |
| Sunk-cost persistence and the long-tail illusion | 100 signed partners with 12 active is a 12-partner channel; prune on net contribution, not tenure |
## Invocation Examples
- "Model the economics of a proposed reseller before we sign - 30% effective discount, $40K ACV, they project 14 deals in year two."
- "Is this marketplace partner worth keeping? Two years in, decent attributed revenue, and I suspect most of it is not incremental."
- "Build both sides of the P&L for an MSP alliance - ours and theirs - so we can renegotiate the split with real numbers."
## Reference
- `mbfinotti/partnerships-skills@alliance-prioritization` - ranks a portfolio of candidate alliances; hands single-partner P&L modeling to this skill.
- `mbfinotti/partnerships-skills@partner-channel-program` - sets the program-wide economics envelope this single-partner model must live within.
- `mbfinotti/partnerships-skills@partner-tiering` - tier criteria and benefits inside an existing program.
- `mbfinotti/partnerships-skills@partner-performance` - ongoing scorecards and QBR performance review once the model is approved.
- `mbfinotti/partnerships-skills@affiliate-commission-structure` - designs affiliate payout tiers for professional affiliates; a different discipline from single-partner modeling.