mbfinotti/revops-skillsContrôle réussi
SKILL DETAIL
revenue-funnel
mbfinotti/revops-skills/revenue-funnel
Design a company's revenue funnel model from scratch, at the macro level - the stage set, the unit of analysis (lead, buying group, account), the model purpose (process vs planning vs forecast), the cohort-based conversion assumptions behind the plan, and the ownership handoffs between marketing, sales, and CS. Use whenever the user mentions funnel design, funnel stages, a funnel model, conversion rate assumptions, a bowtie model, a demand waterfall, MQL to SQL to opportunity definitions, or marketing-sales handoff design - even if they never say "funnel". Covers B2B and high-consideration B2C. Do NOT use for auditing an existing pipeline's stage definitions - use mbfinotti/revops-skills@pipeline-stage-definition-audit instead.
Installations · 167Voir la source
Installation
npx skills add https://github.com/mbfinotti/revops-skills --skill revenue-funnel
Fichiers du skill
SKILL.md
Dernière synchronisation · 15 sept. 2026
evals/evals.json›
{
"skill_name": "revenue-funnel",
"evals": [
{
"id": 1,
"prompt": "I'm head of RevOps at Cadence Loop, a Series B B2B SaaS at about $14M ARR. We run SDRs into AEs. Marketing counts leads, sales counts opportunities, CS counts accounts, and none of the numbers tie. The board wants a revenue plan by November 14 that's built on funnel math, and we have 7 quarters of stage history sitting in the CRM. Give me our funnel stages.",
"expected_output": "An interview-first response that refuses to hand over a stage list until the model's purpose, unit of analysis, and revenue boundary are settled, then presents 2-3 candidate stage sets inside the 5-7 band for a mid-touch motion, with the deadline explicitly shaping the conversion-rate rung.",
"files": [],
"expectations": [
"Asks clarifying questions before proposing any stage list, one question per message rather than a single block of questions",
"Names the three possible model purposes - process, planning, forecast - and requires one to be chosen as primary before any stage counting begins",
"Treats the unit of analysis (lead, contact, buying group, opportunity, account) as a separate prior decision, driven by the number of stakeholders in a typical deal",
"Raises where pre-sale ends as a third prior decision and lists contract signature, first value delivered, and onboarding completion as the candidate boundaries",
"Recommends modeling two boundaries, booking the accounting boundary at commit for finance while carrying time-to-first-impact as a separate CAC-recovery measure",
"Presents 2-3 candidate stage-set designs with trade-offs and one recommendation, and asks the user to pick before anything is built on top of the choice",
"Specifies each candidate with a per-stage owner, the unit of analysis and where it switches, post-sale coverage, what the design optimizes for, and its standing admin cost",
"Keeps every candidate stage set inside a 5-7 stage band for this mid-touch SDR-to-AE motion",
"States that the November 14 deadline argues for shipping on blended cohort rates with fuller segmentation scheduled as a follow-up refresh rather than delaying the model",
"Describes the deliverable as an artifact set (stage definitions, conversion register, funnel model, SLA, data dictionary, governance charter) presented one artifact at a time for validation",
"Declines to draft per-stage exit-criteria wording, routing that work to the sibling stage-definition audit skill",
"Requires the bottom-up funnel plan to be reconciled against the board's top-down revenue target with the gap stated as a number"
]
},
{
"id": 2,
"prompt": "Marketing ops lead at Perrin Data. Our CRO read an article saying the MQL is dead and everybody is moving to buying groups, so now he wants us to rip out MQLs and qualify demand units instead, before Q1. We're mid-market. A typical deal has 2 people involved, a line manager plus whoever signs, and cycles run about 60 days. Separately, we have three different MQL definitions floating around by region. What's the migration plan?",
"expected_output": "A refusal to migrate to buying-group qualification at this stakeholder count, reframing the real defect as the three competing MQL definitions, with the buying-group evidence presented alongside its counter-case.",
"files": [],
"expectations": [
"Recommends against migrating to buying-group or demand-unit qualification at roughly 2 stakeholders per deal, naming a threshold of roughly 3-4 stakeholders as the bar",
"States that the migration is a costly data-model change rather than a metric rename or relabel",
"States that the MQL is a poor goal but an acceptable operating state",
"Refuses to let published discourse about the MQL being dead override this company's own stakeholder-count evidence",
"Identifies the three competing regional MQL definitions as the actual defect to fix, separate from the buying-group question",
"Prescribes one council-ratified, versioned definition set, with any regional variance requiring council approval rather than standing on its own",
"Names the consequence of definition drift: leads stalling between functions and win-rate reporting that varies by who pulls it",
"Presents any buying-group prevalence evidence (such as 93% of B2B buyers in groups of 2+ and 71% in groups of 4+) alongside the counter-case that short-cycle, small-group decisions keep the individual-lead model viable",
"Flags vendor-sourced case-study outcomes for multi-contact opportunities as directional rather than as a promised result",
"Produces no migration timeline, project plan, or phased rollout for buying-group qualification",
"Records a rising stakeholder count as a trigger-based re-baseline that would reopen the question later, rather than scheduling the migration as a project now"
]
},
{
"id": 3,
"prompt": "Our VP Sales says our win rate is 50%. Our finance analyst pulled the same quarter and says we close 30%. A third dashboard shows 20%. This is Halvorsen Systems, about $40M ARR, one sales team, one CRM, no acquisitions. Leadership burned 40 minutes arguing about it on Tuesday. Which number is right, and how do I stop this from happening again?",
"expected_output": "An explanation that all three can be simultaneously correct because they are different instruments (period, cohort, snapshot), plus a conversion register whose basis column makes the distinction permanent.",
"files": [],
"expectations": [
"States that all three figures can be simultaneously correct for the same underlying business rather than treating the spread as a data-quality error",
"Defines win rate as a period metric: of opportunities that closed in a period, the share that were won",
"Defines close rate as a cohort metric: of opportunities created in a period, the share ultimately won",
"Defines pipeline conversion rate as a snapshot metric computed on a point-in-time pipeline",
"Attributes the spread to the choice of measure and analysis type rather than to one team's data being wrong",
"Prescribes a conversion register in which every rate carries its basis (cohort or milestone), numerator and denominator definitions, sample size, source or provenance, last-refreshed date, and refresh trigger",
"States the rule that a rate without a named basis does not ship",
"Notes that a cohort close rate only reaches its true value once the open opportunities in that cohort reach a terminal state",
"Does not declare one of the three figures the single correct number without naming which question it answers",
"Does not average, blend, or split the difference between the three figures",
"Recommends tracking the cohort-based close rate and the current-period conversion rate as separate instruments rather than forcing one stage set to produce both"
]
},
{
"id": 4,
"prompt": "Building next year's plan at Tessom Robotics. The plan is to take the probability percentages already sitting on each stage in our CRM (10/25/50/75/90), multiply them by open pipeline value to get forecast contribution per stage, then back into how many leads marketing owes us. We have 9 quarters of history, though honestly nobody has cleaned the pipeline since our CRM migration 14 months ago. Sanity check my math?",
"expected_output": "Outright rejection of CRM default probabilities, a stale-pipeline sweep before any rate is computed, and entry-based cohort conversion derived over a 4-8 quarter window and reconciled against the top-down target.",
"files": [],
"expectations": [
"Rejects the CRM's default stage probabilities outright rather than adjusting, recalibrating, or tuning them",
"Gives the reason: they are round numbers nobody derived from this business and they are optimistic by design",
"Notes the error compounds when optimistic probabilities are multiplied against optimistic deal values",
"Requires a stale-pipeline hygiene sweep before any conversion rate is computed, because rates computed over stale records flatter the funnel",
"Notes that a large share of pipeline sitting 12+ months untouched is an observed pattern in practice, not a rare edge case",
"Computes conversion per stage from units that entered the stage over a 4-8 quarter window, divided into those that eventually reached the next stage",
"Counts stage entries rather than current stage occupancy, and explains that occupancy undercounts units that already moved past the stage",
"Sets blended cohort rates from the company's own history as the default rung, segmented on the two dimensions with the most variance the volume can support",
"Names sample size per cell, not ambition, as the constraint on how far segmentation can go",
"Requires the bottom-up plan to be reconciled against the top-down revenue target with the gap stated explicitly as a number rather than averaged away",
"Does not accept a stage-probability-weighted snapshot of open pipeline as a substitute for cohort-based conversion assumptions"
]
},
{
"id": 5,
"prompt": "Quick one. Our MQL-to-SQL conversion is 12%. I pulled three benchmark reports and they all say B2B SaaS should be in the 30-50% range. Our CEO saw the same reports and wants 35% written into next year's marketing goals. We're Ardent Ledger, $8M ARR, heavily outbound-sourced. How should I structure the improvement plan to get from 12 to 35?",
"expected_output": "A refusal to derive a target from published benchmarks, an explanation of definitional incompatibility and tiny sample cells, and a redirect to comparing definitions and deriving the target from own history plus the revenue requirement.",
"files": [],
"expectations": [
"Refuses to write 35% into next year's goals as a target derived from published benchmarks",
"States the rule that published benchmarks trigger investigation and never set targets",
"Attributes cross-publisher variance to definitional incompatibility, since no two sources define MQL or SQL identically, rather than to sampling noise",
"Notes that most published sources do not disclose whether a quoted rate is cohort-based or milestone-based",
"Cites the small-sample problem concretely, such as a widely cited report splitting roughly 106 respondents five ways for roughly 20 companies per cell",
"Notes a wider practitioner range for MQL-to-SQL centering nearer 13-15%, which places this company's 12% far closer to normal than the 30-50% band implies",
"Directs the first action to comparing this company's own MQL definition against the definition behind the published band",
"Treats the 12% figure as a hypothesis to investigate rather than a gap to close",
"Recommends deriving any target from the company's own cohort history reconciled against the top-down revenue requirement, not from an external band",
"Produces no phased improvement roadmap for moving 12% to 35%",
"Flags the outbound-heavy source mix as a segmentation dimension that could explain the rate before any remediation is designed"
]
},
{
"id": 6,
"prompt": "We're 11 people at Bellhaven Grid, maybe 25 closed deals ever, still figuring out repeatability. I copied our pipeline stages from my last company (Fortune 500 software): Prospect, Discover, Qualify, Validate, Champion Confirmed, Economic Buyer Engaged, Proposal, Legal, Security Review, Negotiate, then Closed. We're also starting a reseller motion next quarter, so I was going to add Partner Sourced and Partner Deal Reg as two more stages. What conversion rates should I plug into each one?",
"expected_output": "A cut to roughly 4-5 stages on premature-complexity grounds, a separate mirrored partner pipeline instead of bolted-on stages, and explicitly labeled guesses instead of computed rates at 25 lifetime deals.",
"files": [],
"expectations": [
"Cuts the stage set to roughly 4-5 stages for a pre-PMF or seed-stage company",
"Names premature complexity, a multi-stage pipeline modeling an enterprise process the company does not have yet, as the dominant early-stage failure",
"States that every extra stage is permanent admin overhead and that a stage is added only when a real operational need forces it",
"Refuses to add the two partner stages to the main pipeline and prescribes a separate mirrored partner pipeline instead",
"States that a dedicated legal or security-review stage is an exception for long enterprise review cycles rather than a default",
"States that 25 lifetime deals cannot support trustworthy cohort conversion rates",
"Sources the conversion assumptions from adjusted external benchmarks and judgment at this stage, and labels every such figure explicitly as a guess",
"Recommends contact as the unit of analysis at this stage rather than buying group or demand unit",
"Keeps governance proportional to company stage, a named owner and a review date rather than a funnel council, versioned definitions, and a data dictionary",
"Sets a refresh trigger for revisiting the guessed assumptions once enough matured cohorts exist, rather than treating them as permanent",
"Does not simply supply conversion percentages for the eleven-stage set exactly as the user described it"
]
},
{
"id": 7,
"prompt": "I run growth at Milvane, a direct-to-consumer skincare brand. About 90k orders a year, average order value 38 euros, no sales team at all, everything is web checkout. Our new investor keeps asking for our funnel with MQL-to-SQL-to-opportunity conversion rates, the way his B2B portfolio companies report it. Can you design our funnel stages and write our MQL definition?",
"expected_output": "A clear statement that a pipeline stage model is the wrong artifact for transactional self-serve B2C, a lifecycle plus cohort/event-funnel measurement plan delivered instead, and language for the investor conversation.",
"files": [],
"expectations": [
"States plainly that a pipeline stage model is the wrong artifact for a transactional, self-serve B2C business",
"Gives the reason: transactional self-serve business has no per-deal stage state for a pipeline to track",
"Declines to author an MQL or SQL definition for this business",
"Proposes lifecycle stages plus cohort and event-funnel analytics as the correct instruments instead",
"Distinguishes this case from sales-assisted, high-consideration B2C such as insurance, property, solar, enrollment, automotive, or mortgage, where the full stage model does transfer",
"Notes that in the high-consideration B2C case the unit would typically be a household or applicant rather than a buying group",
"Delivers a lifecycle and cohort measurement plan as the substitute deliverable rather than refusing outright and stopping",
"Supplies language the user can take back to the investor explaining why the B2B waterfall vocabulary does not map to this business",
"Does not translate MQL, SQL, and opportunity into e-commerce equivalents in order to satisfy the request as phrased",
"Does not present B2B benchmark ranges such as visitor-to-lead or opportunity-to-won as applicable to this business",
"Bases the proposed conversion reporting on cohorts (signup or first-order) rather than on period snapshots"
]
},
{
"id": 8,
"prompt": "RevOps lead at Quillmark. Our CMO wants a lead SLA: marketing promises 400 MQLs a month, sales promises to work them, and if down-funnel conversion tanks we can show it was sales not touching leads. I've drafted a doc with marketing's commitments plus a 24-hour follow-up expectation on sales. Problem is our CRO won't sign anything and told me flat out we don't do SLAs on the sales side. Our head of sales ops is on board but has no authority. How do I finish this?",
"expected_output": "A refusal to ship the one-sided SLA, the bilateral and system-enforced rungs struck for lack of a sponsor with the gap recorded in the deliverable, and the paper-mapping session plus standardized definitions delivered instead.",
"files": [],
"expectations": [
"Refuses to finish or ship the one-sided SLA as drafted",
"States that a one-sided SLA makes marketing the defendant and sales the judge, poisoning the alignment it was meant to create",
"Given no cross-functional sponsor, strikes the bilateral SLA and system-enforced timer rungs from the design rather than shipping weakened versions of them",
"Delivers the lower rungs instead: the paper-mapping session and standardized handoff definitions as what is achievable without a sponsor",
"Records the struck rungs and the missing sponsor explicitly in the deliverable rather than leaving the gap unstated",
"Prescribes the paper-mapping session first, with both sides of the handoff in one room mapping first touch through to expansion and marking each handoff's owner, criteria, and where it breaks",
"Sizes that session at roughly an hour or two and notes the largest alignment gaps typically surface within the first half hour",
"Specifies handoff triggers as behavior-based criteria rather than a single point-score threshold",
"States that if a bilateral SLA is later ratified, its initial targets are set close to the current baseline because an unmeetable target destroys credibility immediately",
"States that system-enforced timers are promoted immediately once an SLA is ratified rather than parked as a later phase, because enforcement rather than definition is where handoffs fail",
"Names the obligations sales would owe in any bilateral version - acceptance window, follow-up, and disposition with reason codes - alongside marketing's volume and data-completeness commitments",
"Rejects a standalone monthly MQL volume target as a team goal, because per-layer volume metrics drive low-quality lead stuffing and falling down-funnel conversion"
]
},
{
"id": 9,
"prompt": "Fentari Cloud, usage-based infrastructure product, roughly $22M run rate. Contracts are annual commits but customers ramp: some take five months to get a real production workload onto us, some never do. Right now our funnel ends at Closed Won when the contract is signed, and that's the number marketing and finance both report against. The board is asking why bookings look great and revenue doesn't follow. Fix our funnel model.",
"expected_output": "A diagnosis of the single signature boundary as wrong for consumption revenue, two modeled boundaries, and ramp stages made first-class with their own conversion and time metrics plus a named owner.",
"files": [],
"expectations": [
"Diagnoses the single revenue boundary at contract signature as wrong for consumption-based revenue",
"Names the failure it produces: bookings celebrated while the ramp to real revenue goes unowned",
"Prescribes modeling two boundaries, the accounting commit for finance and first value delivered, rather than collapsing them into one",
"Carries time-to-first-impact or first value as a separate CAC-recovery measure rather than as a renamed stage",
"Extends the model past close with ramp stages as first-class, such as commit, first production workload, ramp curve, and steady-state run rate",
"Gives each post-commit stage its own conversion rate and its own time metric",
"States that customers consume the product before revenue books, and that this gap is where reporting breaks",
"Assigns a named owner to the onboarding and ramp segment rather than leaving it unowned between sales and CS",
"Notes that first-value measurement depends on product telemetry and flags that instrumentation cost instead of assuming it exists",
"Requires the single-boundary choice to be argued explicitly in the deliverable if only one boundary is ultimately modeled",
"Does not resolve the bookings-versus-revenue gap by changing the reported bookings number or reclassifying closed-won deals"
]
},
{
"id": 10,
"prompt": "Head of RevOps at Sablon Health, B2B, about $30M ARR. Two things landed on me this week. First, our new growth advisor says linear funnels are obsolete and we should replace the funnel model with a growth-loop model. Second, sales leadership wants the pipeline stages to double as the forecast categories, so Negotiate just means commit. Oh and the CEO added a line to my job description making me the owner of ICP definition and our pricing tiers. Tell me how to build all this.",
"expected_output": "Funnel and growth loops kept as separate artifacts answering different questions, stage and forecast category kept as two orthogonal fields, ICP and pricing declined as outside RevOps scope, and explicit recycle paths designed for non-linear movement.",
"files": [],
"expectations": [
"Keeps the funnel model and the growth-loop model as separate artifacts rather than replacing one with the other",
"States what the funnel answers that a loop model does not: how many SDRs, how much demand-gen budget, and what quota capacity a target requires",
"Characterizes the funnel as a capacity and resource-allocation instrument rather than a model of buyer psychology",
"Places growth loops in a channel-strategy artifact and declines to let either model answer the other's question",
"Refuses to let pipeline stages double as forecast categories",
"Prescribes two orthogonal fields: stage as position in the process, forecast category as confidence",
"Names the consequence of conflating them: both constructs degrade and stages inflate to signal confidence",
"Declines RevOps ownership of ICP definition and pricing posture, assigning both to commercial leaders",
"States what RevOps does own here: revenue-workflow definitions, definition governance, systems configuration and change control, and commercial analytics",
"Designs explicit recycle and regression paths so backward stage moves are recorded rather than treated as noise or blocked outright",
"Names backward-transition rate as a first-class metric to track"
]
}
],
"trigger_queries": [
{ "query": "Design a revenue funnel model for our Series B SaaS from scratch", "should_trigger": true },
{ "query": "we need funnel stages defined, marketing sales and CS each count something different", "should_trigger": true },
{ "query": "what stages should our brand new pipeline have", "should_trigger": true },
{ "query": "should we qualify buying groups instead of MQLs now that deals have 6 stakeholders", "should_trigger": true },
{ "query": "help me pick between the demand waterfall and the bowtie", "should_trigger": true },
{ "query": "is the bowtie model worth adopting for a recurring revenue business like ours", "should_trigger": true },
{ "query": "our board wants a revenue plan built on funnel math", "should_trigger": true },
{ "query": "how many stages should a mid-market SDR to AE pipeline have", "should_trigger": true },
{ "query": "we have no model for how many leads turn into deals and I need one before the annual plan", "should_trigger": true },
{ "query": "set our MQL to SQL to opportunity definitions", "should_trigger": true },
{ "query": "what conversion rate assumptions should go into next year's plan", "should_trigger": true },
{ "query": "where should marketing hand off to sales and where should sales hand off to CS", "should_trigger": true },
{ "query": "I need a bilateral SLA between marketing and sales for qualified leads", "should_trigger": true },
{ "query": "our win rate and our close rate disagree and nobody knows which one to use", "should_trigger": true },
{ "query": "what's the difference between close rate and win rate for planning purposes", "should_trigger": true },
{ "query": "we're moving upmarket and deals now have 5+ stakeholders, does our funnel need to change", "should_trigger": true },
{ "query": "build us the demand unit waterfall", "should_trigger": true },
{ "query": "can you build a funnel model that covers post-sale expansion too", "should_trigger": true },
{ "query": "should closed won be the end of our funnel or should it go further", "should_trigger": true },
{ "query": "our contracts sign months before customers actually use anything, how do we model that", "should_trigger": true },
{ "query": "I inherited a funnel model nobody trusts, rebuild it", "should_trigger": true },
{ "query": "how do I decide whether to model individual leads or whole accounts", "should_trigger": true },
{ "query": "what's the right unit of analysis for our pipeline", "should_trigger": true },
{ "query": "we're a two-sided marketplace, how do we even do funnel math", "should_trigger": true },
{ "query": "PLG company adding a sales assist motion, I need a model that covers both sides", "should_trigger": true },
{ "query": "should PQLs or PQAs be our qualification gate", "should_trigger": true },
{ "query": "need to work out how many SDRs and how much demand gen budget a $40M number requires", "should_trigger": true },
{ "query": "reconcile our bottom up funnel plan against the top down target", "should_trigger": true },
{ "query": "our CRM stage probabilities are just 10/25/50/75/90, is that fine for planning", "should_trigger": true },
{ "query": "where do I get defensible conversion rates out of our own CRM history", "should_trigger": true },
{ "query": "how often should we re-baseline our conversion assumptions", "should_trigger": true },
{ "query": "our MQL to SQL sits way below what the benchmark reports say, what do we do", "should_trigger": true },
{ "query": "set up governance for who is allowed to change our funnel definitions", "should_trigger": true },
{ "query": "we have three MQL definitions by region and it's chaos", "should_trigger": true },
{ "query": "our growth advisor says funnels are dead and we should use loops instead", "should_trigger": true },
{ "query": "our stages are also our forecast categories and it's a mess", "should_trigger": true },
{ "query": "high consideration B2C, mortgage brokerage, do the B2B funnel frameworks apply to us", "should_trigger": true },
{ "query": "I want one model that tells reps what to do and also tells the board how many leads we need", "should_trigger": true },
{ "query": "marketing counts leads, sales counts opps, finance counts bookings, make them tie together", "should_trigger": true },
{ "query": "we need a shared language between marketing and sales for how a deal progresses", "should_trigger": true },
{ "query": "build the conversion register for our planning model", "should_trigger": true },
{ "query": "what does TOFU MOFU BOFU actually map to in a CRM", "should_trigger": true },
{ "query": "60% of our new ARR comes from existing customers, how much of the model should cover renewal and upsell", "should_trigger": true },
{ "query": "seed stage, do I really need ten pipeline stages", "should_trigger": true },
{ "query": "what should the model look like at Series A versus Series C", "should_trigger": true },
{ "query": "help me brainstorm two or three different stage set designs and then pick one", "should_trigger": true },
{ "query": "usage based pricing, where should the funnel end", "should_trigger": true },
{ "query": "our channel partners sell completely differently, do they get their own stages", "should_trigger": true },
{ "query": "hey can you help me map out how a deal goes from first click to signed and who owns each step", "should_trigger": true },
{ "query": "audit our existing stage exit criteria, half of them are named after rep activity like demo scheduled", "should_trigger": false },
{ "query": "are our stage definitions buyer-verifiable, two managers can't agree on what qualified means", "should_trigger": false },
{ "query": "why is everything stuck in one stage", "should_trigger": false },
{ "query": "rewrite the exit criteria wording for each of our seven existing stages", "should_trigger": false },
{ "query": "our stage-skip rate and close-date push counts look terrible, diagnose the stages", "should_trigger": false },
{ "query": "where are we losing deals between demo and proposal, size it in dollars", "should_trigger": false },
{ "query": "our funnel is leaky, find the drop-off points and quantify them", "should_trigger": false },
{ "query": "quantify the recoverable revenue we lost to unworked inbound leads last quarter", "should_trigger": false },
{ "query": "our lead scores are wrong, sales rejects every MQL we send", "should_trigger": false },
{ "query": "set the point weights and the decay curve for our fit and engagement scoring model", "should_trigger": false },
{ "query": "what score threshold should make a lead an MQL", "should_trigger": false },
{ "query": "round robin is lopsided, three reps get everything", "should_trigger": false },
{ "query": "who should this inbound lead get assigned to, design the territory rules", "should_trigger": false },
{ "query": "why did we miss the number last quarter, our commit is never right", "should_trigger": false },
{ "query": "our reps are sandbagging and deals keep slipping, diagnose the forecast", "should_trigger": false },
{ "query": "clean up the pipeline before the QBR and flag every stale deal", "should_trigger": false },
{ "query": "list deals with no stage or amount change in 90 days and tell me what to do with each one", "should_trigger": false },
{ "query": "write the handoff packet CSMs need after a deal closes", "should_trigger": false },
{ "query": "our CSMs start from zero after closed won, fix the post-close transition", "should_trigger": false },
{ "query": "build a weighted account health score with bands and thresholds", "should_trigger": false },
{ "query": "which usage signals actually predict churn for our accounts", "should_trigger": false },
{ "query": "who owns each CRM field and which system wins when they conflict", "should_trigger": false },
{ "query": "our custom fields have sprawled to 400, build a field dictionary", "should_trigger": false },
{ "query": "finance and sales report two different ARR numbers, set the source of truth per object", "should_trigger": false },
{ "query": "build the KPI tree from board level down to IC with owners and guardrail metrics", "should_trigger": false },
{ "query": "what should our north star metric be", "should_trigger": false },
{ "query": "our board deck reads like a data dump, restructure the revenue section", "should_trigger": false },
{ "query": "what goes in the monthly revenue review versus the QBR", "should_trigger": false },
{ "query": "design the discount approval matrix and the margin floor policy", "should_trigger": false },
{ "query": "who has to approve a 40% discount on a non-standard deal", "should_trigger": false },
{ "query": "we have 34 GTM tools, decide what to keep, consolidate, or cut", "should_trigger": false },
{ "query": "new RevOps project, which skill do I need for this", "should_trigger": false },
{ "query": "where do I start with RevOps", "should_trigger": false },
{ "query": "am I ready for a RevOps manager role, review my resume", "should_trigger": false },
{ "query": "write the interview loop and take-home exercise for our first RevOps hire", "should_trigger": false },
{ "query": "which revops newsletters and podcasts should I subscribe to", "should_trigger": false },
{ "query": "recommend a funnel platform, ClickFunnels versus Kajabi versus Systeme", "should_trigger": false },
{ "query": "self-host a funnel builder on my own VPS", "should_trigger": false },
{ "query": "write the sales funnel copy for my course launch, landing page plus upsell plus order bump", "should_trigger": false },
{ "query": "audit my landing page for conversion rate, the hero section isn't converting", "should_trigger": false },
{ "query": "set up server-side conversion tracking with the Meta conversions API", "should_trigger": false },
{ "query": "build the data pipeline that loads our product events into BigQuery", "should_trigger": false },
{ "query": "design our deployment pipeline stages in CI", "should_trigger": false },
{ "query": "map our recruiting pipeline stages from applied through to offer", "should_trigger": false },
{ "query": "run a cohort retention analysis on our January signups", "should_trigger": false },
{ "query": "write the nurture email sequence for MQLs that sales rejected", "should_trigger": false },
{ "query": "our attribution model says paid search drives everything, settle the budget fight between channels", "should_trigger": false },
{ "query": "forecast next quarter's bookings from our current pipeline coverage", "should_trigger": false },
{ "query": "design our customer onboarding checklist for new enterprise accounts", "should_trigger": false }
]
}
references/conversion-methodology.md›
# Conversion Assumption Methodology
How the conversion register's numbers get derived, why the vocabulary matters, and what external benchmarks are actually good for.
## Cohort vs Milestone: the Distinction Most Models Get Wrong
- **Win rate** is a _period_ metric: of opportunities that closed in a period, what share were won.
- **Close rate** is a _cohort_ metric: of opportunities created in a period, what share are ultimately won.
- **Pipeline conversion rate** is a _snapshot_ metric: the rate at which a point-in-time pipeline converts to revenue.
A single worked case (from a Kellogg planning course) shows how much the reported number depends on which measure and analysis type is picked, for the same underlying business:
- Narrow win rate: 50%
- Broad win rate: 43% vs 29%
- Close rate: 30% vs 20%
The true value only emerges once open opportunities reach terminal state. Reporting a "conversion rate" without its basis invites exactly this confusion, which is why the register's basis column is mandatory.
**Decompose the close rate over time, not just in aggregate.** Knowing a cohort eventually closes 40% of its qualified leads is less useful than knowing the timing: derive average in-quarter, first-quarter, and second-quarter close shares, ideally from 6-8 matured cohorts rather than 2. Without the decomposition, lead generation cannot be phase-lagged into bookings - a plan can be arithmetically correct while being temporally wrong.
**The lag breaks naive coverage math.** With a 30-day sales cycle, quarterly pipeline coverage is close to meaningless: roughly two-thirds of the pipeline needed to close during the quarter has not been created yet when the quarter starts.
## Deriving Historical Rates
1. Sweep stale records first: across one vendor's customer base, more than 10% of pipeline sat 12+ months without a change to stage, close date, or amount. Rates computed over that population flatter the funnel.
2. Window: every unit that _entered_ a stage over 4-8 quarters, divided into those that eventually reached the next stage. Entries, never current occupancy - occupancy undercounts slow movers already past the stage.
3. Segment before trusting: a blended rate hides exactly the differences (source mix, price point, opportunity type) that separate two funnels in the same industry. Minimum viable segmentation set is motion x segment (SMB/MM/ENT) x source (inbound/outbound/partner/PLG) x opportunity type (new/renewal/upsell/cross-sell) - four dimensions, and sample size is the real constraint. Pick the two highest-variance dimensions, hold the rest blended until volume supports splitting.
4. Opportunity type moves rates more than most teams model: new-business acquisition shows the longest cycles and lowest conversion (low single digits at the opportunity level in some published figures). Retention/expansion converts faster and far more often (healthy retained share cited around 75-85% at the engage stage). At $50-100M ARR, 58% of new ARR comes from existing customers, 67% above $100M - a pre-sale-only model governs a third of the business at that scale.
## What External Benchmarks Are For
Benchmarks trigger investigation; they never set targets. The variance across publishers is definitional incompatibility, not noise - no two sources define MQL or SQL identically, and most do not disclose whether a rate is cohort- or milestone-based. One widely cited report drew ~106 participants split 5 ways - roughly 20 per cell, which supports no real assumption. Commonly cited B2B SaaS ranges, for the sanity-check role only:
- Visitor-to-lead: 1-5%
- Lead-to-MQL: 15-30%
- MQL-to-SQL: 30-50% (wider practitioner range centers nearer 13-15%)
- SQL-to-opportunity: roughly 42-75%, depending on source
- Opportunity-to-won: 15-30%
A stage far outside these bands earns a hypothesis and a look, nothing more.
| Source type | Good for | Caveat |
| ------------------------------------------------------ | ---------------------------------------------------- | -------------------------------------------------- |
| Practitioner surveys (e.g. Benchmarkit) | Breadth across categories | Self-reported; definitions vary by respondent |
| Long-running banker surveys (e.g. KeyBanc) | Time series since 2010 | Small n, lagged publication |
| Forrester / SiriusDecisions | Standardized definitions enable real peer comparison | Paywalled; requires adopting their stage set |
| Motion-segmented lifecycle benchmarks (Bowtie-aligned) | Post-sale coverage | Vendor-adjacent |
| VC portfolio reports | Expansion economics | Portfolio selection bias |
| Vendor blog "20XX benchmarks" posts | Nothing reliable | Frequently circular citations with no primary data |
## Re-benchmarking Cadence
- **Weekly** - operational monitoring only; no assumption changes.
- **Monthly** - recompute cohort rates as cohorts mature; flag drift beyond the register's tolerance band.
- **Quarterly** - re-ratify assumptions in the funnel council; version the model.
- **Trigger-based re-baseline** - ICP change, pricing change, new segment, motion change, comp-plan change, or stage-definition edit invalidates history. Reset the baseline; never blend old and new data into one number.
Validation to run alongside the cadence:
- Backtest the last 4 quarters of plan against actuals per stage.
- Hold out the most recent complete cohort rather than fitting on everything.
- Reconcile bottom-up against top-down and force the gap to be named.
Planning runs top-down and bottom-up meeting in the middle - the named gap is what surfaces an input shortfall quarters before it becomes a revenue miss.
references/framework-selection.md›
# Framework Selection
The four named frameworks, what each actually is, and the fit tables that narrow the choice. Remember the caveat from SKILL.md: the frameworks differ less than their marketing suggests - the real variance is unit of analysis, post-sale scope, and exit-criteria rigor. Selection buys a vocabulary and access to a benchmark pool.
## The Four Named Frameworks
**Demand Waterfall lineage (SiriusDecisions, now Forrester).**
- 2006: introduced as the Demand Generation Waterfall.
- 2012: re-architected, adding stages for inbound, teleprospecting, and sales-sourced leads.
- 2017: rebuilt around the _demand unit_, the buying group rather than the individual lead - "Prioritized Demand" replaces the marketing-qualified lead with a marketing-qualified account scored at the group level.
Seven stages run from Target Demand (TAM by firmographics) through progressively qualified demand to Pipeline Opportunity (close date and dollar value assigned) and Closed. It is the closest thing B2B has to a standard pre-sale waterfall, and a common language between marketing and sales leadership.
Misapplied when the company still qualifies leads individually: forcing demand-unit stages onto single-threaded deals adds reporting overhead without the account-based motion that justifies it.
**B2B Revenue Waterfall (Forrester, 2021+).** The current successor adds the opportunity mix: the Targeted stage carries targeted accounts, each holding one or more targeted opportunities typed as new-business, renewal, cross-sell, or upsell, so conversion rate and cost can be evaluated per opportunity type and resources allocated against the mix. This is the version with genuine post-sale coverage via opportunity type, fitting multi-product enterprise B2B where renewal/upsell/cross-sell are material.
Forrester itself warns organizations over-rotate on tooling for this migration when the real requirement is aligning process, data, and programs across teams. Do not recommend the swap below roughly 3-4 stakeholders per deal.
**TOFU/MOFU/BOFU.** Shorthand for the top three stages of the Awareness/Consideration/Conversion/Loyalty/Advocacy marketing funnel; TOFU carries the most volume and lowest conversion. A content and demand-gen planning lens (a commonly cited B2B SaaS content-spend mix is roughly 50/30/20 across the three), never a CRM stage model - making it double as the pipeline's stage set conflates marketing funnel stages with buyer-verifiable sales milestones.
**Bowtie (Winning by Design; 2023 "Bowtie Standard" adds a Prioritization stage).** Extends the funnel with an equally weighted post-sale half: acquisition/pipeline/close on the left, onboarding/adoption/expansion on the right, knotted at closed-won. The case for it: in recurring-revenue businesses growth increasingly comes from the right side, since expansion is 40%+ of new ARR above $50M.
Maps to five touch models (No/Low/Medium/High/Dedicated Touch) so one stage set flexes by motion. Misapplied on one-time-purchase or low-NRR businesses with no post-sale expansion economics to instrument.
## Motion Fit
| Motion | Stage count | Unit | Framework fit |
| --------------------- | ----------- | ------------------------------ | ------------------------------------------------------------- |
| No-touch / PLG | 3-4 | User, then account | Growth-loop/pirate-metrics lens plus a product-qualified gate |
| Low-touch assisted | 4-5 | Account (PQA) | Bowtie left side compressed, PQL/PQA gate |
| Mid-touch (SDR to AE) | 5-7 | Lead, then opportunity | 2012 Demand Waterfall or Bowtie |
| High-touch enterprise | 6-8 | Buying group, then opportunity | B2B Revenue Waterfall |
| Channel / partner-led | Mirror set | Partner-sourced opportunity | Separate pipeline, never extra stages |
5-7 stages cover most mid-market and enterprise motions; an eighth stage for long legal/security review is the exception. Prospecting activity belongs outside the pipeline entirely.
## Revenue-Model Fit
- **Subscription** - Bowtie or a Forrester waterfall both work; revenue books at commit, so the pre-sale stage set carries the planning weight.
- **Usage/consumption-based** - extend past close into the ramp as first-class stages: commit, first production workload, ramp curve, steady-state run rate, each with its own conversion and time metric. Customers use the product before revenue books, and that gap is where reporting breaks.
- **Marketplace/take-rate** - two funnels (supply and demand) with a liquidity constraint between them. No mainstream framework handles this; the model is custom by necessity, and the deliverable should say so.
## Scaling by Company Stage
| Dimension | Pre-PMF / Seed | Series A-B | Series C+ / public |
| ------------------ | ------------------------------------------- | ----------------------------------- | ------------------------------------------------------ |
| Stage count | 4-5 | 5-6 | 6-8, separate pipelines per motion |
| Unit | Contact | Lead, then opportunity | Buying group, then opportunity |
| Conversion source | Benchmarks and judgment, labeled as guesses | Own data, wide confidence intervals | Own data, segmented, cohort-based |
| Post-sale modeling | None | Onboard + renewal | Full post-sale side with expansion mix |
| Governance | Founder decides | Named owner, quarterly review | Funnel council, versioned definitions, data dictionary |
The dominant early-stage failure is premature complexity - a 10-stage pipeline modeling a future enterprise process. Add a stage only when an operational need forces it.
## PLG Unit Choice: PQL vs PQA
A product-qualified lead is user-centric - one person's product activity. A product-qualified account is a group of connected users showing purchasing potential as a whole.
Individual users swiping a card for $25-50/month are not where large PLG organizations earn the bulk of revenue, so a PQL-only model systematically misdirects sales capacity. Segment PQLs into at least three buckets by what the account needs next, starting with hand-raisers who actively want to talk to sales.
## The MQL vs Buying-Group Call
The buying-group evidence is real. Forrester's 2023 Buyers' Journey Survey found:
- 93% of B2B buyers in a group of 2+
- 71% of B2B buyers in a group of 4+
The strongest published outcome (a Forrester client story, read as a vendor case) reports multi-contact opportunities 8x likelier to advance and a 17% higher closed-won rate.
The counter-case is underweighted: where decisions are made by an individual or small group and cycles are short, the MQL remains a practical operating model. The migration is a costly data-model change, not a metric rename - below roughly 3-4 stakeholders per deal the individual-lead model is simpler and loses little.
The MQL is a bad goal but an acceptable state.
references/handoff-sla-design.md›
# Handoff and SLA Design Mechanics
The model-level design of the funnel's three cross-functional handoffs. Execution of any single handoff belongs to the sibling skills named in SKILL.md's Reference section.
## The Paper-Mapping Exercise
Put both sides of a handoff in one room and map the funnel on paper - the complete journey from first marketing touch to expansion - marking every handoff point with who is responsible, the criteria, and where it breaks in practice. Practitioners report the biggest alignment gaps surface in the first 30 minutes, which makes this the mandatory cheap first step before any SLA infrastructure is designed. A widely repeated framing of what it catches: misaligned handoffs run "like a relay race where both runners are sprinting in opposite directions."
## Bilateral SLA Structure
A one-sided SLA is not an SLA. Structure every handoff agreement with obligations both ways:
- Marketing commits to what it delivers (volume, criteria, data completeness); sales commits to what it does with it (acceptance window, follow-up, disposition with reason codes).
- Document the consequence, not just the commitment: if sales misses the follow-up window, marketing is not accountable for that lead's conversion; if marketing sends leads below agreed criteria, sales carries no follow-up obligation on them.
- Set initial targets close to current baseline. An aggressive target nobody can meet destroys the SLA's credibility immediately.
- Operationalize triggers as behavior-based criteria, not a single point-score threshold.
**Representative timer structure** (adapt to the user's cycle, never copy blind): sales accepts or rejects each qualified lead within 8 working hours or it auto-reverts to marketing for nurture/redistribution; it advances to sales-qualified within 4 days or reverts again. CRM activity is the elapsed-time counter; unaccepted leads escalate to the sales manager, reroute, or return to marketing.
## Governance Cadence and Versioning
- RevOps watches live response times, data completeness on handed-off records, and rejection reason codes as the early-warning layer.
- Both teams review accepted/rejected/recycled volumes and disputed dispositions monthly.
- Leadership - the funnel council - re-ratifies definitions and targets quarterly. Council membership, decision rights, and the change-control process live in the governance charter artifact.
- The SLA itself is versioned, with old versions kept for audit. Enforcement, not definition, is the usual failure point: an SLA nobody reviews is just a document, so compliance metrics sit on a standing joint agenda, not in a drawer.
**RevOps scope discipline as the owner of this machinery:** own revenue-workflow definitions, definition governance, systems configuration and change control, commercial analytics, and operational enablement. Never own segmentation, ICP definition, offer bets, or pricing posture - those belong to commercial leaders. Owning the definitions and instrumentation while a commercial leader owns the strategy those definitions measure is what keeps accountability from corrupting.
## The Sales-to-CS Handoff as a Data Contract
At the model level this handoff is a data contract, not a meeting. Required payload:
- Buying-group roles
- Identified pain and its metrics
- Compelling event
- Success criteria
- Implementation scope
- Renewal risk factors
The account executive documents renewal risk during the sale so the CSM inherits a risk profile, not just a revenue number.
Enforce with an assignment gate: CS ownership confirms only after the handoff document is completed and reviewed, with RevOps tracking document completeness to spot recurring gaps. The CS-to-sales return path (renewal/expansion) mirrors it: defined expansion signals, a named owner for the expansion opportunity, and an agreed point where CS hands a qualified expansion back into the pipeline.
## Directional Evidence, Not Promises
Formal marketing-sales alignment SLAs correlate with stronger reported program ROI (65% of companies with them report strong ROI in HubSpot State of Inbound data), and analyst-cited research on aligned revenue operations reports roughly 19% faster growth and 15% higher profitability. Quote these as directional support for investing in the handoff layer - never as an outcome to promise a specific company.
references/worked-examples.md›
# Worked Examples
Three shapes of the engagement: a mid-touch B2B design showing the candidate-comparison step, a PLG design showing the unit switch, and one design done wrong. Structures and figures follow the sourced tables in the other references; per-company numbers below are illustrative placeholders to show the artifact shape, and a real engagement derives them from the user's own history.
## Example 1: Mid-Touch Series B SaaS (candidate comparison)
Interview findings:
- SDR-to-AE motion
- Subscription revenue
- 2-3 stakeholders per deal
- ~20% of new ARR from existing customers
- 6 quarters of CRM history
- Planning is the primary purpose
- Hard deadline at the annual plan
Prior decisions settled:
- Model: planning
- Unit: lead, switching to opportunity at qualification
- Boundary: commit for finance, plus time-to-first-value tracked separately
Candidates presented:
- **A. Compressed 5-stage (2012 Demand Waterfall vocabulary):** Inquiry -> Marketing-Qualified -> Sales-Qualified -> Opportunity Validated -> Commit. Optimizes for clean planning math on thin data; lowest admin cost. Post-sale left to a renewal date field.
- **B. 6-stage with post-sale skeleton (Bowtie-leaning):** adds Onboarded/Adopting after Commit. Optimizes for the expansion share the company expects to grow; costs product-telemetry instrumentation it does not have yet.
- **C. 7-stage buying-group model (Demand Unit vocabulary):** qualification at the account level. Rejected in the write-up, not just demoted: at 2-3 stakeholders per deal the migration cost buys little (below the 3-4 threshold).
Recommendation: A now, with B's post-sale stages as the register's first trigger-based extension once expansion share or telemetry justifies them. The deadline promoted the low-admin candidate; the compounding mandate got a scheduled upgrade path instead of being silently dropped.
Register excerpt (shape, not gospel):
```
transition rate basis n source refresh trigger
MQL -> SQL 22% cohort 1,840 own data, 6 quarters quarterly council
SQL -> Opp 48% cohort 405 own data, 6 quarters quarterly council
Opp -> Won 24% cohort 194 own data, 6 quarters quarterly council
close-rate lag 55/30/15% in/1st/2nd quarter, 6 cohorts monthly recompute
```
Bottom-up plan from these rates landed 8% short of the top-down target; the gap was named in the funnel model with two closure options (MQL volume vs SQL-to-Opp improvement), not averaged away.
## Example 2: PLG Moving Sales-Assisted
Interview findings:
- No-touch self-serve core
- A new sales-assist motion for teams
- Usage-based pricing
- Thousands of signups monthly
Design: two connected constructs, not one funnel. Self-serve side measured with cohort/event analytics (signup -> activation -> paid conversion), no per-deal stage state. From the product-qualified gate onward, a 4-stage pipeline at the _account_ unit: PQA identified -> Sales-assist engaged -> Commit -> Ramp-to-steady-state (usage revenue makes the post-commit ramp a first-class stage with its own conversion and time metrics).
PQLs bucketed three ways, hand-raisers routed first. The written model states explicitly where the unit switches from user to account and which side of the gate each metric lives on.
## Example 3: Done Wrong (negative example)
A seed-stage company shipped a 10-stage pipeline copied from an enterprise template, stage probabilities left at CRM defaults, one blended "conversion rate" reported weekly, and an SLA that obligated only sales.
What failed, mapped to the rules it broke:
- Ten stages at pre-PMF is premature complexity - half the stages saw no operational use, and reps parked deals in stage 2 until close, corrupting every rate.
- CRM default probabilities are deleted-menu items: nobody derived them from this business, and multiplied against optimistic deal values the error compounded exactly in the middle stages where most pipeline sat.
- The blended weekly "conversion rate" mixed a period win rate with a cohort close rate depending on who pulled it - the same business read three different numbers in one leadership meeting, and no register existed to say which basis was which.
- The one-sided SLA made marketing the defendant and sales the judge; it was ignored within a quarter because nothing enforced it and no consequence was documented.
The fix followed this skill's workflow: purpose and unit settled first, stage count cut to 4 with guesses labeled as guesses, a register with basis and provenance columns, and the SLA rebuilt bilaterally with a paper-mapping session as the first step.
SKILL.md›
---
name: revenue-funnel
description: Design a company's revenue funnel model from scratch, at the macro level - the stage set, the unit of analysis (lead, buying group, account), the model purpose (process vs planning vs forecast), the cohort-based conversion assumptions behind the plan, and the ownership handoffs between marketing, sales, and CS. Use whenever the user mentions funnel design, funnel stages, a funnel model, conversion rate assumptions, a bowtie model, a demand waterfall, MQL to SQL to opportunity definitions, or marketing-sales handoff design - even if they never say "funnel". Covers B2B and high-consideration B2C. Do NOT use for auditing an existing pipeline's stage definitions - use mbfinotti/revops-skills@pipeline-stage-definition-audit instead.
license: MIT
metadata:
author: Maya-Beth Finotti
version: "1.4.7"
---
# Revenue Funnel Design
You are the strategic partner to a RevOps function head designing the company's revenue funnel model from scratch. The deliverable is a versioned artifact set (Output Shape below), not a slide - a stage set with owners, a conversion register with provenance, and the handoff/SLA layer connecting marketing, sales, and CS. Design the model; never execute inside it.
## Ground Rules
- Macro only. Design which stages exist, what unit moves through them, whose numbers feed the plan, and who owns each handoff. Never write individual stage exit-criteria wording, lead-routing rules, or handoff execution playbooks - route those to the sibling skills in Reference.
- Two decisions come before counting stages: what the model is _for_ (process vs planning vs forecast) and what unit moves through it (lead, contact, buying group, opportunity, account). Conflating the three purposes into one stage set is the root cause of most bad funnel design.
- Conversion assumptions are cohort-first. Win rate is a period metric, close rate is a cohort metric, pipeline conversion rate is a snapshot metric. One worked case reads a 50% win rate and a 30% or 20% close rate for the same underlying business, depending only on which measure and analysis type someone picked.
- Every rate in the model names its basis or it does not ship.
- Published benchmarks trigger investigation, never set targets. Sources disagree because their stage definitions disagree, and sample cells are tiny - one widely cited report splits ~106 respondents 5 ways, leaving ~20 companies per cell. A stage far outside a published range is a question to investigate, not a gap to close.
- The funnel is a capacity and resource-allocation instrument, not a model of buyer psychology. It answers how many SDRs, how much demand-gen budget, and what quota capacity a target requires - questions no growth-loop model answers. Keep loops as a separate channel-strategy artifact; never let either pretend to answer the other's question.
- The MQL is a bad goal but an acceptable state. Migrating to buying-group qualification is a costly data-model change - do not recommend it below roughly 3-4 stakeholders per deal, whatever the current discourse says.
- RevOps owns definitions, instrumentation, and change control for this model - never segmentation, ICP, offer bets, or pricing posture. Those belong to commercial leaders; say so when the engagement drifts there.
- Add a stage only when a real operational need forces it. The dominant early-stage failure is a 10-stage pipeline modeling an enterprise process the company does not have yet; every extra stage is admin overhead forever.
- Label the provenance of every number: the user's own data, a named published source with its caveat, or an explicit guess.
- Every ranked menu below states a default order, not a law - it shifts with context and with who executes it. Re-rank it against the Interview answers and against what this company already has: an analyst who can already run cohort SQL, admin capacity sitting idle, an executive sponsor already in the room. Name which answer moved which option when presenting the reordering.
## Interview
Ask before designing anything. One question per message; multiple-choice where possible; skip anything already answered.
- What triggered this: no funnel model exists, an inherited model nobody trusts, a motion or product change invalidated the old one, or leadership asked for a revenue plan?
- Which question must the model answer first: what people should do at each step (process), how many inputs are needed to hit the number (planning), or what closes this quarter (forecast)? Pick one primary; the others get derived views, not extra stages.
- Which GTM motion(s): no-touch/PLG self-serve, low-touch assisted, mid-touch (SDR qualifies then AE), high-touch enterprise, channel/partner-led? More than one running at once?
- Revenue model: subscription, usage/consumption-based, marketplace/take-rate, one-time purchase?
- How many stakeholders participate in a typical deal: 1-2, 3-4, 5+? This decides the individual-lead vs buying-group unit question.
- Company stage: pre-PMF/seed, Series A-B, Series C+ or public?
- Roughly what share of new revenue comes from existing customers (renewal, upsell, cross-sell)? This decides how much post-sale funnel the model must carry.
- What data exists: how many quarters of stage history, in a CRM or a warehouse, clean enough to compute cohort rates? Has stale pipeline been swept recently?
- What definitions already circulate - is there one MQL definition or several by region/team? Who owns definitions today, and is there any standing forum that could ratify changes?
- B2B, B2C, or both? If B2C: sales-assisted high-consideration (insurance, property, enrollment) or transactional/self-serve volume?
- By what date must the model land, and does a planning cycle (annual plan, board meeting) depend on it? A hard deadline promotes shipping the model on blended rates and scheduling segmentation as a follow-up.
- Is this a one-off plan input or a standing operating model? One-off compresses governance to a named owner and a review date; a compounding mandate makes the funnel council and the re-benchmarking cadence non-optional.
- What is the effort ceiling: analyst time for cohort analysis, admin capacity to instrument new definitions, and political capital to put marketing and sales under a bilateral SLA? No cross-functional sponsor strikes the SLA layer down to a paper agreement and says so in the deliverable.
## Workflow
1. Run the Interview. Confirm the scope boundary: model design, not an audit of existing stage definitions and not execution inside the model.
2. Settle the two prior decisions and the revenue boundary (The Two Prior Decisions below). Do not proceed to stage counting until the user validates all three.
3. Choose the framework vocabulary (Framework Vocabulary below; detail in [references/framework-selection.md](references/framework-selection.md)).
4. Enter explicit brainstorming: present 2-3 candidate stage-set designs per Candidate Designs below, with trade-offs and a recommendation. Get the user's pick before building anything on top of it.
5. Build the conversion register per Conversion Assumptions below, methodology in [references/conversion-methodology.md](references/conversion-methodology.md).
6. Design the ownership and handoff layer per Ownership Handoffs below, mechanics in [references/handoff-sla-design.md](references/handoff-sla-design.md).
7. Set the governance shell: who ratifies definition changes, the re-benchmarking cadence tiers, and version control for the model itself.
8. Emit the deliverable (Output Shape below) one artifact at a time for user validation, grounded in the matching worked example from [references/worked-examples.md](references/worked-examples.md). Never present the whole model as a fait accompli.
9. Check the Pass Threshold; iterate until it holds or every remaining gap is explicitly scheduled (e.g. cohort decomposition waiting on data volume).
10. If your harness has persistent memory, store the settled purpose, unit, boundary, stage set, and assumption register so later planning and reporting runs start from the decided model instead of re-interviewing.
## The Two Prior Decisions
**Model purpose.** A process model (what reps do), a planning model (how many inputs hit the number), and a forecast model (what closes this quarter) have different optimal stage sets. Starting pipeline is a mix of opportunities created over the past 1-4 quarters, so the cohort-based close rate and the current-quarter conversion rate are different instruments - track them separately rather than forcing one stage set to produce both.
**Unit of analysis.** Lead, contact, buying group, opportunity, or account - this single choice determines everything downstream, and it is the actual difference between the named frameworks' generations. Decide it from the Interview's stakeholders-per-deal answer, not from framework fashion.
**Where pre-sale ends.** Three defensible boundaries, each with a real cost:
- Contract signature: standard, auditable, wrong for consumption revenue.
- First value delivered: economically honest, needs product telemetry.
- Onboarding completion: convenient, but a delivery milestone rather than a buyer milestone.
Default recommendation: model two boundaries - book the accounting boundary at commit for finance, and carry time-to-first-impact as a separate CAC-recovery measure. Collapsing them into one is how companies celebrate bookings while losing the customer in year one, with onboarding owning neither a metric nor an owner.
## Framework Vocabulary
This menu is deliberately not ranked by efficiency. The named frameworks differ far less than their marketing suggests: "Engaged Demand", "Detected", and "Awareness" are the same stage under different sponsors, and the real variance across frameworks reduces to unit of analysis, post-sale scope, and exit-criteria rigor. Selection is a mapping from motion and unit, plus a judgment about whose benchmark pool and whose vocabulary the board already speaks - a ranking would be false precision.
- Mid-touch, individual-lead qualification -> 2012-era Demand Waterfall vocabulary.
- Account-based, buying-group qualification (3-4+ stakeholders) -> Demand Unit Waterfall (2017) or, where renewal/upsell/cross-sell mix must be modeled per opportunity type, the B2B Revenue Waterfall (2021+).
- Recurring revenue with material expansion economics -> Bowtie (equal-weight post-sale half; expansion is 40%+ of new ARR above $50M, ~58% at $50-100M, ~67% above $100M).
- TOFU/MOFU/BOFU -> content and demand-gen planning lens only; never a stage set for the pipeline.
- Marketplace/take-rate -> no published framework fits; build two funnels (supply, demand) with a liquidity constraint between them, and say the model is custom.
Adopting a named framework wholesale has one under-appreciated payoff: standardized stage definitions are what make external peer benchmarking meaningful at all. Read [references/framework-selection.md](references/framework-selection.md) before presenting the choice.
## Candidate Designs
Brainstorm before recommending. Present 2-3 candidate stage-set designs; specify each one as:
- Stage list with per-stage owner
- Unit of analysis, and where it switches (e.g. lead -> opportunity)
- Post-sale coverage
- What the design optimizes for
- Its standing admin cost
Follow with a trade-off comparison and one recommendation with reasoning. Let the user pick or blend; validate the pick before the conversion register is built on it.
Keep candidates inside the motion's stage-count band:
- No-touch/PLG: 3-4
- Low-touch: 4-5
- Mid-touch and most mid-market/enterprise: 5-7
- High-touch enterprise: 6-8
An eighth stage for long legal/security review is the exception, not the default. A genuinely different buying process (channel/partner) gets a separate mirrored pipeline, never extra stages bolted onto the main one. Scale complexity to company stage (table in [references/framework-selection.md](references/framework-selection.md)): pre-PMF gets 4-5 stages and guesses labeled as guesses, not a Series-C governance apparatus.
## Conversion Assumptions
Where the numbers come from, ranked by planning accuracy per analyst-hour. Vendor and CRM default stage probabilities are deleted from this menu entirely, not demoted - they are round numbers nobody derived from this business, optimistic by design, and the error compounds when multiplied against optimistic deal values.
- value: `own segmented cohort rates > own blended cohort rates > adjusted external benchmarks`
- effort: `own segmented cohort rates > own blended cohort rates > adjusted external benchmarks`
- efficiency: `own blended cohort rates > own segmented cohort rates > adjusted external benchmarks`
Default rung: blended cohort rates from the user's own history, segmented on the two dimensions with the most variance the volume can support. Move up to fuller segmentation (motion x segment x source x opportunity type) only where entries per cell stay large enough to trust - sample size, not ambition, sets the split. Move down to adjusted external benchmarks only pre-PMF or after a trigger reset wipes usable history, and label every such figure an explicit guess.
The order starves full four-dimension segmentation; promote it as volume grows, because opportunity type alone moves rates more than most teams model (new-business acquisition converting at low single digits while renewal/expansion runs far higher).
Method, whatever the rung:
1. Sweep stale pipeline first - across one vendor's base, over 10% of pipeline sat 12+ months untouched, and it distorts every rate computed over it.
2. Compute per stage over a 4-8 quarter window: units that entered the stage, divided into units that eventually reached the next stage. Count entries, never current occupancy.
3. Record each rate's basis (cohort or milestone), numerator/denominator definition, sample size, source, and refresh trigger in the conversion register.
4. Decompose the close rate over time from 6-8 cohorts: in-quarter, first-quarter, second-quarter shares. Without the lag decomposition a plan can be arithmetically correct and temporally wrong - at a 30-day cycle, quarterly coverage math is near-meaningless because roughly two-thirds of the quarter's closing pipeline does not exist at quarter start.
5. Reconcile the bottom-up funnel plan against the top-down revenue target and name the gap explicitly. Never average it away - the named gap is what surfaces a 10% MQL shortfall two quarters before it becomes a revenue miss.
6. Set the re-check cadence:
- Weekly: operational monitoring, no assumption changes.
- Monthly: cohort recompute with drift flags.
- Quarterly: re-ratification and versioning.
- Trigger-based: full re-baseline on ICP, pricing, segment, motion, comp-plan, or stage-definition change.
A trigger invalidates history, so reset rather than blend.
## Ownership Handoffs
Three handoff points cross functional boundaries:
- Marketing-to-sales: qualified lead accepted.
- Sales-to-CS: closed-won to onboarding.
- CS-to-sales: renewal/expansion.
Each needs an owner on both sides, explicit handoff criteria, and a bilateral SLA. Four investment rungs below, ranked by alignment gained per unit of effort and political capital:
- value: `system-enforced timers > bilateral SLA > standardized definitions > paper mapping`
- effort: `system-enforced timers > bilateral SLA > standardized definitions > paper mapping`
- efficiency: `paper mapping > standardized definitions > bilateral SLA > system-enforced timers`
1. **Paper mapping** - both sides of each handoff in one room, mapping the funnel first touch to expansion, marking every handoff's owner, criteria, and where it breaks. An hour or two; practitioners report the biggest alignment gaps surface in the first 30 minutes. Always do this first, inside the design engagement itself.
2. **Standardized definitions** - one written definition per handoff trigger (what counts as qualified, what triggers the handoff), behavior-based rather than a single point-score threshold. Days of drafting and sign-off.
3. **Bilateral SLA** - commitments, timers, and consequences on both sides, targets set near current baseline. A one-sided SLA makes marketing the defendant and sales the judge, and poisons the alignment it was meant to create. Weeks of negotiation; needs a cross-functional sponsor.
4. **System-enforced timers** - acceptance clocks, auto-reversion, escalation, and reassignment built into the CRM/automation layer, with compliance reviewed in a standing joint cadence. Admin build plus retraining.
The efficiency order starves enforcement, and enforcement - not definition - is where handoffs actually fail: an SLA nobody's system tracks decays into a document within a quarter. Promote rung 4 the moment the SLA is ratified rather than parking it as a later phase; if the Interview found no sponsor for a bilateral SLA, strike rungs 3-4 from the design, deliver rungs 1-2, and record the gap in the deliverable rather than pretending a paper agreement will hold. Mechanics, representative timers, and the sales-to-CS data contract are in [references/handoff-sla-design.md](references/handoff-sla-design.md).
## Output Shape
The deliverable is an artifact set, presented one file at a time for validation:
```
stage-definitions.md : per stage - name, definition, entry criteria, owner,
unit, system-of-record field (exit-criteria wording is
drafted downstream by the stage-definition audit skill)
conversion-register.csv: rate, numerator/denominator, cohort-or-milestone basis,
sample size, source/provenance, last refreshed,
next-refresh trigger
funnel-model : bottom-up plan with lagged cohort conversion, reconciled
against the top-down target, gap named explicitly
sla.md : bilateral commitments, timers, consequences, escalation,
version history - one per handoff point
data-dictionary.md : object model, field ownership, allowed values, reason
codes for the fields the model depends on
governance-charter.md : funnel council membership, decision rights, definition
change control, re-benchmarking cadence
capacity-model : headcount and budget implications derived from the
funnel model, never independently authored
```
Definitions live in the CRM and marketing automation platform; measurement lives in the warehouse, so history can be restated when a definition changes without corrupting past reporting.
## Pass Threshold
- Every stage has a named owner, an entry definition, a unit of analysis, and a system-of-record field. No stage exists without an operational need the user can state.
- Every rate in the conversion register carries its basis (cohort or milestone), numerator/denominator, sample size, provenance, and refresh trigger. No unlabeled defaults survive.
- The bottom-up plan and top-down target are reconciled with the gap stated as a number, not averaged away.
- Where 4+ quarters of history exist, the model is backtested against them and stages whose modeled vs actual rates diverge beyond the agreed tolerance band are flagged with a hypothesis.
- Both revenue boundaries (accounting commit and first value) are modeled, or the single-boundary choice is explicitly argued in the deliverable.
- Every handoff point has obligations on both sides, a timer, a consequence, and an escalation path - or a recorded statement of which rungs were struck and why.
Iterate until all six hold. Cohort decomposition needs 6-8 matured cohorts to be trustworthy; if that volume does not exist yet, ship on the available basis, label it, and schedule the upgrade as the register's first refresh trigger.
## Common Failure Modes
Deliberately unranked: each row is one diagnosis with one fix, not competing options for one goal. Apply every row that matches.
| Defect | Consequence | Fix |
| ------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------- |
| Definition drift across regions/teams (documented case: 3 MQL versions, 30+ lead statuses) | Leads stall between functions; win-rate reporting varies by who reports it | One definition set, council-ratified, versioned; regional variance needs council approval |
| Per-layer volume metrics per team | Marketing stuffs low-quality leads; down-funnel conversion tanks and sales takes the blame | Shared full-funnel metrics alongside layer metrics; review conversion jointly |
| One blended conversion rate per transition | Hides the segment/source/opportunity-type spread that drives the real plan | Segment on the two highest-variance dimensions volume supports |
| Stage set doubling as forecast categories | Both constructs degrade; stages inflate to signal confidence | Two orthogonal fields: stage = position in process, category = confidence |
| Linear-journey assumption | Backward moves and loops recorded as noise or blocked outright | Design explicit recycle/regression paths; measure backward transitions as a first-class metric |
| Single revenue boundary at signature on consumption revenue | Bookings celebrated while the ramp to real revenue goes unowned | Model commit and first-value boundaries separately |
| PQL-only qualification in PLG | Sales capacity aimed at individual card-swipers instead of accounts | Add the product-qualified account; bucket PQLs, hand-raisers first |
| Attribution used to settle channel budget fights | Software attribution structurally favors last-touch lower-funnel channels | Treat attribution as directional; keep the fight out of the funnel model's scope |
| Stale pipeline left in the baseline | Every derived rate flatters the funnel | Hygiene sweep before any rate is computed |
## B2B and B2C
B2B and B2C split into three cases:
- **Sales-assisted, high-consideration B2C** (insurance, property, solar, enrollment, automotive, mortgage): the full model transfers unchanged, since these run true stage pipelines and the purpose/unit/boundary decisions apply as-is. The unit is usually a household or applicant rather than a buying group.
- **Transactional/self-serve B2C**: has no per-deal stage state at all. The correct instruments are lifecycle stages and cohort/event-funnel analytics; the model this skill produces would be the wrong artifact. Say so, and design a lifecycle/cohort measurement plan instead of forcing a pipeline.
- **PLG**: sits between the two - cohort math on the self-serve side, a compressed stage model from the product-qualified gate onward.
## KPIs
- Track whether the model works, not whether it exists:
- Absolute gap between day-one plan and actuals per stage per quarter
- Assumption drift vs the register's tolerance bands
- Backward-transition rate
- SLA compliance per handoff (response time, disposition rate)
- The named bottom-up/top-down gap trending toward zero across planning cycles
- Governance health: the register's last-refreshed dates are current, and definition changes went through the council rather than around it.
- Directional context only, never a promise: formal marketing-sales SLAs correlate with stronger reported program ROI, and aligned revenue operations with faster growth - both are survey/analyst findings, not guarantees for a specific company.
## Invocation Examples
- "We're a Series B SaaS with an SDR-to-AE motion and no real funnel model - marketing, sales, and CS each count different things. Design one from scratch."
- "Our board wants a revenue plan built on funnel math. Help me set the stage model and defensible conversion assumptions from our CRM history."
- "We're moving upmarket and deals now have 5+ stakeholders. Should our funnel qualify buying groups instead of MQLs, and what would the model look like?"
## Reference
- Read [references/framework-selection.md](references/framework-selection.md) when choosing the vocabulary and stage-count band.
- Read [references/conversion-methodology.md](references/conversion-methodology.md) when building the register.
- Read [references/handoff-sla-design.md](references/handoff-sla-design.md) when designing the handoff layer.
- Read [references/worked-examples.md](references/worked-examples.md) when shaping the deliverable.
- See `mbfinotti/revops-skills@pipeline-stage-definition-audit` for auditing an existing stage set's exit criteria - this skill designs the model those stages live in; that one drafts the buyer-verifiable wording per stage.
- See `mbfinotti/revops-skills@lead-scoring` for the scoring logic feeding a qualification gate.
- See `mbfinotti/revops-skills@lead-routing` for the assignment logic feeding a qualification gate.
- See `mbfinotti/revops-skills@sales-to-cs-handoff` for executing the post-close handoff.
- See `mbfinotti/revops-skills@sales-pipeline-hygiene` for the stale-deal sweep that must precede rate derivation.
- See `mbfinotti/revops-skills@revenue-kpi-framework` for the metric hierarchy this model's KPIs roll into.