Build funnel metrics the meeting will actually trust.
Conversion by stage, velocity, and win rate with every definition pinned, and the leak named.
Inside: Stage conversion, time-in-stage, win rate, and a definition ledger.
Install it in one line, or paste it in.
~/.claude/skills/ and runs automatically when it is relevant.Connect your context. Set it to your motion.
it reads the full deal history and stage timestamps automatically, across every cohort.
it can build true entered-stage cohorts instead of inferring from a current snapshot.
pushes the pinned definitions into a dashboard so the whole team reads one number.
This was built for a B2B SaaS org running a staged pipeline. Set these to your stack:
Pin your win definition and denominator rule before anything else. Most funnel arguments are really definition arguments in disguise.
| Set this | What it is | Default / Example |
|---|---|---|
| CRM | your CRM connector | a CRM (Salesforce |
| STAGE model | your ordered stage list | your opportunity stages |
| WIN definition | what counts as won | Closed Wonexcluding no-decision |
| DENOMINATOR rule | what each rate divides by | deals that entered the stagenot all deals |
| VELOCITY basis | how you measure time-in-stage | stage-entry to next-stage-entry dates |
| COHORT window | the period you group deals by | created quarteror entered-stage month |
| SEGMENTS | cuts you compare | segmentsourceownerproduct |
Everything the skill does, in full.
Reads your stage data and builds the core funnel: how many deals convert from each stage to the next, how long they sit in each stage, what share of qualified deals win, and where the biggest drop-off is. Every metric ships with its definition attached, so the conversation is about the leak, not about whether the number is real.
- 1Stage-to-stage conversion
For each adjacent stage pair, divide deals that reached the later stage by deals that entered the earlier one. State the denominator out loud. A rate that divides by "all deals ever" and a rate that divides by "deals that entered this stage" are different numbers, and only one of them is honest about the leak.
- 2Time-in-stage velocity
Measure median days in each stage, not mean. One 400-day zombie deal drags a mean and hides the typical path. Use stage-entry timestamps where you have them and say so where you had to infer from close date.
- 3Win rate
Win rate is wins divided by resolved deals (won plus lost), with no-decisions handled explicitly and stated. Report it on a trailing window big enough to mean something, never on this week's handful.
- 4Leak detection
Rank the stage transitions by drop-off and name the single worst one. The biggest leak is where the funnel loses the most deals relative to what entered, not the stage with the lowest raw count. That is the one place a fix moves the whole number.
- 5Definition ledger
Every metric prints with its definition and denominator beside it. This is the trust layer. When someone questions a number in the room, the answer is already on the page.
- Every rate shows its denominator. No naked percentages.
- Velocity is median, and the date basis is named.
- Win rate states how no-decisions were handled, every time.
- Small samples are flagged, not smoothed. A rate on nine deals says so.
FUNNEL METRICS · created Q cohort · 620 deals Transition Conversion Median days Note Lead -> Qualified 44% 9 denom: leads created Qualified -> Demo 71% 6 denom: entered Qualified Demo -> Proposal 52% 14 biggest leak Proposal -> Won 61% 11 denom: entered Proposal Win rate (trailing 2Q): 34% of resolved deals. No-decisions excluded, counted separately. Biggest leak: Demo -> Proposal. Half of demoed deals never get a proposal. Next: capture stage-entry timestamps to replace inferred velocity on 2 stages.
The cohort window, the trailing win-rate window, and the median-over-mean choice are defaults, not laws. They suited a mid-market SaaS cycle. If your deals run longer or your volume is thinner, widen the windows. The definitions are the point, and they are yours to set.
Where an operator takes this next.
The read works from a pasted export. Here is where the leak gets caught while it's still small.
Once the definitions are pinned, the leak stops being an argument and starts being a fix.
A scheduled Claude task rebuilds this funnel weekly from Salesforce and posts to Slack the moment the Demo-to-Proposal rate drops below its trailing average.
Connect a BI tool so the same pinned definitions run per segment, source, and owner without rebuilding the query each time.
Feed the worst transition into a coaching or deal-review skill so reps get a specific drill on the stage that's actually bleeding deals.
One skill is the on-ramp.
A single skill does one job. Chained into a playbook, or run as a full build, it becomes a system. Here is where this one plugs in.