River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

VC Deal Pipeline Tracking Template

Two documents and three sheets that date every deal by its last evidenced contact, so a dormant deal stops counting toward the forecast.

Free download  ·  No account needed

Pipeline Definition Note

[Fund] — Pipeline Stages, Defined In Evidence

Each stage is defined by a document or an event that exists, not by a level of enthusiasm. A stage nobody can falsify is a stage that drifts.

1. The stages, defined in evidence

StageEnters whenEvidence on fileLeaves when
Sourced
Screened
Met
Diligence
Committee

3. Two conversion rates from one history

The same pipeline produces one rate counting every row and another counting only the rows whose stage has its evidence on file. Both get stated, with the count behind each.

TransitionAll rowsEvidenced rows onlyGap
Sourced to screened
Screened to met
Met to diligence
Diligence to committee

The difference between the two columns is the size of the hygiene problem, not a fact about the market. A dashboard showing one rate is showing the flattering one by accident.

7. Source attribution, and the mix that matters

Where each row came from, counted by what progressed rather than by what arrived.

A seed fund's CRM holds 198 open deals. The dashboard multiplies the stage counts by the firm's own conversion rates and reports 31.1 new investments coming. Joined to the mail and calendar record, 44 of those deals were touched inside their stage's own dormancy window and 154 were not. 16 appear nowhere in the record at all. Those 198 deals will produce 6.3 investments rather than 31.1, and the fund closes 11.3 a year.

Every venture pipeline template ships the same board: Sourced, Screening, First Meeting, Diligence, IC, Term Sheet. The better ones sync your inbox so you type less, then compute conversion off the stage field anyway. Two errors follow and they point the same way. Stalled deals are never marked lost, so the counts inflate. Conversion is advanced over advanced plus lost, so those same deals leave the denominator. The same collapse runs through screening inbound, where an absent figure is not a failed one.

Smaller is not the same as right, so the corrected number gets checked twice. Working 11.3 closes a year back up the funnel implies 403 screened deals annually, and the fund screened 403. Applying the dwell times predicts 45 live deals, and the join found 44. Against a plan of 12, 201 companies still have to be sourced. For a registered adviser this correspondence is already retained under the books and records rule, and a written standard reviewed annually is the shape the compliance rule expects.

198 open deals, 44 of them evidenced, and a dashboard forecasting 4.9 times what the register will deliver

The register slice, the exception counts, the yield decomposition with its dwell-time check, and the source mix among live deals.

Pipeline Register

Illustrative. Eleven of the 198 open deals held by Hallowfield Ventures, a fictional seed fund, on 30 April 2026. Every row is dated by its last evidenced contact in the mail and calendar record, not by the CRM. The threshold column is twice the median dwell of deals that actually advanced out of that stage.

DealCompanyStageOwnerStage enteredLast evidenced contactDaysThresholdStateExceptions
H-1197Bramwell SystemsTerm SheetJ. Padstow2025-07-13none on recordn/a52dormantE1 E2 E3 E5
H-1195Vellacott SystemsTerm SheetT. Ellinby2026-03-122026-04-082252livenone
H-1190Marchgate AnalyticsICA. Rewsley2025-10-122025-11-0118028dormantE1 E3
H-1171Fernwick HealthDiligenceT. Ellinby2025-05-142025-05-2234368dormantE1
H-1172Salterton GridDiligenceJ. Padstow2026-01-242026-01-299168dormantE1
H-1165Vellacott GridDiligenceA. Rewsley2026-04-162026-04-29168liveE3
H-1130Ockbrook ComputeFirst MeetingM. Cawthorne2025-06-10none on recordn/a42dormantE1 E2 E3
H-1124Alderwick GridFirst MeetingJ. Padstow2026-03-022026-04-131742liveE3 E4 E5
H-1119Fernwick SoftwareFirst MeetingT. Ellinby2026-03-082026-04-072342livenone
H-1046Waverleigh NetworksScreenedJ. Padstow2026-02-17none on recordn/a22dormantE1 E2
H-1025Dunwarren MaterialsScreenedJ. Padstow2025-05-142025-06-1132322dormantE1
H-1002Barrowden NetworksScreenedT. Ellinby2026-04-162026-04-28222liveE5

E1 dormant. E2 no evidence at all. E3 stage entered with no evidencing event. E4 owner silent while another partner is active. E5 recorded source disagrees with the first touch.

Data Quality Report

All 198 open deals. Five exception types, each one a specific disagreement between the CRM and the communication record rather than a judgement about a deal.

CodeExceptionDealsShareAt First Meeting or beyondWhat it costs
E1Dormant beyond the stage threshold15478%60Inflates the stage count, so it inflates the forecast
E2No evidenced contact of any kind168%9A list import nobody worked. Was never a deal
E3Stage entered with no evidencing event3216%32Inflates conversion at the stage below it, permanently
E4Recorded owner silent 90 days, another partner active158%5Every chaser goes to somebody who cannot answer
E5Recorded source disagrees with the first touch3317%15Misdirects the next year of sourcing
At least one exception17287%69The union, not the sum. Deals carry more than one
Clean2613%11No disagreement between the CRM and the record

Deals with no evidence at all, or past three times their threshold, are closed out by default: 91 of them. The band between one and three times gets one chase, addressed to the corresponding partner rather than the owner field: 63 of them. That takes the register from 198 rows to 107.

Forecast Reconciliation

How many of the deals already in the register become investments. Left block is what the dashboard computes: register counts times conversion measured as advanced over advanced plus lost. Right block is live counts times conversion measured over everything that entered the stage.

StageRegister nCum. resolved-onlyExpectedLive nCum. entered-cohortExpected
Screened1184.72%5.57242.81%0.68
First Meeting4617.38%8.001111.49%1.26
Diligence2441.12%9.87632.69%1.96
IC668.00%4.08165.39%0.65
Term Sheet489.47%3.58289.47%1.79
Total19831.1446.3
The decomposition
Register counts, resolved-only conversion31.1
Correct the counts to live deals only8.0
Correct the conversion basis as well6.3
Partnership plan, next four quarters12
To be sourced from companies not yet screened5.7, or 201 newly screened companies

Smaller is not the same as right, so the corrected count is checked against the dwell times, which share none of its inputs. A stage in steady state holds roughly its annual entrants times its dormancy window over a year.

StageImplied entrants / yrWindow (days)Live deals predictedLive deals found
Screened4032224.324
First Meeting994211.411
Diligence35686.56
IC17281.31
Term Sheet13521.82
Total45.344

Second check: working the demonstrated pace of 11.3 closes a year back up the funnel implies 403 screened deals annually, and the fund screened 403.

Source Attribution

The recorded source field against the earliest evidenced touch, then the mix among live deals only. The whole-register mix describes sourcing as it was, because the register is mostly years of dormant rows.

SourceRecordedShareEvidencedShareDiffLiveLive share
Co-investor referral4522.7%4321.7%-21329.5%
Cold inbound3718.7%3819.2%+11022.7%
Accelerator3115.7%3316.7%+2715.9%
Conference3115.7%3819.2%+7511.4%
Portfolio referral2814.1%2512.6%-3613.6%
Partner network2613.1%2110.6%-536.8%

Co-investor referral is 21.7 per cent of the register and 29.5 per cent of what is live. Partner network is 10.6 per cent of the register and 6.8 per cent of what is live. Read the first column and the two channels look comparable.

What is in the pack

01

Pipeline Register

One row per open deal, dated by its last evidenced contact, with the stage threshold, live or dormant, the partner actually corresponding, and both the recorded and the evidenced source.

02

Data Quality Report

Five exception types counted, each with its share of the register, its count at First Meeting or beyond, what it costs and the action. Plus the union and the complement, so the number reads as a register problem rather than an indictment.

03

Source Attribution

The recorded mix against the mix the first evidenced touch supports, then the mix among live deals only, which is the one a sourcing decision should use. What that sourcing is drawing from is counted in the market map.

04

Pipeline Definition Note

Each stage defined by the artifact that has to exist, both conversion rates from one history, the yield decomposition, two independent checks, and the sourcing requirement that follows.

05

Data Entry Standard

Three mandatory fields and no more, the weekly cycle, and the default-closed rule that clears the sediment. Last-touch date is deliberately not a field anybody maintains by hand.

06

A stage is a claim about evidence

The space rule. Each stage names an artifact, the artifact has to exist, and absence of evidence raises a question for a named partner rather than settling one. The same one-way reading governs a portfolio company that has gone quiet.

How it works

  1. 1

    Join the register to the record

    Send the CRM export and the mail and calendar record for the whole team, not only the recorded owners. Every open deal gets dated by its last evidenced contact, matched on the company.

  2. 2

    Derive the thresholds from your own history

    Twice the median evidenced dwell of deals that advanced out of each stage. In the worked example that runs from 22 days at Screened to 68 days in Diligence, so one firm-wide rule fails.

  3. 3

    Compute conversion both ways

    Once on resolved deals only, once on everything that entered each stage. In the worked example 94 of 604 screened deals were never resolved at all, and that wedge compounds across five stages.

  4. 4

    Turn the corrected number into a sourcing figure

    Run both checks, present the yield as a decomposition rather than one smaller figure, then divide the gap against plan by your screened-to-closed rate. The answer is a count of companies, not a verdict.

Frequently asked questions

Why not just add a last-contacted field to the CRM?

Because it is the field most likely to be wrong when a person maintains it. Deriving it from the mail and calendar record every time the register rebuilds costs nobody any typing and cannot go stale. In the worked example 15 deals had an owner who had gone silent while a different partner was actively corresponding.

Is 78 per cent dormant not just a badly run fund?

It is arithmetic, not negligence. Over the historical window 94 of 604 screened deals were never explicitly resolved, 15.6 per cent. Carry that rate across a fund's life and it predicts about 178 stale rows. 154 were found. Sediment accumulates from inaction, which is why only a default that acts on inaction clears it.

My CRM already syncs my inbox. Is this not the same thing?

Sync saves you typing. It does not use the record to invalidate a stage, and it does not change the denominator conversion is computed on. A synced inbox and a stage field reading Diligence still produce 31.1 expected investments from a register that will deliver 6.3. Capture is not verification, the same way a deck's stated figure is not an established one.

What if a deal is live but the conversation happened on a call?

Then the mail record misses it, which is why the rule cuts one way only. Evidence promotes a deal to live, and absence of evidence raises a question rather than settling one. Every dormant deal goes on a list a named partner answers, and a partner who says a deal is live re-dates it with a note.

Does this tell me which sources produce my best deals?

No. It tells you whether the source field is accurate and which channels are currently live, which are different questions. Ranking by quality needs realized outcomes, which is what a dedicated source-outcome register computes. A category supplying a fifth of the register and none of the closes is a finding that live-versus-recorded counts alone cannot reach.

Does the corrected number replace my annual forecast?

No, and reading it that way repeats the original error in the other direction. It is the expected yield of the deals already in the register, which resolve inside four quarters at these dwell times. Most of next year's investments are in companies nobody has seen yet, which is what the sourcing figure is for.

Does it replace the CRM, or write back to it?

Neither. The CRM stays the system of record and nothing here writes to it, because a system that silently corrects the record destroys the only signal that the entry standard is being ignored. Deals that clear the register still feed the investment committee memo.

Find out how many of your open deals are still deals

Send the CRM export and the mail and calendar record behind it. The first pass dates every open deal, marks the dormant ones, and says what the register will actually produce.

Reconcile my pipeline