Pipeline Review and Forecast

A deal is only as real as the last thing the customer actually did. You review the pipeline on that principle: what buyers have said and done, not how sellers feel. The forecast is a range with reasons, built from the owner's own conversion history and an inspection of the deals that decide the number. The standard: every committed deal has buyer evidence, every change since last week is explained, and forecast accuracy is tracked so the team learns.

When to use

• "Are we going to hit the number this quarter?"
• "Prep the pipeline review for Monday."
• "Write my forecast for the board."
• "Which deals are at risk of slipping?"
• "Do we have enough pipeline for next quarter?"

What you need

• Open and closed deals with: amount, stage, close date, created date, owner, source, stage history and close date changes if available, last activity date, next step. From the connected CRM or an export. If data quality is poor, run a quick crm-hygiene pass first and say how it limits the forecast.
• The target for the period and what has already closed.
• At least 4 quarters (or 12 months) of closed deals to compute conversion rates and cycle times. With less, say the rates are rough.
• The owner's forecast categories and definitions, if they have them. Otherwise propose the defaults below and record them in memory.

Method

1. Ground the numbers. Confirm stage names, what counts as closed, the period dates and the currency. Read actual field values; do not assume a CRM's default setup.
2. Compute history with execute_code:
   • win rate by stage: of deals that reached a stage, the share that closed won (by count and by value), from your own history rather than generic stage percentages;
   • median sales cycle, overall and by deal size band;
   • median days in each stage;
   • average won deal size.
3. Coverage.
   • Coverage = open pipeline with close dates in the period / (target minus closed so far).
   • Required coverage = 1 / historical win rate for pipeline at a similar point in the period. If 25 percent usually closes from this point, you need about 4 times the gap.
   • Report actual against required.
4. Pipeline creation for future periods. Deals take a median cycle to close, so next quarter's revenue needs pipeline created now. Needed new pipeline = next period target / win rate. Compare with pipeline created in the last few weeks.
5. Velocity (a health trend, not a forecast): (open qualified deals x win rate x average deal size) / median cycle length in days = revenue per day. Track the trend week to week.
6. Changes since last review. List deals added, won, lost, moved stage, changed amount, and close dates pushed out of the period. A pushed deal needs a reason from the buyer, not from the seller.
7. Inspect the deals that matter: every deal in commit and best case, plus any deal above a size threshold the owner sets. For each, check:
   • the last customer action, dated (a reply, a meeting attended, a document returned); seller emails do not count;
   • a next step with a customer-side owner and date;
   • economic buyer known and engaged, champion tested, paper process mapped (see discovery-qualification);
   • a close date backed by a compelling event. Red flags:
   • no customer action in 14 days;
   • close date pushed twice or more;
   • single-threaded;
   • no economic buyer contact in a late stage;
   • paper process unknown with less than 30 days left;
   • stage age over twice the median.
8. Categorise by evidence. Default definitions to propose:
   • Commit: the economic buyer has agreed to buy within the period, terms are agreed or nearly so, and the paper process is mapped with time to complete. You would be surprised to lose it.
   • Best case: a real path to closing this period with an engaged buyer and a mapped decision process, but something material is still open (pricing, legal, approval).
   • Pipeline: qualified and active, but closing this period would need things to go faster than usual.
   • Omitted: not closing this period; move the close date or close the deal. Moving a deal to a lower category when reality changes is good forecasting, not failure.
9. Build the forecast range three ways and compare:
   • Category roll-up: closed + commit + (best case x historical best-case conversion).
   • Weighted pipeline: closed + sum of (amount x historical win rate for its stage and age).
   • Run rate for small, high-volume deals: average closed per week x weeks left. Report a low (closed + commit), a most likely and a high (adding best case), and explain where the methods disagree. A worked example is in references/forecast-template.md.
10. Risks and asks. The top 3 risks to the number with what would reduce them, and specific asks of the owner (an executive call, a pricing decision, help with a stuck legal review).
11. Track accuracy. Record each week's most likely forecast. At period end, error = (forecast minus actual) / actual for each week. Report the pattern (for example, consistently 15 percent optimistic until week 8) and adjust the conversion assumptions.
12. Present.
    • Put the number, range, coverage and top changes on a show_card, with the full review as a doc.
    • Schedule the weekly run with cronjob.
    • Suggested category changes go to the deal owners and the owner; change the CRM only with approval.

Weekly review agenda (30 to 45 minutes)

1. The number: closed, range, coverage against required (5 minutes).
2. What changed since last week and why, buyer reasons only (10 minutes).
3. Commit and best case deals with red flags: one question each, "what did the customer do last, and what will they do next?" (15 minutes).
4. Pipeline creation for next period against need (5 minutes).
5. Asks and owners (5 minutes). Send the pack the day before so the meeting is for decisions, not reading.

Output

Use references/forecast-template.md:

• Headline: target, closed, forecast range (low, most likely, high), coverage against required.
• Changes since last week.
• Deal table for commit and best case: amount, stage, close date, last customer action and date, next step, red flags, category and suggested change.
• Pipeline creation versus need for the next period.
• Risks and asks.
• Accuracy history.

Checks before you finish

• Every commit deal shows a dated customer action and a mapped paper process.
• Conversion rates come from the owner's own history, with the date range stated.
• The three methods were compared and differences explained.
• Pushed deals have a buyer reason recorded.
• Figures recompute from the data; currency and period are stated.

Pitfalls

• Seller optimism as data. "Feels good" is not evidence. Ask what the customer did.
• Generic stage probabilities. Default CRM percentages rarely match a real business. Use history.
• Coverage without quality. Three times coverage of stale deals is not three times coverage.
• Sandbagging. Hiding deals to look good later destroys trust as much as optimism does.
• Ignoring next quarter. If pipeline creation is short now, next quarter is already in trouble. Say so early.
• Dead deals left open. They inflate coverage and hide the real gap.

See also: crm-hygiene, discovery-qualification, win-loss-analysis.