Pipeline review is Thursday. You have eleven open opportunities, a number you said out loud three weeks ago, and ninety minutes before the meeting to work out which deals will not survive the questions.
Almost every page ranking for prompts on this subject is a programmatic directory: a template, a dozen interchangeable prompts, written for a revenue operations audience rather than for the person who has to defend a commit. This page is about the meeting. Forecasting as a discipline is covered elsewhere — see pipeline review prep for the routine around it and how to run a pipeline review for the management side.
Before the meeting: inspect your own pipeline honestly
Export your open opportunities with their stage, value, close date, last activity date and your notes. Then run this before you look at anything else.
Here are my open opportunities with notes. For each one, state the single piece of evidence that it will close in the stated month. If the only evidence is my own optimism, say so. Do not fill gaps.
What comes back is usually shorter than you expect, and the deals with no sentence under them are your meeting. Everything else on this page is detail.
The stage-integrity prompt
Stage inflation is the most common reason a pipeline looks healthy and is not. A deal sits in a late stage because it moved there in March, not because anything happened.
Here are our stage definitions and my open deals with notes. For each deal, say which stage the evidence actually supports and quote the note that supports it. Where a deal is in a later stage than the evidence supports, say which exit criterion has not been met.
For every deal, list the exit criteria for its current stage that I have no recorded evidence for. Return it as a checklist I can work through.
That second prompt turns a vague worry into a to-do list. Qualification discipline for the same deals, element by element, is in ChatGPT prompts for MEDDPICC.
Commit and best case, with evidence
The split between commit and best case is where credibility is won or lost, because a manager remembers which bucket you used last quarter.
Sort these deals into commit, best case and pipeline. For each, give the two facts that justify the bucket and the one fact that would move it down a bucket. A deal with no confirmed next meeting and no named approver cannot be commit — apply that rule strictly.
Here is what I plan to call for the quarter. Build the case that I will miss it, using only the evidence in these notes. Then tell me which two deals that case depends on.
The second prompt is the one the directory pages never include, and it is the reason to run any of this. A model asked to assess your deals will inherit the optimism of whoever wrote the notes, so you have to ask it to argue the other side explicitly.
Deals that have quietly stopped moving
A stalled deal rarely announces itself. It just stops appearing in your week.
Rank these deals by how long since a buyer-side action, not a seller-side action. Count emails I sent with no reply as no activity. Flag anything over 14 days.
This deal has had no buyer response for [n] days. From the notes, list the three most likely reasons, the evidence for each, and the lowest-risk message to test the most likely one.
The distinction in the first prompt matters more than it looks. Seller activity is not pipeline movement, and a CRM that counts your own follow-ups as engagement will tell you a dead deal is active.
The questions your manager will ask, before they ask
You are my sales manager reviewing this pipeline. Ask me the eight hardest questions, starting with the deal where my case is weakest. One at a time. Do not accept an answer that contains no evidence.
Run it, answer out loud, and note which answers you could not give. That list is the only prep that reliably changes how a review goes. Prompts for the stakeholder gaps it exposes are in Deal Strategy & Stakeholder Management.
For managers: running the review itself
Inspecting one rep's pipeline is a different job from inspecting a team's, and the useful prompts are comparative.
Here is the pipeline for my team of [n] reps. For each rep, identify the deal most likely to slip and the one pattern in their notes that repeats across deals. Then name the single coaching conversation with the highest return across the whole team.
Compare stage distribution across these reps. Where does a rep's pipeline differ from the team's in a way that suggests stage inflation rather than a different territory?
From last quarter's reviews and what closed, identify which of my team's commit calls were reliable and which were not. Quote the evidence that was present at the time.
That last one is the only honest way to calibrate a team, and it is worth running once a quarter rather than weekly. The surrounding manager workflow is in AI prompts for sales managers, with the reporting layer in RevOps, Data & Analytics Support and the coaching layer in Leadership, Coaching & People Management.
Turning the review into next week's actions
Most reviews produce agreement and no change, because nothing written down survives the meeting.
Here are my notes from today's pipeline review. Extract every commitment I made, who it is to, and the date. Separate anything that was discussed but not committed to. Then list the three actions that affect the deals most likely to slip.
Pair it with the CRM hygiene prompts in Sales Operations, CRM & Productivity so the updates land where the next review will read them.
Take one prompt free, with no account, at /free-prompt. Pro is $12 per month or $89 per year, and lifetime access is $149 — See pricing.
Why a forecast built on a model's optimism is worse than no forecast
An assistant reading your notes has no access to the buyer. It has access to your account of the buyer, written by you, on days when you felt good about the deal. Ask it whether the quarter looks achievable and it will reflect that account back at you with more confidence than you had.
That is worse than guessing, because it feels like analysis. Three habits keep it useful: constrain every prompt to recorded evidence, require a quote rather than a judgement, and always run the prompt that argues against your own number. A model that tells you a deal is strong has told you nothing. A model that names the two deals your quarter depends on has told you where to spend Friday.
The rest of the catalogue, by role and by stage, is mapped in AI prompts for sales, and the account-executive view of the same deals is in AI prompts for account executives. The same evidence discipline applied to a book of renewals rather than a pipeline is in ChatGPT prompts for renewal conversations. Browse the library → or read the full taxonomy.
Frequently asked questions
What are good ChatGPT prompts for pipeline review?
The four that earn their place are a stage-integrity check against your own exit criteria, a commit-versus-best-case sort that requires two facts per deal, a stalled-deal check that counts only buyer-side activity, and a prompt that argues the case you will miss your number. The last one changes more reviews than the other three together.
How do I use AI to forecast sales more accurately?
By using it to test evidence rather than to predict. Have it state what evidence exists for each deal's close date, mark what is assumed, and build the case against your own call. Accuracy comes from removing deals that have nothing under them, not from asking a model for a number.
Can ChatGPT tell me which deals will close?
No. It has no access to your buyer, only to your notes about your buyer, which were written on your good days. What it can do reliably is point out which deals have no recorded evidence for their next step or their timeline, which is the same information in a more useful form.
How do I prepare for a pipeline review meeting?
Export your open deals with stage, value, close date, last buyer activity and your notes. Run the evidence prompt, then the stage-integrity prompt, then have the model play your manager and ask the eight hardest questions one at a time. The answers you cannot give are your prep list.
What should a manager inspect in a pipeline review?
Evidence rather than confidence. Whether each deal's stage matches its exit criteria, whether commit deals have a confirmed next meeting and a named approver, how long since buyer-side activity, and whether a rep's stage distribution differs from the team's in a way that suggests inflation. Calibrating last quarter's commit calls against what actually closed is the highest-value pass.