Diagnose a stalled deal and get a ranked 14-day action plan

What this does

You summarize a stuck deal. The AI lists possible reasons it stalled, what evidence in your notes supports or contradicts each, the questions you still need answered, and a ranked 14-day action list.

Who it’s for

AEs with a deal that went quiet or stuck in a stage; managers using it to prep a deal review with a rep. Not a forecast, and it can’t know what the buyer is thinking.

What you need

• Deal facts: stage, size band, age, last contact
• Who’s involved and their roles
• What’s happened (timeline in your words)
• What the buyer said they wanted
• Tool: Any AI chat tool (for example ChatGPT, Claude, Gemini, or Grok), Your CRM or notes
• Cost: $0 with a free AI chat tier (limits change)
• Setup: 5 minutes (estimate)

Steps

1. Write a short timeline: what happened, when, with whom. Use fake names or titles if you’re using a consumer AI tool.
2. Add what the buyer said they wanted and what you’ve sent.
3. Run the prompt.
4. Check each ‘possible reason’ against your notes. Strike any with no evidence.
5. Pick the top two actions, assign dates, and do them by hand. For managers: use the ‘questions to ask the rep’ list in your next one-on-one.

The prompt

You are an experienced sales manager reviewing a stalled deal. Use ONLY the facts I give you. Don't invent buyer motives, competitors, or events. Where you are guessing, label it "Possible."

NO-ADDITIONS RULE: Do not add facts, tips, statistics, scarcity or urgency, promises, references or testimonials that I did not supply. Refer to the buyer and everyone else involved as "they" or by title. Do not use "he," "she," "his," or "her" unless I gave that person's gender. Evidence FOR or AGAINST a reason must come ONLY from my notes below. Do not use outside knowledge, "typical" buyer behavior, or what companies usually do as evidence. If my notes have no evidence either way, write "No evidence in my notes."

Deal summary: [STAGE, ROUGH SIZE, AGE IN DAYS, LAST CONTACT DATE]
People involved (titles only): [TITLES AND ROLES]
Timeline: [WHAT HAPPENED, IN ORDER, WITH DATES]
What the buyer said they wanted: [QUOTES OR PARAPHRASE]
What I've sent or promised: [ITEMS]
What I think is going on: [YOUR THEORY, OR "NO IDEA"]

Output:
1. POSSIBLE REASONS IT STALLED (up to 5). For each: evidence in my notes FOR, evidence AGAINST or missing, and how likely (low/medium/high, labeled as a judgment).
2. QUESTIONS I NEED ANSWERED (things I don't know that matter).
3. RANKED 14-DAY ACTION PLAN (max 6 actions): action, owner (me / buyer / my manager), target day (Day 1-14), and what result would tell me to continue or stop.
4. A "BREAK-UP" TRIGGER: what signal means I should close this deal as lost or park it.
5. QUESTIONS A MANAGER SHOULD ASK THE REP about this deal (3 questions).

Plain words. No pressure tactics. Don't suggest deceptive tactics (fake deadlines, fake discounts, going around someone behind their back).

Sample input (fake data)

All names and companies are fictional. This is the input for run 4 (the output below), on the current prompt. Runs 5 and 6 used different inputs; see the test log.

Deal: Negotiation stage, about $45k, 80 days old, last contact Oct 3
People (titles only): VP of Operations (champion), Procurement Manager (reviewing contract), Owner (signs)
Timeline: Aug 1 intro call. Aug 20 demo with VP and Procurement. Sept 10 sent proposal. Sept 24 Procurement asked for a revised payment schedule; I sent one Sept 26. Oct 3 VP said "I'll get back to you this week." Two emails since (Oct 8, Oct 12), no reply.
Buyer wanted: Faster onboarding of new warehouse hires
Sent: Proposal, revised payment schedule, one customer reference sheet
My theory: Procurement is slowing it down

Sample output

Real output from run 4 on 2026-10-14, using the current prompt, from an unspecified AI assistant (Grok Bot agent session; model name not shown). Copied as produced. Nothing was fixed after the fact; the problems we found in it are listed under “How it went”.

**1. POSSIBLE REASONS IT STALLED**
1. Possible: Procurement is still reviewing the revised payment schedule.
 - For: Procurement asked for it Sept 24 and you sent it Sept 26; nothing in your notes shows a response. Against/missing: No sign Procurement has gone silent on you; the last contact was from the VP. Likelihood: medium (judgment).
2. Possible: The Owner hasn't approved yet.
 - For: The Owner signs, and no note shows the Owner has seen the proposal. Against/missing: Nothing says the Owner is involved yet. Likelihood: medium.
3. Possible: The VP's priorities shifted after Oct 3.
 - For: The VP said "I'll get back to you this week" on Oct 3 and has not replied to two emails. Against/missing: No notes on why. Likelihood: low.
4. Possible: The proposal doesn't fully answer their onboarding goal.
 - For: Their goal was faster onboarding of new warehouse hires; your notes don't say the proposal addressed it. Against/missing: Your notes don't list what the proposal covers. Likelihood: low.
5. Possible: Your theory (Procurement slowing it down).
 - For: Procurement asked for a revision Sept 24. Against/missing: Nothing after Sept 26 shows Procurement is the one holding up. The silence since Oct 3 comes from the VP. Likelihood: low.

**2. QUESTIONS I NEED ANSWERED**
- Did Procurement accept the revised payment schedule?
- What did the VP mean by "get back to you this week"?
- Has the Owner seen the proposal?
- Is there a date by which they want new hires onboarded faster?

**3. RANKED 14-DAY ACTION PLAN**
1. Day 1, me: Email the VP asking whether the revised payment schedule resolved Procurement's question. Continue if they reply; stop if silent by Day 4.
2. Day 3, me: Ask Procurement directly whether the Sept 26 schedule works. Continue if they answer.
3. Day 5, me: Ask the VP who needs to approve and what the steps are before the Owner signs.
4. Day 7, my manager: Offer a short call with the Owner, with the VP's knowledge. Continue if the VP agrees.
5. Day 10, me: Send a one-page summary tied to their onboarding goal for the VP to forward.
6. Day 14, me: Send a closing note asking for a yes, no, or date.

**4. BREAK-UP TRIGGER**
No reply from the VP or Procurement by Day 14 and no new contact by Day 21: park the deal and set a reminder to revisit.

**5. QUESTIONS FOR THE REP**
- Why do you think Procurement is the delay, given the VP is the one who went quiet?
- Who has actually talked to the Owner?
- What did Procurement say about the revised schedule?

Check before you send

• Each ‘possible reason’ has evidence in your notes. Strike the rest.
• Actions are things you’re willing to do honestly (no fake deadlines or discounts).
• The ‘break-up trigger’ fits your sales process.
• Use real names only in your own CRM, not in the AI tool, unless policy allows.

How it went

• Test date, tool, model: 2026-10-14. Unspecified AI assistant (Grok Bot agent session; the model name is not shown to the agent). One tool only.
• Runs logged: 6, on 6 different fake inputs (see the test log). Runs 1-3 used the original prompt. Runs 4-6 are a second round on new fake inputs after the prompt was fixed, as listed below. Every run was done by the same AI assistant session that wrote the fake input, and in round 2 the same session also edited the prompt, so these are not blind or independent runs. A script counted words and characters against the prompt’s limits. No second person re-ran it from the written steps.
• Why the label is “Tried in an AI tool,” not “tested”: we reserve “tested” for plays with at least 3 logged runs in at least 2 different AI tools, and we ran one tool. We also have no independent reproduction of the steps and no timed manual baseline. The runs are real, but this is one tool’s behavior, and 3 runs per round is too few to say a problem is gone.
• Time saved: Not measured. We did not time a manual baseline or an assisted run.
• What we changed after round 1: Added: no-additions rule; ‘they’ for everyone unless gender is supplied; evidence FOR/AGAINST only from my notes, ‘No evidence in my notes’ otherwise, no outside knowledge or ‘typical buyer’ behavior.
• Round 2 (current prompt, runs 4-6). What worked:
  • No gendered pronouns for the buyer in any run (the first round used ‘he’ 8 times for an ‘Owner’). Titles and ‘they’ were used throughout.
  • Evidence FOR/AGAINST came from the notes. ‘No evidence in my notes’ (run 6) or a close paraphrase (run 5) appeared where nothing supported a reason, and none of the ‘finance teams are busy during close’ or ‘small-business pattern’ outside-knowledge lines from the first round appeared.
  • In run 4 the model told the rep their own theory (Procurement) was weakly supported by the timeline, and in run 5 it said so for ‘owner doesn’t see value’. Caps held: up to 5 reasons, up to 6 actions, exactly 3 manager questions.
• Round 2: what failed or needed fixing:
  • Run 6 computed ‘only 4 days since the proposal’ from a current date the input never gave. That is a small outside-the-notes fact, and the prompt now says notes only.
  • Action plans are still fairly generic (‘send a short message asking whether it arrived’), especially for the thin deal in run 6.
  • Run 4 has 5 reasons and 6 actions with ‘for’ and ‘against’ lines; checking all of them against the notes takes more than 2 minutes, so edits is No.
  • The model’s likelihood labels are still judgment calls (for example ‘high’ for one reason in run 5 on one line of notes).
• Checked against “Check before you send”: ‘Each possible reason has evidence in your notes’ passed in runs 4 and 5 and mostly in run 6; titles used; actions honest; break-up triggers fit.
• Edge cases: Thin deal (run 6) is now honest about missing evidence. Only one input without a gender hint was run thoroughly; a deal where the rep’s notes use ‘he’ or ‘she’ was not run.
• Round 1 (original prompt, runs 1-3): what failed, kept for reference:
  • Run 3: gender assumed for the buyer (‘he’ x8, ‘his’ x1) when the input said only ‘Owner’. A small hallucinated fact that would carry into follow-up emails.
  • Evidence ‘FOR’ a reason was sometimes general knowledge, not the notes (run 1: ‘finance teams are busy during close’; run 3: ‘small-business pattern’, ‘spouse or partner or bookkeeper’). The prompt says use ONLY my facts.
  • Run 2 misread the notes: ‘VP Sales has not been in contact since Aug 28 per your notes’. The notes say a proposal went out Aug 28 and nothing about VP contact after that.
  • Action plans were a bit generic (‘send a one-line summary’, ‘offer a short call’), especially in the thin run 3.
• Not yet observed: how a different model or tool handles the same prompt (a second-tool run is still to do).

Watch-outs

Privacy: use titles instead of names, and avoid pasting confidential pricing or contract terms into a consumer AI tool unless your company allows it. Any outreach that follows still needs your review and the usual email/text consent rules.

See the Safe Use page: AI drafts, you decide, you send.

Variations

• Manager version: run it for the top 5 stuck deals before a pipeline review.
• Owner version: replace ‘people involved’ with ‘who in the household or business decides.’
• Short version: ask for only sections 2 and 3.

Sources

None. Original wording. (Add any public framework or post used, with link and checked-on date.)

Retest

Re-check by: 2026-12-13 (60 days after the 2026-10-14 runs), or sooner when run in a second tool.

Test log

Changelog

• 2026-10-14: Draft written. Not run yet.
• 2026-10-14: Ran 3 times in one AI assistant on fake data. Filled in sample input and output, how it went, and the test log. Front matter set to runs_logged 3, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested. The prompt text was not changed after the runs.
• 2026-10-14: Prompt revised after round 1. Added: no-additions rule; ‘they’ for everyone unless gender is supplied; evidence FOR/AGAINST only from my notes, ‘No evidence in my notes’ otherwise, no outside knowledge or ‘typical buyer’ behavior. Re-ran 3 fresh runs (4-6) on new fake inputs in the same AI assistant. Updated the sample input and output to run 4, the “How it went” section, and the test log (original runs 1-3 kept, marked). Front matter set to runs_logged 6, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested.
• 2026-10-15: Published with the evidence label “Tried in an AI tool” (6 runs, 1 tool, fictional inputs). Earlier wording that implied more evidence than one tool’s runs was replaced with the label; the runs, results, and limits above are unchanged.