Ask a model to "fill out MEDDPICC for this deal" and it will. Every field populated, every letter answered, and perhaps a third of it invented. It is trained to complete the form, and an empty field reads as a question rather than a finding.
That is the whole problem with AI qualification, and it has a one-line fix. These prompts go letter by letter around a single rule: quote the evidence or mark the gap. They work in ChatGPT, Claude, Gemini or Copilot. For the framework itself, see what MEDDPICC is or how to score a deal against it.
The quote-or-gap rule
Paste this above any qualification prompt you run. It is doing most of the work on this page.
Rules for this task. For every element, you must either quote the exact words from the source that support your answer, or write GAP and nothing else. Do not infer, do not generalise from similar companies, and do not fill a field because it looks incomplete. A GAP is a useful finding, not a failure. If the source contradicts itself, quote both lines and say so.
Run the same prompt with and without those rules once. Without them you get a tidy scorecard; with them you get a list of what you do not know, which is the only output that changes what you do next.
Your source matters as much as your prompt. Use raw call notes or a transcript, not a CRM summary — a summary has already dropped the buyer's exact words, which is what the rule depends on. Getting a call into usable shape is covered in prompts for sales call notes.
Metrics and economic buyer
These two carry the most invention risk: a model can generate a plausible ROI figure and job title with no evidence at all.
From these call notes, extract Metrics: the quantified business outcome this buyer wants. Quote their words. Separate what they measure today from what they want it to become, and mark which figures came from them and which I supplied. If no number was discussed, write GAP and draft the one question that would surface it.
Identify the Economic Buyer: the person who can approve this spend without asking anyone else. Quote the evidence that they hold that authority. Distinguish who has been named from who I have actually met. If nobody has confirmed budget authority, write GAP.
That named-versus-met distinction is worth the prompt on its own: a deal where the economic buyer has been named but never met is a common shape of loss, and it looks fine on a scorecard.
Decision criteria, decision process and paper process
The three middle letters are where deals slip a quarter, and where reps guess most.
Extract Decision Criteria: what they said they will judge vendors on, in their words, ranked if they ranked them. Mark any criterion that came from me rather than from them. Then list criteria a buyer like this usually has that were never mentioned, clearly labelled as untested.
Extract the Decision Process and the Paper Process as two separate timelines: who evaluates, who approves, in what order, by when — and then, separately, what happens after they say yes. Security review, legal, procurement, signature authority. For each step, quote what they told me or write GAP.
Keeping paper process separate is not pedantry. Deals with a solid decision process and an unmapped paper process are the ones that commit for a quarter and close in the next. Prompts for working that stretch are in Negotiation, Procurement & Closing.
Identify pain and champion
Identify Pain: what is broken, what it costs them, and what happens if they do nothing. Quote their words. Flag pain I described that they never confirmed in their own language.
That flag catches the most comfortable illusion in a deal review: our own pain repeated back to us.
For Champion, resist the prompt that asks "who is our champion" — the model will nominate whoever spoke most. Test instead:
Assess the Champion claim. For [name], list evidence they have taken action on our behalf when we were not in the room: introductions made, internal meetings arranged, material shared onward, language they used with colleagues. Quote each. If all evidence is enthusiasm expressed to me directly, say that plainly: this is a supporter, not a champion.
Champion and multi-threading prompts sit in Deal Strategy & Stakeholder Management.
Competition
Extract Competition, including the status quo and internal build as options. For each, quote what they said about it, note which of our differentiators they have actually heard, and list the criteria where we are weaker. If they named no alternative, write GAP and tell me what that silence most likely means.
A model told to find competitors will invent a shortlist from the category. Naming the status quo as a competitor is what makes the answer honest, since that is what most deals lose to.
Running a full deal inspection
Once each letter is answered individually, the useful move is not a score. It is a ranked list of what to do this week.
Here are my eight MEDDPICC findings, with GAPs marked. Do three things. One: name the two gaps most likely to cost us this deal, and say why those two. Two: for each, write the single question that would close it and the person to ask. Three: argue the case that this deal will not close this quarter, using only my own evidence. Do not reassure me.
The third instruction is the one worth keeping. A model assessing a deal tends toward optimism, because the notes were written by someone who wanted it. Asking for the opposing case surfaces what a manager will ask in the forecast review — often before the manager does. The manager's side of that conversation is in AI prompts for sales managers, and the wider AE workflow in AI prompts for account executives.
Run it across a pipeline rather than one deal and it becomes a self-inspection on the review cadence — see Sales Operations, CRM & Productivity.
MEDDIC vs MEDDPICC, and custom GPTs
MEDDIC is the original six: Metrics, Economic buyer, Decision criteria, Decision process, Identify pain, Champion. MEDDPICC adds Paper process and Competition. For a short transactional cycle the extra two are overhead; for an enterprise deal with security review and procurement, paper process frequently decides the close date. If your team runs MEDDIC, delete the two prompts and keep the rule.
There are several MEDDPICC custom GPTs, and they are fine — as long as they follow the quote-or-gap rule. Before trusting one, run it on a call where you know a letter is genuinely unknown. If it returns a confident answer instead of a gap, it is completing a form rather than qualifying a deal, and the letters it invents are exactly the ones you needed.
Where these prompts live
Qualification prompts only pay off when the same version runs on every deal; one rewritten from memory each time produces findings you cannot compare across a pipeline. Promptifi keeps them organised by the job. Browse the library →, see the full taxonomy, or take one free prompt at /free-prompt. Pro is $12 per month or $89 per year, and lifetime access is $149 — See pricing. Discovery question design, where most gaps get closed, is in Meeting Prep & Discovery and prompts for discovery calls.
Frequently asked questions
Can ChatGPT do MEDDPICC qualification?
It can extract and organise MEDDPICC evidence from call notes reliably. It cannot judge a deal, and left unconstrained it invents answers for letters your notes do not cover. The fix: require a direct quote for every element or the word GAP, and run one letter at a time.
What is a good MEDDPICC prompt?
One that names a single letter, points at a specific source such as a transcript, and forbids inference. The rule that makes it work: quote the buyer's exact words, or write GAP and nothing else. Prompts that ask a model to "complete MEDDPICC" produce a full scorecard and a false picture.
How do I find MEDDPICC gaps in a deal?
Answer each letter separately with the quote-or-gap rule, then give the model your findings and ask for the two gaps most likely to cost the deal, the question that would close each, and the case against the deal closing this quarter using only your own evidence.
Is there a MEDDPICC GPT?
Several exist and some are good. Test one before trusting it: run it on a call where you know a letter is unknown. If it returns a confident answer rather than a gap, it is filling in a form. One that enforces quote-or-gap is useful; one that does not is worse than no tool.
What is the difference between MEDDIC and MEDDPICC?
MEDDPICC adds Paper process and Competition to MEDDIC's six. Paper process covers everything after the yes — security review, legal, procurement, signature authority — and often sets the real close date. Short transactional cycles rarely need either; enterprise deals usually do.