Pipeline is thin, so somebody exports the closed-lost list and tells you to work it. Two hundred rows, lost six to eighteen months ago. Some are winnable now. Most are not, and the campaign fails because it treats all two hundred the same way.
The work is triage before outreach: which lost deals have actually changed, which to leave alone, and how to say so to a manager who wants the list worked. For the deal that stalled mid-cycle rather than closing, when enterprise deals go dark owns that; for the post-mortem on why you lost, win-loss analysis owns that.
A deal you lost is not a lead, and treating it as one is why the campaign fails
A lead has no history with you. A closed-lost deal has a verdict: somebody evaluated you, compared you to something else and said no. They remember it even when your CRM does not record who.
So the first prompt is not an email: "Here is a closed-lost record with its notes. Summarise what this buyer concluded about us, in their words where the notes have them. Then say what would have to be different now for that conclusion to change." If nothing has to be different, you are emailing someone who already knows the answer.
Prompts that find the few lost deals where something has actually changed
Change is the whole qualifier. Four kinds matter: the person who said no has moved on, their chosen vendor has failed or been acquired, their business has shifted into the problem you solve, or you shipped what they said was missing.
Prompt the sort, not the list: "Here are 40 closed-lost records with loss reasons and dates. Group them by which of these four changes would make them winnable: decision-maker gone, incumbent failing, business shifted, our gap closed. Put anything fitting none in a fifth group and say why." The fifth group is the point. Trigger events is how you detect the first three without guessing.
Reading your own loss reasons honestly, including the ones that were excuses
CRM loss reasons are written by the person who lost, the week they lost, for an audience of their manager. "Lost on price" often means "never built the business case." "No budget" often means "no champion." A re-approach built on a fiction fails for the original reason.
Try: "Here are our loss reasons by frequency. For each, give the most likely real reason it conceals and the one piece of evidence that would tell the two apart." Then: "Which of these would I expect to be over-reported by a rep writing for their manager?" The answer is usually price, and it changes who you re-approach.
Prompts for the re-approach that does not pretend the loss never happened
The failure mode is an email written as though you have never met. The buyer remembers, and the pretence reads as incompetence or bad faith.
So name it: "Write a short email to a buyer who chose a competitor eighteen months ago. Acknowledge the decision in one clause without relitigating it, name the specific thing that has changed since — [change] — ask one question about whether it matters now, and give them an easy no. Under 90 words." That last constraint does the most work. An easy no is what makes the yes credible.
When the person who said no has left, and when that is the whole opportunity
A departed decision-maker is the strongest signal on the list, because the verdict left with them. It is also the one most often worked badly: the successor inherits a vendor they did not choose and has no loyalty to it.
"The person who rejected us has left. Write an approach to their successor that does not mention the previous evaluation at all, and instead asks how they currently handle [problem]." Deliberately different from the prompt above: there is no shared history with this person, so inventing one is strange. Treat it as a first conversation with unusually good context, and qualify it like one using lead qualification prompts.
Prompts for what to do with the deals you lost on price
Price losses are the biggest group and the most tempting to re-approach with a discount. That teaches the buyer your first number was not real. Decide what actually happened first.
"This deal was lost on price. From the notes, decide which happened: they could not fund it, they valued it below our price, or they used price to decline for another reason. Say which, and what evidence supports it." Only the first is fixed commercially. The second needs a different business case; the third needs the real objection found before you quote again.
The lost deals to leave alone, and saying so to a manager who wants the list worked
Some rows should never be touched again: you lost on a capability you still lack, the relationship ended badly, they signed a multi-year contract recently, or they have twice told you to stop. Working them costs hours and does lasting damage.
Make the case in writing: "Here are the records I am excluding and why. Write a short note to my manager stating how many of the 200 I am working, the criterion I used, and what I expect from that subset — without arguing the rest are impossible, only that they are worth less than my next-best use of the time." Managers accept a criterion far more readily than a refusal. Losing better covers the discipline that makes next quarter's list shorter.
Why a model handed a closed-lost list will tell you to re-engage all of them
This is the honest limit, and the reason to build the triage prompt before the email prompt. Ask a model whether a lost deal is worth re-opening and it will find a reason, because a plausible reason is always available. Two hundred rows in, you have two hundred reasons and no priority order.
The fix is to make the model choose and to force the negative case. Ask for a ranked subset with a stated cutoff rather than an assessment per record, and ask the question in reverse: "Which of these should I not touch, and what in the record tells you so?" A model that cannot produce a no is not helping you triage. If a deal comes back to life, the breakup email is the adjacent move, and the prompts for the rest sit in Prospecting & Pipeline Creation, Deal Strategy & Stakeholder Management, Account Research & Buyer Intelligence and Outreach & Messaging.
Edit every prompt above before running it; keep the bracketed parts specific. Start with a free prompt, then Browse the library →, the full taxonomy, or the wider set in AI prompts for sales. Pro is $12 per month or $89 per year, and lifetime access is $149. See pricing.
Frequently asked questions
What are good ChatGPT prompts for re-engaging closed-lost deals?
Triage prompts first, outreach second. Ask the model to group records by which change would make them winnable (decision-maker gone, incumbent failing, business shifted, your gap closed), with a fifth group for records fitting none. Then ask which you should not touch and what in the record says so. Write the email only for the subset that survives.
When should I go back to a deal we lost?
When something specific has changed, not when the quarter is thin. The four changes worth acting on: the person who said no has left, their vendor is failing or was acquired, their business has moved into the problem you solve, or you shipped what they said was missing. Absent one of those, the re-approach fails for the original reason.
How do I re-approach a prospect who chose a competitor?
Acknowledge the decision in one clause without relitigating it, name the one thing that has changed since, ask a single question about whether it matters now, and give them an easy no. Keep it under 90 words. If the decision-maker has left, treat their successor as a first conversation instead: the verdict left with them.
Should I mention that we lost the deal last time?
If you are writing to the person who decided, yes: briefly, once, without arguing, because they remember and pretending otherwise reads as bad faith. If you are writing to a successor who inherited the vendor, no: there is no shared history with them, so raising it invents one and puts them on the defensive about a choice they did not make.
Which closed-lost deals are not worth re-opening?
Ones lost on a capability you still lack, ones where the relationship ended badly, ones that signed a multi-year contract recently, and anyone who has asked you twice to stop. Exclude them explicitly and give your manager the criterion rather than refusing the list: a stated cutoff with an expected outcome lands better than an argument about hopeless rows.