Sales Call Prep Prompts for Competitive Deals

The battlecard exists. The deal record exists. Somebody has already written the competitor profile. What nobody has is twenty minutes to work out which eight lines of all that matter for the call starting in an hour.

So the work is not writing. It is choosing what to leave out. And a choice is only visible when something actually gets left out.

That is why the word count is the instruction rather than the formatting. Hand a model the same pile with no limit and it gives it back to you in a tidier order with better headings. That reads like work. It is not: a reordered pile still has to be read in full by somebody who has two minutes.

Most AI prompts ask a model to produce something you do not have. This one asks it to throw things away. Which means the interesting question is never what the brief says. It is what the brief dropped to get there, and a brief never tells you that by itself.

Two blocks go in, and the order changes the result. Put the account record first. Whatever a model reads first becomes the subject, and the subject of this brief is a deal, not a competitor.

The account half is usually thinner than it should be. The useful sentences were spoken, not logged. Nobody creates a CRM field for a buyer mentioning that their new head of sales used the competitor at their last company, so it sits in a call recording and only reaches the brief if somebody goes and gets it.

The third pile is where prep goes wrong

When a brief comes back thin, the reflex is to paste in more. Two more battlecards. The market overview. Last quarter’s competitive digest. That adds general material to rank and no new information about this account, so the balance tips further against the half you needed. One card for the competitor actually in the deal beats five cards for the ones who might be.

Four prompts, meant to run in sequence inside one conversation rather than chosen between. The first produces the brief. The second makes its omissions visible. The third picks the one question to ask. The fourth is for the ten minutes before dialling, when there is time to check two facts and no more.

The last prompt ranks claims by how easily the buyer could check them, not by how important they look. That inversion is the point. A published price, a plan name and a free-tier limit are one search away from the person sitting opposite, so a month-old version gets corrected in the room. A claim about how the competitor implements something is slower to check, and safer to be slightly behind on.

The account note for this run was written for it. We say so in the run condition as well as here, because there is no public version of somebody else’s deal notes, and a page in this format that shows you one has invented it. The competitor block is real and every figure in it is checkable. We asked for the brief, then asked the same conversation what it had left out.

Read the first output as a brief a rep would be glad to get. It is clean. The prices are right. The closing question is a good question. A rep could walk into the call with it.

It would also have been the same brief for any six-person sales team evaluating the same competitor, anywhere. Which is the definition of not having prepared.

General facts look more important

A tier price is clean, quotable and obviously competitive intelligence. A note that the buyer’s new head of sales spent three years at the competitor is messy and specific to one account. It does not look like the thing that was asked for. Under a limit, a model keeps whatever resembles the category in the request, so the cut runs straight along the line between what anyone could know and what only this deal knows.

The fix costs nine words and it is the second prompt above. Asked directly which discarded item would most change the call, the same model named the new head of sales without hesitating. It had the ranking all along. Nothing in a request for a two-hundred-word brief ever asked for it.

A landmine question is one you ask so the buyer finds the gap themselves. You do not make the claim about the competitor. They do, out loud, in their own words, which is the only version they will believe.

A battlecard holds four or five of them, and it should. A call brief should carry one. A rep holding five asks none of them well, and the right one is not a property of the competitor. It is a property of this account.

Two more rules keep them usable. The buyer has to be able to answer from their own situation, not from something only you know. And the question must not contain the claim: “How do you cope with their setup times?” is an accusation with a question mark on it. “How long did your last implementation take, end to end?” is a question, and the buyer’s own number is worth more than any assertion you could make.

One honest caution. If you do not already know roughly what the answer will be, it is not a landmine, it is a risk. Run the question against what you know about the account before it goes in the brief, which is what the third prompt above does. A competitive analysis checklist keeps the underlying list maintained so there is something to pick from in the first place.

Prep is the most disposable competitive document a team produces, and that makes it the riskiest place for an old fact. A battlecard gets reviewed, argued over and dated by section. A prep brief is read once, spoken from, and closed. Nothing goes back to it, so a price that moved last month gets said out loud and then deleted, with no record that it was wrong.

The decay is not really in the brief. It is in the card the brief was cut from, which is where the review cadence belongs. A battlecard drafted once goes stale section by section, and every brief cut from it inherits that silently and at full confidence.

One half of this job cannot be watched by anything, and it is worth stating plainly. The account record is private. The sentence about the new head of sales was said on a call. No amount of monitoring a competitor produces either. That is the same half the model cut, so both failures point the same way: towards the competitor, and away from the deal.

Flares keeps the competitor half current and dated, so the brief starts from this morning’s pricing and packaging rather than from whatever was true when the card was written. Carrying the account half into the prompt stays a job for whoever was on the last call.

Which is the argument for asking what was cut, every time, even when the brief looks finished. A finished-looking brief is the normal output. The four lines it removed to get there are the ones that made this call different from every other call against the same competitor.