Executive Summary Prompts for Competitive Analysis

A one-page brief does two jobs that look like one. It shortens an analysis, and it says what to do about it. The first is arithmetic: rank, cut, keep the dates. The second needs facts about your own company, and those are not in the analysis.

Ask for a brief and you get both, in one voice, at one confidence level. The deprecation date and the recommendation to act on it sit in the same paragraph, set in the same type. One of them can be checked in thirty seconds. The other cannot be checked at all.

This is not the usual problem of a missing source. Every other gap in competitive work closes by pasting something in. There is nothing to paste here. What a recommendation needs is cost, capacity, sequencing and appetite, and an analysis of a competitor holds none of it.

A ranking is the judgement your material supports

So the useful output is narrower than a brief and more useful than a summary. Rank the findings against one named decision. That ranking is arguable in the way a colleague’s ranking is arguable, and every line in it points at something you supplied.

Then ask what the material does not contain. That list is short, it is nearly always about you rather than about the competitor, and it is the closest thing to the recommendation you can honestly get from a model. It also tells you who to go and ask. Where the whole period rather than one decision is the subject, a competitive intelligence report is the document that takes it, and a brief is cut from that rather than the reverse.

Three inputs, and the order matters. The decision goes above the analysis, not after it. A model reading a long document answers whatever it was asked first, so a decision stated at the end arrives too late to filter anything.

Most people skip the second input, and that is what turns a brief into a newsletter. “Understand their strategy” is not a decision. “Whether to fund a migration offer before the second quarter” is. The difference is that somebody can say no to the second one. Where the question arrives from another team, a written request brief is where that sentence gets captured before any work starts.

Three prompts, and the third one is a separate conversation rather than a final bullet. That is deliberate, and it is the hardest-won instruction here: an instruction asking for commentary after an output does not change the output.

The second prompt forbids three things by name: a recommendation, an owner, a deadline. Forbidding them works better than asking for caution, because caution is a tone and the prohibition is a format. A model told to be careful still names a plausible job title.

Cutting is the other half of the job, and it is where a word limit does its damage. A model cuts by generality, so the lines only your situation knows go first. Competitive sales call prep runs on a two-hundred-word limit and loses all four account-specific facts to exactly that.

We briefed a real analysis and read the result back against it. The material was three public sources on one competitor: two years of its annual filing, an earnings call, and its dated developer changelog. The decision was named in the prompt.

The audit line only looks backwards

The brief ended by listing what it had left out. Two items, both real, both correctly judged irrelevant. It listed none of the four things it had added.

That asymmetry is built into the instruction. “Say what you left out” asks a model to compare the output against the input and report what is missing from the output. Additions are missing from the input, and nothing asked about those. So the request that feels like a safeguard audits one direction and is silent on the other.

A brief that is candid about its omissions and quiet about its inventions is worse than one that is candid about neither. The cut list reads as evidence that somebody checked. The third prompt is the check the cut list is mistaken for, and it has to run in its own conversation on the finished text.

Changing the reader changes the order, not the invention

Naming a different reader does real work. Briefed for a finance leader, the material led on the deadline and the pressure on upgrade rates. Briefed for a product leader, the same material led on the three competitive factors the filing had added, and the deadline dropped to fourth. Both rankings were defensible and they were genuinely different.

Both also closed with a recommendation and a named owner. The variable you supply moves the emphasis. The variable nobody supplies gets filled in either way.

The product brief added something else worth naming. It read the filing’s list of competitive factors as a statement of what the company is building. A filing says what a company competes on, which is a narrower claim, and the way to test the difference is to read that list beside the same period’s release notes and filings. A brief has no room for that step and no way to show that it was skipped.

A brief is only as current as the analysis under it, and an analysis ages from the day it is written. The gap between reading a competitor’s filing and briefing the exec team on it is often three weeks. In that time a price moves, a page changes, or the thing you are briefing on ships.

The dates are what protect you, which is why they are required on every claim above. A brief saying a competitor announced something on 15 September can be checked. The same brief without the date invites a question nobody in the room can settle, and the meeting moves on to something else.

No tool fixes the second half of this. The recommendation still comes from people who know what the quarter can absorb, and time to insight is a property of your own process rather than of your sources. Where a brief recurs on a rhythm, the quarterly competitive update is the document that carries it, and last quarter’s version is what makes this quarter’s readable.

Flares keeps a dated record of what competitors change across pricing, product, documentation and messaging. The analysis behind a brief is then current on the morning it is read, and every claim in it already carries the date it became true. What a competitor said in a filing, and what it means for next quarter, is still two separate reads — and only the first one can be kept fresh for you.

Between briefs, a weekly competitor digest is what stops the next one opening with something the exec team heard three weeks ago from somebody else.