Using the Analyst to Iterate a Positioning Brief

A positioning brief takes most PMM teams three weeks to land. Two of those weeks are spent waiting — for the CEO's edits, for sales leadership's reaction, for the third round of "can we sharpen the category line." The work itself is maybe six hours. The rest is calendar friction.

An AI analyst, used correctly, collapses the calendar. Not by drafting the brief — that's the wrong job for it — but by acting as the patient, well-briefed second reader who's available at 11pm on a Tuesday when you want to test whether the new category line actually holds.

3 weeks

What the analyst is actually for

Most PMMs reach for an AI tool and ask it to write the brief. That produces something polished, generic, and immediately rejected by the CEO — because the CEO can tell, within two paragraphs, that no one has thought hard about the category. The brief is a thinking artifact. Generated thinking is uncanny in a specific way that executives are good at detecting.

The analyst is for the second pass. You write the brief. The analyst stress-tests it. Specifically, it does four things humans are bad at being available for:

• Reads the entire 2,000-word brief without skimming
• Holds every prior version in working memory
• Asks the dumb-but-load-bearing question without social cost
• Plays the buyer, the competitor, and the skeptical board member in sequence

Step 1 · Load the strategic context first

The analyst can't pressure-test a brief if it doesn't know what was true last quarter. Before you paste the draft, paste the inputs that produced it: the win/loss notes from the last twelve deals, the three competitors you're actually losing to, the analyst-relations transcript from March, and the pricing page as of this morning.

This is not optional. An analyst with no context will produce reasonable-sounding feedback that ignores the two facts that matter most: that your largest competitor just repositioned, and that your last six losses were all on the same objection. Loaded with context, it will surface those threads unprompted.

DO

The first version of the brief I tested with the analyst came back with one question: "Why does the category line ignore the objection that closed your last three losses?" I'd missed it. My VP would have caught it in week two.

Step 2 · Write the brief without the analyst in the loop

Draft v1 by yourself, or with one human co-author. Forty minutes, longhand or in a doc. Don't prompt the analyst to "help you draft." The brief is a record of what you believe, and the analyst doesn't believe anything — it pattern-matches. A pattern-matched brief reads like every other brief.

The output of step two is a brief that is yours, has rough edges, and contains at least one claim you're nervous about. That nervous claim is what the analyst exists to test.

Step 3 · Run the four-lens interrogation

Once the brief exists, run it through four passes. Each pass is a different prompt, a different role for the analyst, and a different output.

The four passes take about ninety minutes total. They produce roughly forty distinct critiques. About a third are noise — the analyst over-generates. The other two-thirds are the same critiques you'd get from your VP, your CRO, and your CEO over the next three weeks, compressed into a single afternoon.

Step 4 · Triage and revise

Now the brief is your problem again. The analyst gave you forty notes; you'll act on twelve. The triage is the hard part and it's where PMM judgment lives — knowing which buyer concern is real, which competitor parallel is coincidence, which CEO pushback is principled and which is preference.

Triage criteria for analyst feedback

Revise. Run the four passes again on v2. The second cycle takes thirty minutes and produces maybe eight new critiques, mostly second-order. By v3 the analyst has stopped finding meaningful gaps, which is the signal to send the brief to humans.

Step 5 · Use the brief to brief the analyst

Once the brief lands, save the prompt pack — the four pass prompts, your context dump, and the version history. The next time you iterate (a launch, a repositioning, a category shift), you re-load the same context plus the new inputs. The analyst doesn't remember across sessions, but your prompt pack does.

This is the compounding move. The first brief takes three days. The second takes one day. The third is a same-afternoon job, because the context is reusable and the four passes have become muscle memory.

What this doesn't replace

The analyst doesn't replace the live conversation with your CEO, your CRO, or the three customers whose renewal is six weeks out. Those conversations produce signal the analyst can't generate — emotional weight, political nuance, the half-formed objection a buyer hasn't articulated yet. You still run them.

What changes is the order. Before, you ran those conversations to find gaps in v1, then revised toward v2, then ran them again. Now you find most of the gaps before the first conversation, and the live time gets spent on the two or three things only humans can answer: "is this who we want to be," "will sales actually use this," and "what are we willing to give up."

That reorder is the compression. Three weeks becomes three days because the slow part — discovering you missed something obvious — happens on Tuesday at 11pm, not Friday at the leadership review.

What to do Monday

Pick the brief you're least confident in — the one sitting in a doc with a comment from your CEO that says "let's discuss." Pull together the context dump: last six losses, top two competitors, current pricing. Block ninety minutes Tuesday afternoon. Run the four passes. You'll have a sharper v2 by 5pm than you'd have by next Friday.