Founder research stack for positioning, market maps, and pricing analysis

How to use this pack

Every prompt below is meant to be copied, filled in, and kept somewhere you can paste from, so you run it again next quarter rather than rewriting it. Bracketed sections are yours to replace. If your research is academic rather than commercial, the researcher workspace covers the reading and citation side of the same setup.

Two rules make the difference between useful output and a nicely formatted guess. First, every factual claim needs a source you have opened, because a fabricated market size in front of an investor is a memorable way to end a meeting. Second, paste real customer language wherever you have it. The gap between research that reads like a consultant's template and research that is about your company is almost entirely the presence of real quotes.

Job	Model	Why
Market map and category scan	Gemini	Recent web material with links you can check
Long reports, filings, transcripts	Gemini	Largest context window, so the whole thing fits
Competitor teardown	Gemini, then Claude	Retrieve first, then synthesise
Pricing structure	GPT	Reliable structured output and arithmetic you can verify
Positioning and messaging	Claude	Writes like a person, holds a consistent argument
Customer language synthesis	GPT for extraction, Claude for the write-up	Two different jobs

Market map

Prompt 1: the category scan

Research the [category] market. Return: (1) size estimates from the last 24 months, each with the publishing organisation, year, and URL; (2) the 8 to 12 most relevant players with any public revenue, funding, or share figures; (3) material changes in the last 12 months, meaning entries, exits, funding, acquisitions, and regulatory changes, with dates; (4) the two or three structural forces most likely to reshape the category in 24 months. Where sources disagree, show both figures rather than averaging. Mark anything unsourced as UNVERIFIED.

Prompt 2: the segmentation cut

Using the scan above, propose three different ways to segment this market: by buyer, by job to be done, and by company characteristic. For each segmentation, say which segment is currently underserved and what evidence in the scan supports that. Reject any segmentation where the segments would buy the same product for the same reason.

The second prompt is the one that produces something usable. A market map that only lists companies tells you who exists; a segmentation tells you where to aim.

Competitor teardown

Prompt 3: the teardown

Build a teardown of [competitor]. Cover: stated positioning in their own words, target segments, pricing where public, distribution channels, the flow for [the job to be done] as documented in their help centre, product or messaging changes in the last 12 months with dates, and complaint themes visible in public reviews. Cite every claim with a URL. Separate what the company states from what third parties observe.

Prompt 4: the positioning map

Place each competitor from the scan on two axes that genuinely differentiate this category. Propose the axes and justify why they matter to a buyer, then reject any axis pair where one axis is a restatement of the other. Show where we sit, which space is currently uncontested, and what would have to be true for that space to be worth owning.

Prompt 5: the vulnerability read

For each major competitor, identify the customer their positioning implicitly excludes and the complaint theme they have not addressed in 12 months. Which of those gaps could we credibly serve given [our capabilities]? Be specific about what we would need that we do not have.

Pricing

Prompt 6: structure before number

Propose three pricing structures for [product]. For each: the value metric, the tier boundaries, who is priced out, who gets a bargain, the failure mode when a customer grows, and what the structure signals about who we are for. Do not recommend one until I have reacted to all three. Context: what it does [x], who buys [y], what it saves them [z], what competitors charge [paste].

Prompt 7: pressure test the price

We are considering [price] for [tier]. Argue that this is too high, then argue that it is too low. For each, state what evidence would settle it and what we could measure in the next 30 days to find out. Do not hedge by concluding it is about right.

The value metric is the part founders most often get wrong and the hardest to change later. Ask specifically: What are we charging for, and does the customer's use of that thing grow at the same rate as the value they get? If those two diverge, the pricing breaks as customers succeed.

Customer language

The most underused research input a founder has is the pile of things customers have already said. Sales notes, support tickets, churn reasons, and review text are all sitting there unread as a set.

Prompt 8: mine the language

Here are [sales call notes / support tickets / review text]. Return: (1) the exact words customers use for the problem, copied verbatim, never paraphrased, with counts; (2) the objections in order of frequency; (3) what they compared us to; (4) the outcome they say they want, in their words. Flag any theme appearing in fewer than three sources as a single observation rather than a pattern.

Verbatim is not a stylistic preference. The words customers use are the words that should appear on your landing page, and any paraphrase converts their language into yours, which is exactly the thing you were trying to remove.

Prompt 9: from research to copy

Using only the verbatim language above and the positioning we chose, draft a landing page: hero headline stating a specific outcome, subhead, the problem section in the customer's own words, how it works in three steps, proof, the strongest objection handled, and a final call to action. No claim beyond the evidence provided. Flag anything you had to invent.

The one-afternoon sequence

Run in this order, in one Whizi thread so each step sees the previous ones.

1. Category scan and segmentation, in Gemini. About 40 minutes including checking the sources.
2. Teardowns of your three closest competitors, in Gemini. About 30 minutes.
3. Positioning map and vulnerability read. About 20 minutes.
4. Customer language mining from whatever notes you already have, in GPT. About 20 minutes.
5. Pricing structures, in GPT. About 20 minutes.
6. Landing page draft, in Claude, which now has all of the above in context. About 20 minutes.
7. Open every source. This is not optional and it is the step people skip.

Keeping it in one thread is what makes step 6 work. The drafting model can see the actual competitor language and the actual customer quotes rather than your summary of them, which is the difference between copy that reflects the research and copy that reflects a memory of it.