Thirty vendors, one schema, a source on every cell

What is the exact prompt?

Three stages, run as one instruction:

• "I am mapping the market for [category, e.g. warehouse-native customer data tools]. Stage 1, roster: find every vendor that plausibly competes, from category pages on G2 and Capterra, 'alternatives to [known leader]' articles, and the sponsor lists of the two main industry conferences. Target 25–40 names; list borderline cases separately rather than silently dropping them."
• "Stage 2, schema: create a sheet with columns: vendor, one-line description IN MY WORDS not theirs, segment (propose 3–5 segments after seeing the roster, then assign each vendor to exactly one), target customer size, pricing model and published entry price if any, deployment (cloud/self-hosted/both), three named flagship features, most recent funding or acquisition if stated on their site or press page, and source URLs."
• "Stage 3, fill: for each vendor, read their site, pricing page and docs. Rules: blank cell over guess, every cell traceable to a source, mark any claim older than 18 months as possibly stale. When done, write ten lines: the segments with member counts, where the crowding is, and which two segments are adjacent enough that vendors will cross between them."

Segments come after the roster, not before

Segments invented before the data are your priors wearing a diagram.

Letting the roster arrive first, borderline cases included, forces the segmentation to explain actual vendors.

The borderline list is where the strategic conversations live: the vendor that does not fit cleanly is usually either dying or about to redefine the category. The "in my words" rule on descriptions is the anti-slop clause, because thirty homepage sentences all say the same four words.

How much can you trust an afternoon's output?

Trust the structure fully, the cells mostly, and the capability claims least, because vendor sites overstate in exactly the ways a transcription preserves. Treat flagship-feature cells as claimed until competitor battlecards verify the two vendors that end up mattering.

Audit it like a scraped directory: ten random cells, click the sources. The staleness markers earn their place the first time someone quotes a 2023 funding round in a 2026 deck, and a quarterly re-run makes it a recurring research report.

What the map feeds

Positioning.

The crowding analysis in the final ten lines is the deliverable: "eleven vendors in mid-market cloud, two in self-hosted enterprise, nobody under $100/month" is a strategy sentence, and the sheet is its evidence.

Analyst reports and win/loss data are market research automation, and the tool comparison is the market research tools roundup. Keep the sheet with its sources and anyone can re-draw the picture honestly.

Frequently asked questions