Run a competitor sweep

Step 1 — Define the arena, not the enemy list

I’m mapping the competition for [my thing], because I need to decide [the decision]. Interview me first. Then list who actually competes for the same customer money — not the companies I fear, but the alternatives my customer genuinely weighs, including “do it themselves” and “do nothing.” Cap the list at 5–8, ranked by how often my customer actually considers them.

The arena is a customer’s-eye view. The spreadsheet your customer already uses is often your biggest competitor, and no rival’s website will tell you that — the interview is what surfaces it.

Step 2 — Design decision-shaped columns

Given my decision, design the table’s columns: the questions I’d need answered before deciding — not a generic feature list. Start from: THE OFFER in one line; ENTRY PRICE and pricing model; WHO IT’S REALLY FOR; BEST AT; VISIBLE WEAKNESS; MOMENTUM (last thing they shipped or announced); SOURCE + DATE for every cell. Add or cut columns based on my decision, and tell me why.

Columns are the sweep’s steering: ask for “info about competitors” and you get the average blog post; ask the questions your decision depends on and you get evidence.

Step 3 — Sweep: one competitor per fresh window

Now the move that makes this lab scale. Each competitor gets its own fresh window, same brief every time:

Using web search, fill this row for [competitor]: [paste the column brief]. Rules: every cell carries its source and the source’s date; prefer their own pricing page and docs over articles; where a fact can’t be found, write UNKNOWN — and record “call for pricing” as exactly that, it’s a data point, not a blank to fill. Return just the row.

Why per-competitor windows: one long window sweeping eight companies fills and rots, and worse, it cross-contaminates — competitor five’s pricing starts echoing competitor two’s. Eight clean windows, one brief, eight clean rows. If this feels familiar, it should — it’s the multi-agent shape with you as the queue.

Step 4 — Assemble, then audit the load-bearing cells

Merge the rows into your table. Then open one more fresh window, paste the table, and audit — not everything, the cells that would change your decision (usually pricing and the BEST AT claims):

Audit these cells against their cited sources using web search: SUPPORTED, STRETCHED, or UNSUPPORTED — the Build a cited research brief scale. Worst first.

Then the human pass: open the two most decision-critical sources yourself. Pricing pages change quietly and get cached loudly; thirty seconds of your own eyes on the two cells that matter is the difference between a map and a rumor.

Step 5 — Read the map, then give it a home

Looking at the completed table: where is the white space — what does everyone charge for that nobody does well? Where am I actually differentiated, per the evidence rather than my hopes? And what would have to be true for me to win at [my intended price]?

Land the table where decisions live — Sheets or Notion, one row per competitor, the checked-date on every row. Save the sweep prompt next to it: next quarter, eight fresh windows re-run the rows in minutes, and the diff — who raised prices, who shipped, who went quiet — is the market telling you what changed.

---

Check your understanding: why does each competitor get its own window — and which cells earn the audit?

Use this process for other comparisons. Arena → decision-shaped columns → one subject per fresh window → audit what’s load-bearing → a dated living table. That’s vendor selection, tool bake-offs, the job-market scan, the grant landscape — any decision that deserves a map instead of a vibe.