Track Competitors

Start with two or three brands that come up in lost deals. Shared buyer, comparable price band, same category language.

A giant you never compete with will dominate every chart and teach you nothing.

Competitors are measured inside the same answers as you, so coverage across prompt stages matters. Category, comparison, problem, and brand prompts each expose a different weakness.

Compare visibility, position, and sentiment. Look for concentration: a rival who beats you everywhere is a positioning problem, one who beats you in a single group is a targeted content problem.

Go to the raw AI answers for those prompts. The model usually states its reason in plain language — a capability, an integration, a price point, a use case fit.

That sentence is the most direct competitive intelligence in the product. It is what the buyer reads.

For each losing answer, list the cited sources. You are looking for pages that carried your competitor into the response and left you out.

Sort by how many of your answers each source influenced. That order is your work queue — continue in Find Citation Gaps.

Source changes propagate on the model's schedule, not yours. Give it time, keep the prompts unchanged, and watch relative movement against the same competitors.