Answer Ad Academy
Feeds, Evidence and Trust
9 min read

The evidence an engine needs before naming you

Six evidence types, and which prompts each one unlocks.

Engines assemble recommendations from evidence, and not all evidence is equal. Six types do most of the work: specifications, pricing, fit statements, third-party corroboration, recency signals and exclusions. Knowing which type unlocks which prompt family turns vague "improve the content" advice into a scheduled workstream.

Specifications unlock comparison prompts. When a buyer asks which of two products handles a given requirement, the engine needs numbers on both sides. Missing specs do not produce a cautious mention — they produce silence, because the engine simply cannot complete the comparison with you in it.

Pricing unlocks budget-constrained prompts, which are a large share of all commercial prompts. Published pricing, even as a range or a starting point, outperforms "contact us" dramatically in answer engines. If the client cannot publish exact prices, publish bands and the factors that move them.

Fit statements unlock situational prompts: who this is for, at what size, in what circumstances. These are the statements your hints mirror, and they must exist in public material, not only in the hint. A hint is a targeting instruction; the fit statement is the evidence that backs it.

Third-party corroboration unlocks trust-sensitive prompts. Reviews, analyst mentions, customer stories with named companies and directories that repeat your specs all raise confidence. Self-published claims echoed nowhere else are treated as weaker, particularly in regulated or high-consideration categories.

Recency and exclusions are the two most neglected. Dated, regularly updated pages are increasingly favoured over undated ones, so a visible last-reviewed date is cheap and effective. And explicit exclusions — what you are not for — increase trust measurably, because they make the rest of your claims read as calibrated rather than promotional.

Evidence type → prompt family unlocked

Specifications → "which handles X better, A or B?"

Pricing → "cheapest option that still does X under $500/month"

Fit statements → "best tool for a 40-person firm doing X"

Third-party corroboration → "is A actually reliable for X?"

Recency → any prompt where the engine prefers current, dated sources

Exclusions → "what should I avoid if I need on-premise?" (and calibration everywhere else)

Common mistakes
  • "Contact us" pricing in a category full of budget-constrained prompts.
  • Fit statements that exist only in hints, never in public material.
  • Undated pages competing against dated competitor sources.
Key takeaways
  • Six evidence types, each unlocking a different prompt family.
  • Published price bands beat "contact us" for answer visibility.
  • Explicit exclusions make every other claim read as calibrated.
Check yourself
Three questions, instant feedback. Worth 30 points.

1. What happens on a comparison prompt when your specs are missing?

2. What should replace "contact us" when exact pricing can't be published?

3. Why do explicit exclusions raise trust?