Docs/ people/ Run a Weekly Evidence Review
guidev1stableVerified 2026-08-01

Run a Weekly Evidence Review

Review observed AI request changes, choose a bounded follow-up, and verify the next week against the same context.

Prerequisites

  • A monitored domain with server-side request evidence.
  • A stable comparison range and the Evidence Model.
  1. Open the weekly trend for one domain and confirm the selected domain, date range, and request class.
  2. Note only material changes in observed request share or per-platform requests.
  3. For each change, select one matching check from the weekly trends guide: important-page distribution, robots.txt, adapter health, classification evidence, or sampled citations.
  4. Record the evidence type and observation window with the proposed action.
  5. Recheck the same view next week before attributing a change to that action.
  6. For an assistant-led review, connect through MCP and ask for the same domain and range.

Expected result

You leave with a short list of diagnostic follow-ups, not a claim that crawler activity caused citation or referral growth.

Verify

Use the dashboard weekly view, or retrieve the same data directly:

curl "https://api.seellm.link/api/analytics/weekly-trend?range=90d&domain=example.com" \
  -H "Authorization: Bearer sk_live_your_api_key_here"

Compare the returned domain and range with the dashboard selection before recording the review.

Troubleshooting

  • If the comparison is noisy, keep the selected window stable rather than changing it until a preferred result appears.
  • If requests decline, inspect deployment health and access policy before attributing intent to a platform.
  • If an important page is missing, confirm that the middleware matcher covers the route.