Decide whether a reassessment is warranted

A re-run refreshes the evidence the loop reasons from, which matters when the business or its public footprint has actually changed. It also costs credits, and each user has a daily limit on runs. Re-running without a material change mostly produces noise: small differences in what public research could reach, easily misread as progress or decline. The Learn step needs real change in evidence, tied to work done and outcomes recorded, not a second opinion on the same facts.

After a material change: a new website or domain, a corrected business type, newly connected integrations, new channels launched, or a long stretch since the cover's "Assembled" date. Not to check whether something improved.

• The current run's "Assembled" date and the count of runs tracked in the page header.
• A written note of what changed since that date, and why it should change what public research can see.

1. 1Open Marketing → Diagnostic, read the cover date, and ask whether the evidence still describes the business. If nothing material changed, stop here.
2. 2Follow the History link next to the run count to "Diagnosis History". Check the "Evidence map" column so you know whether earlier runs were grounded reads or direct scans only.
3. 3If the change is material, press "Re-run Diagnosis"; the button states its credit cost. The website, business type, and context you entered carry over, so update any that changed before pressing "Start Diagnosis".
4. 4When the new run finishes, return to Diagnosis History and open the earlier run. Its "Diagnosis Snapshot" shows "Evidence changes since this run", with any "Changed Inputs", "Manual Overrides", and "Recorded outcomes in this window".
5. 5Read each change against your note. New evidence where you did the work is worth recording; new evidence elsewhere may simply be better coverage.
6. 6If a plan exists, the report's "What now" section offers "Update my plan", a short check-in meant only for material change. A re-run does not change the plan by itself.

A new run in Diagnosis History with its own date and evidence-map label, an updated "Assembled" date on the cover, and, on the earlier run's snapshot, a count of updates with one line per changed area (for example, that public evidence for an area is now observable, or could not be verified in the latest read).