Intelligence Source Vetting (QA)

Intelligence Source Vetting (QA)

1. 01

   Classify each data source by what it can and cannot prove

   Traffic panels, ad-spend estimates, and tender indexes each carry different error bars. Label every source as directional, benchmarked, or auditable before it enters a client deliverable.

2. 02

   Verify the vendor's refresh cadence against the decision timeline

   A quarterly-refreshed dataset cannot support a weekly campaign pivot. Confirm update frequency in writing and note the lag on any chart that reaches a client deck.

3. 03

   Re-read the vendor's monetization and usage terms before publishing derived figures

   Monetization-relevant terms shifted across 105 of 175 AI tools between June and October 2026, so a clause that permitted reselling benchmark data last quarter may not survive this one.

4. 04

   Cross-check any single-source claim against a second independent dataset

   If a competitor's spend estimate from an ad-tracking platform cannot be corroborated by traffic or procurement data, present it as a range, not a point.

5. 05

   Strip unverifiable vendor claims from the analysis layer

   Tools that market themselves as autonomous agents often are not, and repeating that framing in a client report transfers the credibility risk to the agency.

6. 06

   Log every source, pull date, and known limitation in a shared register

   The register becomes the audit trail when a client challenges a number six months into the retainer.

7. 07

   Assign a confidence tier to each insight before it leaves QA

   High confidence requires two or more independent sources; single-source findings ship as hypotheses with the gap stated plainly.