When the ad platform says zero conversions you did get
Setting up a TikTok campaign trial for Binderdex, the handoff surfaced something I did not expect: one campaign objective reported conversions from SKAN, while another reported zero, even though TikTok itself attributed conversions to the campaign. SKAN is not globally absent. It is incomplete per objective.
That matters because if you trust the dashboard zeros, you kill a campaign that is working. The handoff now documents this gap so the next person (or agent) picking up the campaign does not rediscover it the hard way.
What a durable handoff looks like
The commit is a single durable artifact: an HTML report plus the SQL queries that produced it, checked into the repo. The point is that a future agent can pick up the trial without re-deriving what was already learned. Three SQL files cover performance, summary, and selected-post views.
Public view counters are rounded, so they are documented as creative-screening evidence rather than paid conversion proof. The handoff is explicit about what each number can and cannot tell you.
The open gap: product-side attribution, production event emission, and the SKAN mapping remain unverified. The handoff calls those out rather than pretending the trial is closed.