Establish truth
Make the product claim, data grain, and evidence source explicit before automation amplifies them.
A system cannot be more trustworthy than the meaning and provenance of the data it acts on.
See the failure before the pattern
The arithmetic was correct. The grain was wrong.
Why censored detailed rows cannot reconstruct an exact coarse total, and how to make aggregation grain part of the product contract.
Two screenshots became a failing data contract
A debugging workflow that normalizes two conflicting claims, traces one value to source, and preserves the finding as an executable contract.
Copy the control, then adapt it
Screenshot to data contract
A playbook for tracing a visible discrepancy through UI, cache, query, storage, parser, and source.
Aggregation-grain data contract
A runnable contract for exact, censored, and lower-bound values across detailed and coarse grains.
What this pattern is grounded in
Proof items describe real findings or working controls. Sanitized and reconstructed examples say so explicitly. They are evidence for the lesson, not universal guarantees.