Skip to main content

Overview

Most measurement problems do not begin in the model. They begin in the data.

Common issues

Missing days

Gaps in spend, impressions, conversions, or revenue can distort timing, trend, and contribution views.

Zero-spend days that are not real

Unexpected zero rows can make channels look paused or broken when they were not.

Inconsistent naming

Naming drift makes mapping and interpretation harder and can create false fragmentation.

Hierarchy shifts

When the same media source moves between channel levels or naming conventions, trend comparisons become harder.

Partial uploads or stale files

A file can be structurally valid but still incomplete or older than expected.

Prevention habits

  • Keep delivery cadence predictable
  • Preserve stable naming
  • QA new sources before relying on them in reports
  • Review Data Layers and Connector QA after changes
  • Treat anomalies as a data question first, not only a performance question