Sandbox test pass rate for AI-drafted Salesforce changes: what it counts
A sandbox test pass rate can be 99% or 0% on the same data, depending on the denominator. Here is where the figure is made, and what a PR gate with human approval does to it.
SECTION 03 · 8 ENTRIES
Reporting
How attainment and pipeline numbers get reported — the dashboards revenue teams read, and the definitional choices buried inside each metric.
- Entries
- 8
- Updated
- 26 October 2026
- Written by
- P. Oduya
How to read a GEO share of voice dashboard before you hire anyone
A GEO share of voice percentage is a ratio over a prompt set and an engine list that someone chose. Here is how to find both, and what a RevOps team should conclude before paying a consultant.
Built-in reverse ETL: tracing a CRM tile back through two hops
A weighted pipeline tile and its warehouse model disagree by 300,000 dollars. The gap lives in two hops, and built-in reverse ETL keeps both in one history.
Salesforce agent approvals: which denominator is your approval rate using?
The approval-rate tile reads 75 percent. Divide by drafted changes and it reads 60. The audit log cannot settle it, because a denied call never reaches it.
What counts as a Salesforce governed change in a DevOps dashboard
The governed share on a DevOps dashboard can read 100, 80 or 50 percent for the same quarter. The difference is the definition, not the work.
What a credits-used tile is actually counting
The tile says 67 percent of credits used. The unit behind it is an AI response, and that answers a different question from the one the label implies.
The pipeline tile after a fractional exit
The tile still renders after a fractional exit. The field it was built from stopped being written weeks ago, and the number is now a snapshot nobody labelled.
What an AI citation count dashboard is actually counting
A dashboard tile reporting a rising citation count is reporting whatever the underlying tool decided counts as a citation — a definitional choice, not a neutral measurement.