Sample audit brief

Fictional scenario on seeded e-commerce data: weekly conversion rate drops from 3.42% to 3.14% (−8.2% week-over-week). The audit isolates where the drop lives and quantifies each driver's contribution.

Variance decomposition (excerpt)

The North-Star move is split into additive driver contributions; each node carries its share of the −0.28pp drop.

Conversion rate −0.28pp WoW (−8.2%)

Mobile web checkout: −0.20pp (72% of the drop)
Step-3 payment call p95 latency rose from 2.1s to 8.4s at release 41.2; desktop checkout is flat over the same window.
Paid-search traffic mix: −0.05pp (18%)
Brand/non-brand mix shifted 4pt toward non-brand, which converts 38% lower; within the campaign's planned range.
All other segments: −0.03pp (10%)
Within normal weekly variance (robust z-score < 1.5 on every remaining segment).

WBR brief (excerpt)

Finding: The conversion drop is a mobile checkout engineering regression, not a demand problem.
Evidence: Latency and conversion break at the same deploy timestamp; desktop is flat; the decomposition attributes 72% of the drop to mobile step 3.
Recommended action: Roll back or hotfix the step-3 payment call; expected recovery of ~0.20pp of the 0.28pp drop within one release cycle.

This is a redacted, seeded-data sample, clearly labeled as such — no client names, tables, or figures. In a real audit, every figure above links to the query and result rows that produced it.

Book a scoping call