Resources
Guides and worked examples
Short, concrete walkthroughs. Every example uses sample data and is labelled as a demo unless it is connected to a live account.
Why did CPA increase this week?
A worked diagnosis that separates the observed change from the hypothesis behind it.
Example below
Reading an evidence panel
How ranked contributing factors, comparison windows, and data caveats fit together.
Written guide
Reviewing a spend-affecting action
What to check before approving a budget, status, or targeting change — and how it is reversed.
Written guide
Empty, stale, and partial data
Why the workspace refuses to show missing data as a zero, and what it shows instead.
Written guide
Worked example
Why did CPA increase this week?
The same structure the product uses: observation, evidence, an explicitly labelled hypothesis, a recommendation, and a caveat.
“Why did CPA increase this week?”
Observation
CPA moved from $21.10 to $26.17 while spend increased 6.4% and conversions fell 4.1%.
Evidence
Broad — Lookalike 3%
49% of prospecting spend; CPA up 38%
$12,480 spend · CPA $31.40
Retargeting — 30d visitors
Flat spend; most efficient set
$7,110 spend · CPA $12.05
Creative — “Bundle offer v2”
Only creative in the affected ad sets
CTR 1.1% (down from 2.4%)
Conversion volume
Attributed, 7-day click window
1,842 (−4.1%)
Spend shifted toward two broad ad sets while conversion volume stayed flat in retargeting.
Likely creative fatigue in the broad ad sets rather than an audience or bid change. This is a hypothesis, not a confirmed cause.
Recommendation
Current
“Bundle offer v2” active in 3 ad sets, $12,480 of 7-day spend
Proposed
Pause “Bundle offer v2” in the two broad ad sets and reallocate 15% of their daily budget to retargeting
Caveat. Attribution windows differ between ad sets; 2 days of data arrived after the last sync. Treat the conversion delta as approximate.
See it run on your own account.
Request access and connect an advertising account.