The review is complete only when it produces decisions with owners, evidence windows and reversal rules. A dashboard walkthrough is not a weekly operating meeting.
Use this resource when
Download the working file.
This is the complete editable template—not a PDF preview. Add your own account data, owners and decisions.
1. Open with commercial outcomes
Record spend, qualified leads, opportunities, pipeline and revenue before CTR, CPM or CPC. Compare the previous complete period and the same weekday mix. Do not compare an incomplete Monday with a full prior week.
- Spend and pacing
- Cost per SQL and opportunity
- Pipeline created and weighted pipeline
- Lead rejection rate by source
2. Explain what changed
Write one sentence for the movement, one for the likely mechanism and one for the evidence. Separate observed facts from hypotheses so the team does not turn correlation into a campaign edit.
3. Review tests at the right unit
Creative tests should name the variable—hook, proof, offer or format—not simply list ad names. Search tests should identify query class, match type or landing-page change. Record the minimum spend or conversion count required before calling a result.
4. End with owned decisions
Every action needs one owner, one deadline, one expected mechanism and one success threshold. Limit the weekly queue to the three actions most likely to affect qualified pipeline.
- Keep / change / stop decision
- Owner and due date
- Expected metric movement
- Reversal condition if performance deteriorates
5. Control generated journeys after automation changes
Use the Automation & URL Control sheet after an AI Max migration, generated-asset change, landing-page expansion or other material automation change. Record the exact system, generated asset, selected URL, buyer intent, active control and lag-safe review date before choosing KEEP, MONITOR, EXCLUDE, FIX or ROLLBACK.
- Separate lead-generation qualified-pipeline evidence from ecommerce contribution evidence
- Name AI Max and Performance Max explicitly rather than combining them
- Require an approval owner and rollback trigger
- Wait through the evidence window before calling a result
Practical questions
Questions that change the decision
Should the team review every campaign weekly?
No. Review exceptions, material spend and active tests. Stable low-spend campaigns can be handled monthly.
What if B2B sales cycles are longer than a week?
Use cohort and stage-progression views. Weekly pipeline created is useful, while closed revenue should be read over a longer window.
How many actions should come from one review?
Usually three. A list of 15 changes creates noise and makes causal learning impossible.
Does the automation sheet recommend turning AI off?
No. It records the generated journey and business evidence, then supports KEEP, MONITOR, EXCLUDE, FIX or ROLLBACK. The decision depends on the control state, evidence window and commercial outcome.