Stop 4 of 9 · Weekly
Performance Max diagnosis
Asset groups, search categories and the brand-traffic question, plus the three changes worth proposing and the ones the data cannot support.
What it is
Performance Max is the campaign type Nour clients ask about and the one she can least easily explain. It reports less than Search, it buys across placements she cannot fully choose, and it produces a cost per lead that looks excellent until you ask what is inside it. This stop is the weekly diagnosis: what the reporting will tell you, what it will not, and the three changes worth proposing.
What the reporting gives you
| You can see | You cannot see |
|---|---|
| Spend and conversions by asset group | Which asset drove which conversion |
| Search category themes | A conversion split by query |
| Placement and channel where impressions ran | The auction without Performance Max |
Nour writes findings only from the left column. Anything from the right column is a hypothesis, and a hypothesis goes in the review as a test to run, never as a number in a client table.
The brand-traffic question
Every Performance Max diagnosis starts here. A campaign that can serve on brand queries will win the cheap conversions first, and the blended cost per lead drops for a reason that has nothing to do with the work. The test is a brand exclusion applied for a fixed period, not an assertion about percentages. The exclusion is a targeting change, so a person applies it, and it goes in the guardrail log with a review date.
Using the Google Ads connector on [account name], read only, last 30 days against the previous 30. For the Performance Max campaigns: spend, conversions and cost per lead by asset group; the search category themes with their share of spend; the top placements and channels by spend. Then answer three questions and say which data supports each answer: is spend concentrated in one asset group, is there a brand signal in the search categories, and where is spend running with no conversions at all. Mark clearly anything you cannot determine from the available reporting. Change nothing.
From that diagnosis, give me the three changes worth proposing this week, sorted by dollars at stake. For each: the change, the evidence with rows, what we expect to happen, what would tell us in fourteen days that we were wrong, and who approves. Include the brand exclusion test only if the search categories support running it, and say what data would settle it if they do not.
Write four sentences for the C09 review explaining what Performance Max is doing this month, using only figures from the asset group and placement reports. One sentence on where the spend went, one on the result, one on the change we propose, one on what we will know by the next review.
The measurable
Every diagnosis produces one number Nour can hold across weeks: the share of Performance Max spend sitting in asset groups with a cost per lead above the account target. Track it in the log. If it does not move after a change, the change was cosmetic.