Strategy
PPC ad optimization: the 14 days after an account structure audit
After a PPC account structure audit, hold budgets and bids for 14 days, then read the result against the same two numbers, and when to audit again.
What to take away
- A rebuild is not finished when the ad groups are merged and split. It is finished when the result has been read.
- Hold budgets flat for 14 days, and change bids only after two weeks of stable structure.
- Expect about two weeks of unstable CPA, because bidding strategies relearn after major ad group changes.
- Read the result against the same overlap and conversion numbers that drove the audit, and wait for at least 15 conversions before judging a target CPA.
Why a rebuild needs a hold
Restructuring changes the conversion history that automated bidding reads, so the first two weeks are noise. Changing bids in the same window hides which change moved CPA. Structure first, bids second: the hold keeps that order after the rebuild as well as before it.
Bidding strategies such as Target CPA and Maximize Conversions push spend toward whatever the system believes will convert. When ad group history changes, the system has to learn it again. Which strategy fits is a separate decision from structure, covered in Target CPA against Maximize Conversions.
What to do during the 14 days
- Finish the rebuild in Google Ads Editor, with every ad group labeled split, merge or retire beforehand.
- Hold budgets flat for 14 days.
- Leave bids alone until the structure has been stable for two weeks.
- Note the date of every change, so a later swing can be traced to one of them.
Overlap and conversion test
High overlap + thin conversions?
Merge ad groups
Low overlap + healthy counts?
The temptation in week one is to fix a wobbling CPA with a bid change. That swaps a measurable structure change for an unmeasurable pair of changes.
If the number looks wrong, write it down and wait. Fixing what search queries triggered, if anything, belongs to the next audit and not to the hold. The decision rules behind each label are laid out in split, merge or retire.
Read the result against the thresholds
When the hold ends, read the same two numbers that drove the audit. Query overlap is the share of queries appearing in more than one ad group, and conversion depth is how many conversions each ad group recorded in the window. Treat them as thresholds, not verdicts.
| Check | Before the rebuild | After the hold |
|---|---|---|
| Query overlap | Over 60 percent suggested a merge | No query triggers ads from two ad groups |
| Conversions per month | Under 15 combined, no ad group could carry a target | At least 15 conversions before judging a target CPA |
| Blended CPA | Within 10 percent of target for a merge | Falling once the bidding model has volume |
| Landing page | One page serving both intents, or one intent per page | Matched to the ad group promise |
Split, merge, retire thresholds
Split
- Query overlap
- Under 20%
- Conversions/mo
- 15+ each
- Target CPA
- Each beats target
- Landing pages
- One intent per page
Merge
- Query overlap
- Over 60%
- Conversions/mo
- Under 15 combined
- Target CPA
- Blended within 10%
- Landing pages
- One page serves both
Retire
- Query overlap
- Any, no conversions
- Conversions/mo
- Zero for 90 days
- Target CPA
- Above twice target
- Landing pages
- Page no longer live
One signal rarely decides, and two or three together usually do. If one row looks good and the rest do not, wait for more conversions rather than reversing the change.
An example: the merge that took three weeks to show
A US plumbing advertiser ran one ad group for water heater repair and another for water heater replacement, both on phrase match. The search terms report showed 68 percent shared queries and 11 combined conversions per month, so neither ad group could carry its own target CPA.
The merge produced one ad group with two responsive search ads, one per intent, plus negative keywords separating them. Conversions held flat for three weeks, then CPA fell once the bidding model had volume to work with. Judging the merge in week one would have missed the later fall.
Merged ads need a claims check as well. The FTC states that claims in advertisements must be truthful, cannot be deceptive or unfair, and must be evidence-based, so each ad in a merged group must promise only what its page delivers.
A post-rebuild checklist
- One intent theme per ad group
- No query triggering ads from two ad groups
- Negative keyword lists checked for cross ad group conflicts
- Landing page matched to the ad group promise
- At least 15 conversions before judging a target CPA
- Bids unchanged until two weeks of stable structure
The FTC's policy statement says advertisers need a reasonable basis for objective claims before they are disseminated, so run that check before the new ads go live.
If the list is clean and CPA is still high, the cause may lie outside structure. That is the point to look at Quality Score, since ad relevance and expected click-through rate are scored per keyword and per ad group.
When to audit again
Run the audit quarterly for stable accounts, then leave the account stable until the next data window closes. Run it monthly when spend moves by more than 20 percent or a new campaign launches. The wider search advertising routine puts the audit alongside the other checks.
Scripts can flag new query overlap every month, so the next audit starts from a list and not from a guess.
If budget is limited, retire dead ad groups first. They spend money, add no conversions and distort the conversion data bidding relies on. The common search advertising questions cover what else belongs in that first pass.
Common questions
Does restructuring reset the learning period?
Yes. Bidding strategies relearn after major ad group changes, so expect about two weeks of unstable CPA.
When can bids change again?
Only after two weeks of stable structure. Changing both at once hides which change moved CPA.
How do I know the merge worked?
Read overlap and conversion depth again after the hold, and wait for at least 15 conversions before judging a target CPA. In the plumbing example the improvement arrived after three weeks.
What should I fix first if budget is limited?
Retire dead ad groups first. They spend money, add no conversions, and distort the conversion data bidding relies on.
