When we audit a Google Ads account that is not performing, the causes are rarely exotic. Three problems account for the overwhelming majority of wasted spend, and they compound: each one makes the others harder to see.
1. The conversion data is wrong before anything else is
Automated bidding optimises toward the conversions you tell it about. If those conversions are counted twice, or if a form view is recorded as a submission, or if half your enquiries arrive by phone and are never recorded at all, then the algorithm is optimising toward a distorted picture of your business.
The specific failures we find most often:
- Duplicate conversion actions. The same enquiry counted by a thank-you page rule and a form-submit event, inflating reported performance by roughly double.
- Every conversion valued the same. A newsletter signup and a quotation request are both recorded as "1", so bidding treats them as equally valuable.
- Untracked phone calls. Common in service businesses, where the phone is the main enquiry route and none of it reaches the platform.
- No offline import. Leads are tracked, but which leads became customers never returns to the platform, so it optimises for volume rather than quality.
Fix this first. Every other optimisation rests on it, and doing the rest first means measuring your improvements with a broken instrument.
2. Broad match without governance
Broad match is not the problem people often claim it is. It can surface valuable queries you would never have thought to add. The problem is broad match with no negative keyword strategy and no regular search term review, which is a mechanism for spending money on searches that were never going to convert.
A workable discipline looks like this:
- Review the search terms report weekly while the account is young, then fortnightly once it settles.
- Maintain negatives at two levels: an account-wide list for terms that are never relevant, and campaign-level lists for terms that are relevant elsewhere but not here.
- Watch for the classic categories: "free", "jobs", "salary", "diy", "how to", and competitor brand terms you have decided not to bid on.
- Promote consistent winners into their own exact match ad groups where the volume justifies it.
3. Segmentation that does not reflect margin or intent
Many accounts are structured by product category because that is how the website is organised. That is convenient, and it is usually the wrong axis. The useful segmentation is by intent and margin, because those are what should drive different bids.
Someone searching "emergency plumber near me" and someone searching "how much does a boiler service cost" are at different points and are worth different amounts. Grouped into one campaign with one budget, the cheaper, higher-volume, lower-intent queries will absorb the budget and the expensive high-intent ones will be starved.
A structure that tends to hold up
- Separate campaigns for high-intent and research-stage queries, each with its own budget cap.
- Brand traffic isolated in its own campaign so it does not flatter the performance of everything else.
- Geography split where cost per click or conversion rate differs materially between locations.
- Shopping segmented by margin and bestseller status using custom labels, not by supplier category.
Where Performance Max fits
Performance Max can work well once you have reliable conversion data and a genuine reason to reach beyond search. Deployed as an account default, it tends to absorb budget that should have been protected for high-intent search, and to claim credit for brand traffic that would have arrived anyway. If you run it, use brand exclusions, apply audience signals, and keep a separate search campaign so you can still see what search is doing on its own.
The order that matters
If you do nothing else, do these in this sequence: fix the conversion tracking, add the negative keyword governance, then restructure the campaigns. Reversing the order means you rebuild the account and then discover the data you judged it on was wrong.
