Ad accounts decay quietly. A campaign set up eight months ago for a promotion is still running. An exclusion someone removed for a test never came back. An automatic expansion was switched on during setup because it was on by default.
None of that is a mistake in the strict sense. Together they make the difference between an account that works and one that burns budget. Here are the eight places where I most often find something.
The eight points
1. Search terms nobody ever read
Broad match keywords pull in queries that have nothing to do with your offer. Without a regular look at the search terms report you pay for them for months.
Spot it: Google Ads → campaign → Insights and reports → Search terms, last 90 days, sort by cost.
Fix it: Exclude everything off-topic. Turn recurring patterns into account-level negative keywords, not just campaign-level ones.
2. Brand campaigns buying clicks you would get for free
Bidding on your own brand name often buys visitors who were coming anyway — visible as a suspiciously cheap CPA that flatters the overall numbers.
Spot it: Pause the brand campaign and watch organic traffic and total enquiries for a week.
Fix it: If total enquiries stay flat, that budget was redistribution, not growth. You need brand terms where competitors bid on your name — otherwise not.
3. Automatic expansions nobody asked for
Display expansion on search campaigns, Advantage+ audience expansion on Meta: both stretch delivery beyond what you selected. Sometimes that is good. Often it is where half the budget goes.
Spot it: Look through campaign settings for anything starting with “expansion” or “Advantage”. Then check the placement report for how much budget lands there.
Fix it: Switch the expansion off and compare for one to two weeks. If results hold, it was waste.
4. Conversion tracking that counts twice
When the same conversion arrives via the pixel and via a GA4 import, every campaign looks twice as good. The bidding strategy then optimises towards a number that does not exist.
Spot it: Check which conversion actions are set to primary. Two entries measuring the same event are the warning sign.
Fix it: Keep exactly one source per event as primary, set the rest to secondary. Then let the bidding strategy settle again.
5. Placements nobody excluded
Audience Network on Meta, display partners on Google, in-app placements in games: clicks there rarely lead anywhere, and often come from people who tapped by accident.
Spot it: Open the placement report and sort by cost. High click counts with near-zero conversions stand out.
Fix it: Exclude the bad placements. On Meta it is worth running a feed-only test as a benchmark.
6. Campaigns that never left the learning phase
Too many ad sets on too little budget means none of them gets enough data. Everything learns forever, nothing gets good.
Spot it: Multiply active ad sets by the minimum viable budget and compare with the total. Meta shows “learning” in the status directly.
Fix it: Consolidate rather than split. Three ad sets with enough budget beat twelve that never exit learning.
7. Audiences bidding against each other
Retargeting and prospecting with overlapping audiences compete in the same auction. You bid against yourself and push up your own price.
Spot it: Check audience overlap on Meta. On Google: audiences running in several campaigns without mutual exclusion.
Fix it: Exclude existing customers and site visitors from prospecting. Make it clear which campaign owns which audience.
8. Ads pointing at broken pages
The most expensive click is the one landing on a 404, an expired offer, or a page that does not work on a phone. Meanwhile the account looks perfectly healthy.
Spot it: Open every active landing page, on a phone. Also filter Google Ads for ads with a “destination not reachable” status.
Fix it: Switch dead pages off immediately rather than fixing them later. A monthly slot for this check beats any alert.
A checklist for a Friday afternoon
45 minutes, in this order
- Sort the last 90 days of search terms by cost, exclude the obvious nonsense.
- Go through conversion actions: is exactly one primary per event?
- Open the placement report, exclude the most expensive ones without conversions.
- Search campaign settings for “expansion” and “Advantage”.
- Open every active landing page once, on a phone.
That is enough for a first pass. Points 6 and 7 are structural work rather than tidying — they belong in a quiet hour, not a Friday afternoon.
Where cutting is the wrong reflex
So this does not read as an austerity piece — three cases where less budget is the wrong answer:
- During the learning phase. Cutting a fresh campaign after five days because CPA is still high prevents exactly the learning that would bring it down.
- With seasonal tailwind. When demand is rising anyway, budget is the cheapest lever you have.
- When the real problem is the page. A conversion rate below one percent does not get fixed in the ad account.
The honest conclusion
None of these points is new, and most appear in some guide somewhere. The reason they turn up in almost every account anyway is mundane: tidying up is on nobody's objectives. New campaigns are visible work, excluded search terms are not.
Which is exactly why a fixed slot pays off. Once a month, one hour, always the same order. Unspectacular, and more reliable than the next campaign idea.