In almost every advertising account, part of the budget goes out without producing anything. The problem is not knowing it: it is locating it precisely, quantifying it, and telling apart what truly does not convert from what converts without being measured. Here is the sequence of questions that does this work.
Step 1: Check that measurement is reliable
Before judging performance, judge the instrument. If a conversion has not been recorded for weeks, or if an important action is not counted in the goals, the rest of the analysis is skewed.
On my Google Ads account, list every conversion action: its name, its status, whether it is included in the account goals, and the date of the last conversion recorded for each. Flag the ones that have recorded nothing for more than 14 days.
Two anomalies come up constantly. The first: an active conversion that has been silent for weeks, the trigger is broken. The second: an action that is properly recorded but NOT included in the goals, so the bidding algorithm does not optimize for it, and your reports understate it.
Step 2: Isolate what spends without producing
Once measurement is reliable, the question becomes simple: where does the money go, and what does it bring back?
Over the last 90 days, give me per campaign: cost, number of conversions, cost per conversion and share of total budget. Sort by decreasing cost. Highlight the campaigns that account for more than 5% of the budget with fewer than 3 conversions.
The ninety-day window is not arbitrary: over thirty days, a campaign with a long sales cycle looks like it brings nothing back when it actually produces. And the “more than 5% of the budget” criterion avoids wasting time on campaigns that cost three euros.
For the campaigns you just identified, calculate the total spent over 90 days and bring it down to a monthly amount. That is the amount I recover if I stop them today.
This last question changes everything in a discussion with a client or a management team. “Three campaigns are underperforming” triggers no decision. “These three campaigns cost €480 per month for zero sales” triggers one immediately.
Step 3: Go down to the search level
A campaign can be profitable overall while burning part of its budget on searches unrelated to what you sell.
Over the last 90 days, give me the 30 search terms that cost the most without generating any conversion. For each, show the cost, the number of clicks and the campaign concerned.
Review the list yourself: this is the one moment where your knowledge of the business beats any algorithm. The classic patterns to look for:
- Job searches, “recruitment”, “job offer”, “internship”, when you sell a service.
- Searches for free resources: “free”, “pdf”, “template”, “tutorial”.
- Repair or spare-part searches when you sell new products, and vice versa.
- Competitor brands, if you have not deliberately decided to target them.
- Cities or countries outside your service area.
Step 4: The invisible placements
On campaigns that run beyond the search network, part of the budget can go to mobile apps and unrelated sites, often through accidental clicks.
For my Display and Performance Max campaigns, give me the placements that consumed the most budget without any conversion over 90 days. Separate mobile apps from websites.
Step 5: Decide
Not all the savings you identify are equal. Tell apart what can be cut without a second thought from what requires a decision.
| Finding | Decision |
|---|---|
| Off-topic search terms | Immediate exclusion, no risk |
| App placements with no conversion | Immediate exclusion |
| Campaign with no conversion AND reliable measurement | Pause, after checking the sales cycle |
| Campaign with no conversion AND doubtful measurement | Fix measurement first, decide afterwards |
| Awareness campaign | Not to be judged on cost per conversion |
That is the most useful lesson of this exercise: in one account out of two, the first saving is not a campaign to cut, it is a measurement to fix.