Where ad spend waste hides (and how to measure it)
The eight places ad spend waste usually hides in Google Ads and Meta, and how to size it honestly: exposed spend, estimated savings and measured savings.
On this page
Most ad spend waste isn't one big mistake. It's spread across search terms nobody reviewed, placements nobody chose, default settings, broken tracking and budgets running at the wrong pace.
Eight places waste hides
1. Unreviewed search termsBroad match and AI Max expand your queries. Sort the search terms report by cost, filter to zero conversions, and read it: job seekers, students, DIY, "free", competitor brands.
2. PMax inventory you didn't pickPMax can spend across Search, Shopping, Display, YouTube, Discover and Gmail. Check the channel and placement data, and use account-level exclusions where they fit.
3. Network and location defaultsDisplay Network on in a Search campaign, Search Partners never evaluated, or "Presence or interest" location targeting.
4. Hours when nobody answersLead-gen spend at 3 a.m. buys leads that cool off before anyone calls back. Compare spend by hour with when you can respond.
5. Tracking that teaches the wrong lessonThe most expensive kind, and rarely labeled waste. Add-to-cart counted as primary, or one purchase counted twice, and Smart Bidding chases the wrong signal while ROAS looks great.
6. Overlapping campaignsTwo campaigns on the same queries, or PMax bidding on your exact-match terms, blur what works and can raise what you pay.
7. Broken landing pagesA 404 behind a sitelink, a redirect that drops the click ID, a slow mobile page. Every click is paid in full.
8. Pace, not just amountA budget gone by the 20th starves the last ten days. Forecast to month-end, not month-to-date.
The three numbers, kept apart
| Number | What it is | What it isn't |
|---|---|---|
| Exposed spend | Spend that sat on a problem in a past period | Money you would have kept |
| Estimated savings | A forward-looking estimate of what fixing it could save | A result |
| Measured savings | An outcome observed after the change, with a baseline and method | Lower spend alone |
Worked example: sizing it honestly
Worked exampleIllustrative example; the numbers are not real account data
An e-commerce account spends $20,000 a month. A review of the last 30 days finds:
| Finding | Exposed spend |
|---|---|
| Irrelevant search terms, no conversions | $1,400 |
| PMax spend on app placements, no conversions | $1,100 |
| Overnight spend, no conversions | $600 |
| Naive total | $3,100 |
Deduplicate
$300 of the overnight spend was on the same app placements. Counted once, exposed spend is $2,800, not $3,100.Estimate, conservatively
Part of the budget flows elsewhere rather than disappearing. Assuming about 70% becomes real savings while conversions hold gives an estimated saving of roughly $1,960 a month. Label it as an estimate.Measure
A month later, compare against the baseline, adjusting for seasonality. Spend down $1,700 with flat conversions is an observed improvement you can report with its period and method. If conversions also fell, it may not be a saving at all. A truly measured, causal figure needs a holdout or platform experiment.
- Naive total$3,100
- Deduplicated$2,800
- Estimated$1,960
- Observed, conversions flat$1,700
The gap between $3,100 and the number you can defend is where trust in waste reports gets lost.
Make it a habit
| Cadence | Check |
|---|---|
| Daily | Silent conversion tracking, spend spikes, broken pages |
| Weekly | Search terms, PMax placements, budget forecast |
| Monthly | Settings and auto-applied changes, overlap, estimated vs. measured results |
Do
- Deduplicate findings before totaling
- Label estimates as estimates
- Report measured results with their period, baseline and method
Don't
- Report exposed spend as money saved
- Declare success because spend went down
- Mix estimates and measured results in one total
How to keep the last two apart: Estimated vs. measured savings.