Reviewed by Muze AI · 21 October 2026
Start with a list to investigate, not an automatic pause list
Wasted ad spend is advertising cost that does not contribute enough value to justify the cost. A zero-conversion row is a useful investigation lead, but it is not proof of waste: the conversion may be delayed, untracked or credited elsewhere. Start by checking measurement, then rank the largest credible problems.
Use one date range, account currency and definition of a conversion. For ecommerce, distinguish completed purchases from add-to-cart or page-view events. Keep prospecting, retargeting and brand search separate when deciding what good performance looks like.
1. Check tracking before judging performance
Compare the purchase or lead action in the ad platform with your store or CRM records. Check conversion status, value, currency and duplicate events. A campaign cannot be evaluated reliably if its purchase event stopped arriving. Native conversion diagnostics help investigate the tag; a reported ROAS number alone cannot confirm that tracking works.
2. Use a margin that includes the costs you actually pay
For a simple first-order model, break-even ROAS is 1 divided by the contribution margin before advertising. If revenue is $100 and product, fulfilment, payment and expected return costs total $60, the remaining margin is 40% and break-even ROAS is 2.5x. This is an illustrative calculation, not an industry benchmark.
A gross margin that excludes shipping, payment fees or returns can make your threshold too optimistic. Fixed overhead, tax, repeat purchases and customer lifetime value require a broader model.
3. Build a small evidence table
Export ad-level rows from Meta and search-query or keyword rows from Google where available. Record the chosen conversion action and reporting window next to the numbers. Do not add platform-attributed revenue together as if every platform reported different orders.
| Item | Spend | Purchases | What to check |
|---|---|---|---|
| Example: irrelevant search query | $180 | 0 | Does the query describe a product you do not sell? |
| Example: new prospecting ad | $90 | 0 | Has its normal conversion delay elapsed? |
| Example: retargeting ad | $220 | 3 | Check margin, frequency and attribution overlap. |
4. Separate irrelevant demand from inconclusive performance
An obviously irrelevant query is a different problem from a relevant query with no recent purchase. Review the actual search terms before proposing negatives. Excluding an entire phrase can also remove relevant demand, so inspect the scope of the change.
On Meta, investigate the path from impression to landing-page visit to checkout. Increasing frequency, falling response and repeated creative can suggest fatigue; none proves the cause alone. If clicks look healthy but purchases disappear, inspect the offer, page speed, checkout and tracking before replacing every ad.
5. Allow for conversion delay and small samples
Recent spend is visible before all resulting conversions arrive. Review the account’s historical time to conversion and use a mature comparison period. Spending one target CPA without a purchase is not a universal reason to pause an ad.
For illustration, if purchases occurred independently at an expected CPA of $50, a $50 spend window could still have zero purchases about 37% of the time under a simple Poisson model. Real auctions are more complicated; the example explains why one small window should not decide a campaign’s future.
6. Record the decision and check what happened
- Choose the highest-impact issue supported by the data.
- Write the evidence, proposed change, owner and review date.
- Review the exact entities and budget affected before applying it.
- Compare the result after enough conversion time has elapsed.
- Keep a change log so simultaneous edits do not hide what caused the result.
Where Muze fits
Muze AI helps review connected Meta and Google advertising data and turn account questions into an investigation. Start with a specific account, date range and business goal. Ask which findings are supported by the data and which need a manual tracking or website check.
A useful first request is: “Review this account for the last 30 days. Rank the largest spend concerns, show the source metrics and explain what additional evidence is needed before changing anything.” This is an example request, not a guaranteed output or savings claim.
Review your ads with Muze
See the analysis workflow and the account checks to make before acting.
Explore ad analysis