Nobody wakes up planning to spend 400 dollars on an ad set with zero conversions. It happens because the account has 60 line items and the dead ones don't announce themselves. They just sit there, quietly billing.

The free Claude skill wasted-spend-finder flags every ad set and ad that has spent past a judgment threshold with nothing to show, groups the results, and totals what the leak becomes over the next 30 days if you leave it running.

This post covers the thresholds it uses, why it separates a dead ad from a dead ad set, and how to run your first audit.

Key takeaways

• 2 default triggers: spend at 2x your target CPA with 0 conversions, or spend at 3x target CPA with ROAS below breakeven.

• Everything gets sorted into CUT, BLEEDING, or WATCH, so you know what to pause now and what to keep an eye on.

• Ad sets still in Meta's learning phase are excluded by design. They aren't waste, they're unfinished.

• The audit runs at ad level and ad set level, because 1 dead ad inside a healthy set is a creative problem, not a pause the set problem.

The 2 thresholds that define waste

Waste needs a definition before it can be found, and the skill uses 2 that you can edit to your margins.

The zero result flag fires when an item has spent at least 2x your target CPA and produced no conversions at all. If you're paying 40 dollars per purchase and something has burned 80 with nothing back, it has had its chance.

The unprofitable flag fires when an item has spent at least 3x target CPA and its ROAS sits below breakeven. This one is converting, which is exactly why it survives audits. It looks alive. It's just losing money on every sale.

Both thresholds need your real numbers, so the skill asks for target CPA or breakeven ROAS before it flags anything, and it won't guess if you don't know them.

CUT, BLEEDING, WATCH

CUT BLEEDING WATCH

Cut is a zero result item past threshold that has already exited learning. There's no case left to make for it.

Bleeding is converting below breakeven for 14 days. Slower to kill, worth a real decision rather than reflex.

Watch is under the threshold but trending toward it, with the trend stated so you can see which direction it's moving.

Then comes the number that makes people act: the total monthly run rate of everything flagged. Seeing 190 dollars of daily waste written out as what it costs over the next 30 days changes the conversation faster than a list of underperformers.

Why ad level matters as much as ad set level

This is the part most spend audits skip. If you only look at ad sets, a single dead creative inside an otherwise healthy set is invisible: the set averages out fine, so nothing gets flagged, and the dead ad keeps taking its share of delivery.

The skill pulls both levels for exactly that reason. A dead ad inside a healthy ad set gets routed to the creative fatigue scanner as a creative problem. A set where everything is dead gets staged for a pause. Same audit, 2 very different fixes.

A dead ad is a creative problem, not a budget problem

Pausing the loser is half the job. Something has to take its slot. Vibelets generates on-brand product ads from your store in minutes, so the replacement is ready when you cut.

Start free with Vibelets →

How to set it up in Claude

Download the skill file from the skill's Google Drive folder, free, no email gate. Add it to Claude by uploading it as a skill under Settings, or attach the file to a Claude project or chat. Anthropic's Help Center guide to skills covers the upload path.

Export from Ads Manager with name, delivery status, spend, results, cost per result, and purchase ROAS for both the last 7 and last 14 days. The delivery status column is what lets the skill exclude learning phase ad sets on its own, so don't drop it.

Paste it in, give it your target CPA, and you'll get the pause list. Nothing executes without your explicit yes on the specific items, and in paste mode you're the one making the changes in Ads Manager anyway.

Conclusion

Most accounts don't have a traffic problem. They have a leak that stopped being visible months ago. Run the audit, look at the 30 day projection, and cut the items you can't defend.

Find the leak before it funds another month

Free Claude skill, yours in 1 click.

Get the skill here →

Frequently asked questions

How much spend is enough to call an ad set wasted?

The defaults are spend at 2x your target CPA with 0 conversions, or spend at 3x target CPA with ROAS below breakeven. Both are editable to your margins, and both need your real target CPA, which the skill asks for rather than guessing.

Why does the skill ignore ad sets still in the learning phase?

Because they aren't waste, they're unfinished. An ad set needs roughly 50 optimization events in a 7 day window before delivery stabilizes, so judging one mid learning kills things that never got a fair read.

Should I pause the whole ad set or just the ad?

It depends which level is failing. A dead ad inside an otherwise healthy ad set is a creative problem, so replace the creative. An ad set where everything is flagged is a structural problem, so pause the set.

Does the skill pause anything automatically?

No. It presents a pause list and waits for your yes on the specific items. In paste mode you execute in Ads Manager yourself; in live mode Claude only acts on what you approved.

How often should I run a wasted spend audit?

Weekly or every 2 weeks, ideally alongside your scale hold kill pass so cuts and reallocations happen in the same sitting rather than as separate impulses.

Sources

1. Meta Business Help Center: About the learning phase

2. Meta Business Help Center: Significant edits and the learning phase

3. Claude Help Center: Use skills in Claude