How the audit works
Every threshold is published, because an audit you can't inspect isn't an audit. These are defaults — reasonable starting points, not neutral facts. Your category and margins may justify different ones.
clicks ≥ 10 AND orders = 0 AND spend ≥ $10
Both a click floor and a dollar floor: one expensive click on a high-ticket item shouldn't hide, and five cheap clicks shouldn't cry wolf. Severity bands at $50 and $200.
same rule → recommended negatives
A search term report doesn't show your existing negative list, so this can't prove a negative is missing — it recommends terms worth excluding. Exact at ad-group level by default; phrase-level only above $50.
orders > 0 AND sales / spend < target
Each term is judged against the target for its own segment — 1x unbranded, 8x branded by default, both yours to change. Opportunity is the spend above what that target would have allowed for those sales.
term contains a brand term → branded
Matched on letters and digits only, so spacing and punctuation don't matter. Branded terms are never recommended as negatives, whatever their numbers say.
Why branded and unbranded are measured separately
Someone searching your brand name was already looking for you. That traffic converts cheaply and at high ROAS almost regardless of how well the account is run — so blending it into a single number flatters the whole account and hides how the unbranded half is doing. The unbranded half is the part that has to win customers who weren't already coming.
This isn't an argument for switching brand campaigns off. Brand defense is usually worth paying for, and a competitor bidding on your name will happily take the click if you don't. It's an argument for knowing which number is which — particularly if the account is managed on a percentage of ad spend, where branded traffic earns a fee for demand you generated somewhere else.
The defaults are deliberately lopsided. Unbranded sits at 1x — the point where the ad spend exceeds the sales it produced — because that is the one reading nobody argues with, and a stricter default would flag half of every healthy account and make this tool look like it was built to alarm you. Branded sits at 8x because brand traffic that cannot clear a high bar is worth a question. Both are starting points, not rules: raise the unbranded target to your real contribution margin and the findings list gets longer and more honest.
When the audit refuses to answer
A verdict drawn from too little data, or from a report that didn't export cleanly, is worse than no verdict — it's a confident wrong answer you might act on or forward to someone. So the audit withholds one when any of these is true:
under 25 terms · under $500 spend
Findings are still listed, because they may be worth a look. The scorecard verdict isn't, because at that size one row moves it.
no term with 10+ clicks and $10+ spend
Nothing statistically meaningful was testable. That is not the same as a clean account, and the tool won't report it as one.
zero sales account-wide · no orders column
Usually a date range too recent for attribution, or a column that didn't export. Findings are suppressed entirely — every term would otherwise look like waste.
In every one of those cases the page says which check tripped and what to re-pull, and the draft email is withheld — that one gets forwarded, so it's the last place an unsupported figure should appear.
What this audit will not tell you
- Whether a term is genuinely bad. A term with no orders may be doing real work — brand defense, a halo on other ASINs, a long consideration cycle the attribution window misses. The tool sees a 14-day attribution window, not your strategy.
- Whether your agency is bad. It reports the account's numbers. Whether those numbers reflect neglect, a deliberate call, or a bad quarter is a conversation to have with whoever manages the account.
- Anything about the terms below the floors. Sub-10-click terms are statistically thin. They're counted in your total spend, never flagged as findings.
Where to get the file
Amazon. In the Amazon Ads console: Measurement & Reporting → Sponsored ads reports → Search term, report period Last 30 days, time unit Daily, CSV or Excel. There is a custom date range that reaches further back, but it is unreliable on this report — the 30-day preset is the one that works, and it is what the audit is calibrated for. Amazon walkthrough, including the wording to send whoever manages the account if you don't have console access yourself.
Walmart. In Walmart Connect Ad Center: Reports → On-demand → Request report → Item Keyword Performance, attribution window 14 days, grouping Daily, the last 60 days. Walmart doesn't record in the file which window you chose, so pick 14: it is what the audit is calibrated for. Walmart walkthrough, with the same wording for whoever manages the account.