Free Amazon PPC audit: how much of your ad spend is working?

First, one number: the share of your Sponsored Products keyword and target spend that went to keywords and targets selling at or under your target ACoS. The rest of that spend splits three ways: selling over your target, no orders after the click threshold, and too early to judge. Each part shows its dollars and the rows behind it.

Then the rules. Fourteen rules read the Sponsored Products tab and the search term tab, and each one says it passed, what it found (with the dollars and the rows) or what your file was missing. Rules that found something sit in three groups so different kinds of dollars aren't ranked together: spend to look at (largest first), room to grow (sales and budget) and setup to check. They cover:

• Keywords, targets, product ads and search terms that reached the click threshold with no orders.
• Keywords and targets selling above your target ACoS, with a lower bid for each (enter a target to run it).
• Campaigns spending about all of their daily budget, and selling search terms that are not exact keywords yet.
• Words that only show up in searches with no orders, negatives that block your own keywords, and enabled items under paused campaigns.
• Duplicate keywords, bids far above what clicks cost, and placement adjustments.

The headline is a share of your dollars because points out of 100 need weights, and a bulk file can't back them up. It has no margins, no organic sales and no goals, and there's no fair pool of other accounts to grade yours against. Your own spend, split by what it bought, needs none of that. Every dollar sits in exactly one group, so the groups add up to 100% and you can open the rows behind each one. Each rule is a plain sentence you can check against your file.

The upload file holds only the suggestions you leave ticked: negative exact keywords in the ad group where a search ran without orders, negative product targets for product pages, campaign negative phrases if you tick them, and lower bids. It never creates campaigns or keywords, never pauses anything and never touches budgets.

Bids only go down, by at most 30% at a time, and only on keywords and targets with at least two orders. A search that sells is never negated in its own campaign, even when it also runs as an exact keyword somewhere else: those rows stay in the findings for you to judge.

Uploading the file changes your live campaigns, so read every row first. Upload it at Bulk operations, then check Amazon's processing report for anything it rejected.

• Thin data. Twenty clicks with no orders is a rule of thumb, not proof. Raise the threshold for products that convert slowly or cost a lot.
• The last few days. Orders can still be credited to recent clicks after you download the file, so a fresh search term may not be finished.
• Relevant searches. If a shopper searched for exactly what you sell and didn't buy, look at price, listing and bid before you block it.
• Bids on products whose margins you know better than the file does. Work out what you can afford with the breakeven ACoS calculator and the target CPC calculator first.

For a closer look at search terms on their own, the Search Term Report Analyzer and the PPC N-Gram Analyzer read the search term report instead of the bulk file.

Your file never leaves your browser. The page reads it in a background worker on your device, runs the checks there and builds the upload file there too. There's no account and no upload, and closing the tab clears it.

Files up to 150 MB work, which is about a million and a half rows. A file near that size can take half a minute and over 1 GB of your computer's memory. Past it, download a shorter date range with "Campaign items with zero impressions" unticked, or leave Sponsored Brands and Sponsored Display out. The page finds the tabs by their names and columns, so Amazon can reorder them without breaking it, but the console needs to be in English.