Amazon PPC A→Z · Lesson 46 · 2:37
How to judge an AI PPC tool Test any PPC tool, Amazon's own included, against similar products it didn't touch in the same weeks, and judge it on profit after ads, not on the ACoS its own report shows.
The video plays from YouTube (no cookies until you press play). 1 Split similar products in two, several on each side: the tool (or Amazon's Apply button) runs one half, the other stays as it is. 2 After 4+ weeks, from your own reports: units × profit before ads − ad spend − tool fee, per half. Wait until late sales are in. 3 The tool earned only the difference. Ask it why for 3 of its changes: no clear reason, don't scale it. Judging a tool by its own before/after report, and by ACoS instead of profit after ads: the season, prices and competitors get counted as the tool's work.
0:00 The tool's report vs an untouched product 0:25 Tools that change your ads for you 0:42 The water bottle example 1:07 The test, side by side 1:28 What the tool really earned 1:42 The common mistake 1:49 Amazon's recommendations: good and watch out 2:06 Do this today Show Hide 0:00 Your AI PPC tool says it's working. How do you check? 0:04 Its report: ACoS down from thirty to twenty five percent, and five hundred and fifty dollars more profit. 0:11 But a similar bottle it never touched made six hundred more, in the same weeks. 0:17 So the tool earned about nothing. Here's how to test it. 0:20 Lesson forty six: how to judge an AI PPC tool. 0:25 Amazon is rolling out its own recommendations in the US: apply, edit, or skip. AI bid tools make the same promise. 0:34 Each one reports its own before and after. But that mixes in everything else that moved: the season, prices, competitors. 0:42 Here's the example. Two similar steel water bottles, A and B. Forty dollars each, with fifteen dollars left before ads. 0:51 In the four weeks before, each sold two hundred units and spent fifteen hundred dollars on ads. 0:57 Thirty percent ACoS, and fifteen hundred dollars of profit after ads. 1:02 Then a tool runs bottle A for four weeks. Bottle B stays as it is. 1:07 A sells two hundred and thirty units and spends fourteen hundred. Twenty five percent ACoS. Two thousand and fifty profit: five hundred and fifty more. 1:17 B, untouched, sells two hundred and forty, and spends fifteen hundred. Also twenty five percent. Two thousand one hundred: six hundred more. 1:28 The tool earned only the difference. Five hundred and fifty, minus six hundred: minus fifty dollars. About zero. 1:35 The season lowered ACoS on both bottles. The tool's report counted it as its own work. 1:42 The common mistake: judging a tool by its own report, and by ACoS instead of profit after ads. 1:49 Use the same test for every tool. Amazon's recommendations are free, and each one comes with a reason you can edit or skip. 1:57 But Amazon also sells the clicks, and it hasn't published results. Other tools need your account data, and their fee comes off the profit. 2:06 So: split similar products in two, several on each side. Let the tool run one half for four weeks or more. 2:14 Then, from your own reports, compare profit after ads, once the late sales are in. 2:20 And ask the tool why, for three of its changes. No clear reason? Don't scale it. 2:26 Judge the tool by the difference, not by its report. 2:30 Next lesson: Amazon Ads Agent, and what its AI does for your campaigns. See you there. Not affiliated with Amazon. Numbers in this video are a worked example. Published 27 October 2026.