Amazon's new AI automatically adds up to four keywords or targets to manual campaigns every 12-24 hours. A home-goods brand saw ACOS jump from 18% to 26% after 62 auto-added targets appeared, many irrelevant or low-margin. The hosts say treat it like a junior analyst: test it in isolation, audit weekly, and turn it off if it creates noise.
This is the High Voltage Business Builders Podcast, daily intelligence for serious e-commerce portfolio builders across Amazon, TikTok Shop, Shopify, Walmart, and every channel that moves the needle. Neil Twa and his Voltage team, all day, every day since 2012. Let's get into it.
Who actually controls your Amazon ads right now? You or a checkbox you scrolled past last week? There's a new targeting strategy beta inside manual Sponsored Products campaigns. One option is easy to miss. Turn it on, and Amazon's AI starts adding up to 4 new keywords and product targets every 12 to 24 hours. These targets are at or below your current bids and are based on your existing campaign data. That sounds helpful, and it might be. Or, it might be the slow, quiet start of Amazon deciding where your ad dollars go. I am breaking down what this beta actually does, what you need to watch, and how to stay in the driver's seat. Look, I want to be clear about something before we get into the mechanics here. I am not against AI. We use it across our 30 brands every single day. But there is a massive difference between AI that works for you and AI that works for Amazon's revenue goals. Those 2 things are not always the same. This new targeting strategy toggle inside manual Sponsored Products campaigns is framed as a convenience feature. Amazon's AI reads your existing campaign data, spots what it thinks are gaps, and adds up to 4 new keywords or product targets every 12 to 24 hours. It does this at or below your current bids, automatically. Here is what most operators will do: they will see it, think it is free optimization, check the box, and walk away. For the first week, it will probably look fine, maybe even good. ACoS ticks down a little, impressions go up, it feels like a win. Then a month in, you look at your search term report and see 40 targets in that campaign you never approved. Some of them are bleeding, some of them are cannibalizing your other campaigns. You have no idea when it started because you were not watching. This is the IDQ problem applied to ads. In the book, we talk about not touching a new listing for 7 to 21 days because the algorithm needs clean data to calibrate. The same principle applies here. When you layer automated targeting on top of a manual campaign without a review system, you are feeding the algorithm noise and calling it optimization. The operators who get burned by features like this are not dumb. They are busy. They trust the platform a little too much. They find out 3 months later when margin has quietly evaporated, 30% of your ad spend on a typical account is already going to targets that will never convert profitably. I see it constantly across our portfolio. This beta, unchecked, could accelerate that number quickly. Let me give you a real picture of how this plays out. I was talking with an operator running a home goods brand, doing around $40,000 to $60,000 a month in revenue. Solid margins. Well-structured campaigns. She had been running clean manual Sponsored Products campaigns for about 18 months. She knew her top-performing keywords cold and had negative match lists dialed in. She sees the new targeting beta, reads the description, and thinks, Amazon knows my account better than anyone. This could save me time. She turns it on across 3 of her campaigns and does not change anything else. 4 weeks later, she is on a call with me, and her ACOS has climbed from 18% to 26%. She is confused because she has not touched anything. We pull the search term report. There are 62 new targets Amazon added across those 3 campaigns in 30 days.
62.
Some of them are adjacent categories she specifically excluded because she tested them 18 months ago And they do not convert for her product. Amazon did not know that. Amazon saw traffic and made a guess. Here is the thing that stings. Some of those targets were actually decent. Maybe 8 or 10 of them showed real promise, but they were buried inside 62 auto-added entries with no structure, no negative match discipline, and no bid logic tied to her actual margin targets. The fix was not complicated. We turned the beta off. We pulled the 12 targets worth keeping, built them into a new structured campaign with proper bids and negatives, and killed the rest. But she lost 4 weeks of clean data and a chunk of margin she did not need to lose. That is the real cost of letting Amazon figure it out. The operators who win with AI-assisted targeting are the ones who treat it like a junior analyst. You review the work, approve what fits, and cut what does not. You do not hand over the account and go to lunch.
While saving 17 hours a week doing it? That's exactly what we built Voltage Business Builders for. It combines the power of experience and being in the right room with the power of AI to compress time and increase your business profits. More than AI for the sake of using AI. It's a room full of senior sellers with over 15 years in e-com showing you the ropes of AI, business building, and creating generational wealth through business. Visit voltage dm.com to see what's inside the Voltage Business Builders Room.
3 moves, right now. Whether you're running 1 SKU or 100, move 1: go find the beta. Open your manual Sponsored Products campaigns in Amazon Ads and look for the targeting strategy setting. It may not be in every account yet, but it is rolling out. If you see it, do not touch it until you have a review process in place. Not because it is evil, Because you need to know what it is doing before it does it at scale. Boring advice. Correct advice. Move 2: Build a weekly search term audit into your calendar. This is not optional anymore. If Amazon is going to add targets to your campaigns automatically, you need a standing review every 7 days. Pull the search term report. Sort by spend. Anything over $5 with 0 conversions gets a negative match added or gets cut. This one habit will save you more margin than any optimization feature Amazon ever ships. Move 3: If you decide to test the beta, isolate it. Pick one campaign. One. Not your top revenue driver, not your brand defense campaign. Pick a mid-tier campaign on a product you understand well. Let it run for 14 days. Then pull the data and decide if the targets Amazon added are ones you would have chosen yourself. If yes, great, scale the test. If no, you have your answer without blowing up your whole account. I know nobody wants to hear to test small and review manually. Everyone wants the checkbox that fixes everything while they sleep. I get it. I would love that too. But that checkbox does not exist. What exists are operators who watch their numbers and operators who wonder what happened to their margin. Pick which one you want to be. If any of this resonates with you, you are likely running Amazon Ads with more guesswork than you realize. New betas, changing targets, and spending you did not approve are all part of the chaos. You have the same 24 hours to manage it all. Most operators are already overwhelmed with tabs— ads over here, listings over there, inventory in another window, and pricing in a spreadsheet somewhere else. Then AI features start promising to help, but bad data in leads to bad calls out. You do not save time, you make costly mistakes faster. That is not freedom, that is chaos with no one steering. Here is what works. Cayman Data pulls your live Amazon numbers into one clear picture. Ads, listings, sales, inventory, all of it. You see what is working and what is costing you money. No more spreadsheets that consume your Sunday night. You stay in charge and see the reason before you say yes. Nothing runs without your approval. That is how it should work. You are the operator. The data should serve you, not the other way around. That level of account review used to take hours every week, chasing reports, cross-referencing tabs, and trying to determine if that spending spike was a problem or a fluke. Cayman Data cuts that down with one live connection to your account. That is how Voltage helps operators save time, protect margins, and grow without losing control of their brands. If you wanna see how it works, head to voltage dm.com. Discover what is possible when your numbers are finally in one place. Thanks for spending part of your day with me on the High Voltage Business Builders Podcast. We will see you back here tomorrow. Until then, stay high voltage.
EP405: Amazon put eBay, Shopify, TikTok, and Walmart orders inside Seller Central. What that costs single-channel Amazon FBA brands on a $1,552,793.82 August book.
