Amazon PPC mistakes we see even in established brands

We've run Amazon ads for years; what are we missing?

Usually, your blended numbers hide some of your spend from you. A blended ACoS or ROAS (return on ad spend), is basically your best spend and your worst spend averaged together into a single number.

You can have dozens of losing campaigns; however, your branded searches and a few strong exact match keywords will keep your account looking healthy on paper.

We saw this with a supplement brand doing $500K a month on Amazon. They came to us with detailed reports from their agency and flat sales for two years. They had spent $100,000 on ads, and after we audited the account, we found that $60,000 of that didn’t generate a single sale.

In the first month, we turned off that spend and didn’t change anything else. The sales remained unchanged, and the profit was up.

Another brand that was making about $1 million a month had been running broad matches for three years straight. They had no negative keywords in the account, and there were 7,000 keywords with zero sales that nobody had ever gone back to remove.

The exact match was actually the best performer; but it was not the part of the account getting the budget. We estimated that the waste was somewhere between $20,000 and $25,000 a month.

Neither of the problems appeared in a campaign-level report. To find these, you actually need to pull your spend by search term and by target and then sort the spend with zero sales.

Why do established Amazon accounts quietly break down?

Usually this is a slow process. For example, someone adds a campaign for a new product that has just launched, and then three more arrive in a seasonal push. Then, a new manager takes over the whole lot and keeps adding to what is there. In fact, nobody really owns the job of redesigning the account, so three things never get done.

1. Branded vs. generic traffic still gets mixed. When consumers type your brand name into the search bar, they are already looking for your product, which often makes those clicks cheap to convert. When you have both branded and generic terms in one campaign, the branded sales pull ACoS down and hide the losses from the generic terms.

Auto campaigns are even worse. If you don’t exclude your brand, they go after the easiest sales they can find, and very often these are brand searches.

2. Negatives are never added, and those are the exact keywords that help you to avoid spending money on searches that don’t align with your goals (read Amazon’s keywords targeting guide for more details). If you don’t add them, any irrelevant search that has already been clicked will continue to be eligible to spend your money again.

3. Auto campaigns are slowly creeping in. The whole goal for the auto campaign is to find new search terms. However, after a few years of data, there is nothing much left to find, but the budget keeps going there anyway.

What happens when hundreds of targets sit in one campaign?

We found one home goods brand that had 200 targets in one single campaign. When a setup like this occurs, it is usually because the account was built by someone with limited experience. Their agency had been reporting a blended ROAS of over 4 in one marketplace and over 5 in another, so from the outside, the account looked perfectly healthy.

The problem here is not the number of targets but the fact that all 200 of them are sharing one budget within that campaign and Amazon pushes that budget to the terms that have the most searches, instead of the ones that bring the most back. In this particular case, some of the targets were returning almost twice what the top spender did, and they still didn’t get the budget.

Amazon’s own Sponsored Products targeting guide describes the same pattern with products. In their example, one pack size ate up nearly 100% of the campaign's impressions and Amazon’s advice was to break it out into separate ad groups or campaigns.

The same goes for keywords. You should let your proven keywords have their campaigns, with their own budgets and bids. If you want to see which layout we use, make sure to check out our guide to Amazon PPC campaign architecture, which walks you through it in detail.

How do you know your auto campaigns are taking too much budget?

For the accounts that we run, auto campaigns tend to take a slice of the budget in the 10-15% range. One of the consumer brands we recently audited was nowhere near that range.

They had 5 or 6 campaigns that were taking the majority of the spend and every one of them was a discovery campaign, so together they were taking about 99% of the budget. For months, the brand had asked its agency to move its proven terms into exact-match campaigns.

Most of the budget in discovery campaigns goes to placements other than keywords and top of search. In this case, the top of the search was actually the best placement, running at 15% to 21% ACoS. However, none of the campaigns had a top-of-search bid adjustment, so Amazon had no reason to spend more there.

Our September 2026 data from 66 Amazon ad accounts shows the same gap: top of search converted at 13.1% with a 26% ACoS, while product pages converted at 5.5% with a 39% ACoS, and top of search had the lower ACoS in 82% of 63 accounts.

Auto and product targeting took 44% of spend at 31% ACoS, so loose targeting isn’t a bad thing on its own. The problem occurs when loose targeting captures almost all the spend.

Amazon’s guide to keyword targeting recommends a blend of automatic and manual targeting. Run auto first, then use the search term report to find the keywords that perform and move them into manual campaigns.

Just think of it this way. Auto does the research; manual spends the money on what the research has proven. If auto is taking more than 10-15% of the budget, it is time to start moving your proven terms to manual campaigns and then decrease the bids or budgets on auto.

Irrelevant search terms hiding in auto

Going back to the home goods brand, it was actually their team that reviewed the auto campaigns and found their ads showing for “dab spoon,” which is a smoking accessory, and something that the brand doesn’t sell at all. The term wasn’t flagged in the agency’s blended reports.

This is precisely what auto looks like when there are no negatives added. Amazon ends up choosing the matches, and if there is an irrelevant term that gets clicks, it just keeps spending until someone excludes it.

To find these, you need to go to the search term report for each of your auto campaigns, sort them by spend and read through the top 50 terms. Every term that is not relevant should go into a negative and every term that converts moves into a manual campaign.

How much wasted spend is normal in the search term report?

In our September 2026 data across 66 Amazon ad accounts, the search terms that had 10 or more clicks without any sale made up approximately 5% of the spend in a typical account. We use that number as a benchmark for what is normal waste.

You should remember that your total zero-sale spend will be higher than that because it also includes terms with fewer than 10 clicks. If your zero-sale spend is above 15 to 20% of your total spend, then you are dealing with the same problem that the accounts in this article had.

The targeting guide from Amazon suggests the auto search term report to help you select keywords for your manual campaigns, and our guide to Amazon keyword harvesting shows how we do this in practice.

How do you restructure Amazon PPC campaigns without starting over?

A yerba mate brand came to us with about 500 campaigns running. Once we turned them all off, we ran just one campaign with 5 keywords, like “yerba mate” and “organic yerba mate.” After a month passed, the product began ranking for “yerba mate," and sales and profit increased.

This is just one example though, and most accounts don’t require nearly that much, and either way, you keep your data. The search term history in your old campaigns tells you which keywords to rebuild around.

For most existing accounts we take over, we rebuild them in this order:

1. Audit spend by search term and target. First you need to start with your largest campaigns and sort them by spend with zero sales so you can see where your money is going without bringing anything back.
2. Give proven keywords their own campaigns. For those keywords that you need to rank for, we set up one campaign, one ad group, and one keyword, so this way Amazon cannot move the budget to high-volume terms, which return less.
3. Separate branded and generic traffic. You need to run them as different campaigns, give each one its own ACoS target, and make sure the brand is excluded from auto.
4. Add negatives from the zero-sale list. First, start with those terms that have 10 or more clicks without a sale, since those clearly are making the biggest waste in the account.
5. Bring auto back to 10 to 15% of the budget. Move the rest of the budget to manual campaigns built on proven keywords, and add top-of-search bid adjustments wherever the placement performs best.

You need to switch off the old campaigns once the new ones start generating the sales. At Olifant Digital, our PPC management service first starts with an audit that senior specialists run using our in-house AI platform, Olifant AI. To see how we do bidding and keyword tactics, see our Amazon PPC strategy guide along with our broader Amazon advertising strategy guide.

Frequently asked questions

How often should I audit my Amazon PPC structure?

We recommend a weekly review of your search terms and a full structure audit once a quarter. The weekly review is there to catch the irrelevant terms early, before they’ve had much time to spend.

The quarterly audit catches slower decay, such as campaigns stacking up again, branded and generic traffic getting mixed again, and auto creeping back above 10 to 15% of the budget.

It’s also worth running an extra audit every time a new person takes the account over. New campaigns usually end up stacked on top of the old ones.

What counts as too many keywords with zero sales?

It is more useful to look at the figure as a share of spend instead of just a count of keywords. While a list of 7,000 might sound alarming, what really matters is how much money these keywords took.

In our data from 66 accounts, terms that have 10 or more clicks without any sale are about 5% of the spend, and when all your zero-sale spend adds up to more than 15 to 20% of the total, the account has a structural problem instead of just a couple of bad keywords.

Should I pause underperforming campaigns or restructure them?

Restructuring is nearly always the better alternative. By pausing, you stop the waste, but you also kill the keywords within those campaigns that were actually converting.

Instead, we pull the search term data first, move the proven keywords into their campaigns, and then turn off the rest.

What share of budget should auto campaigns get?

We usually keep auto campaigns at about 10-15% of the budget because their job is to find new search terms, but when a term has proved itself, it should be spending in a manual campaign where you control the bid and placement. The only exception is a brand-new product with very little search-term history, because in the first few weeks the auto still has a lot to learn.

What to check in your account this week

First, you need to start with a search term report for your top three campaigns by spend. Then, sort it by spend with zero sales, and work out what percentage of your total spend that is. If it is more than 15 to 20%, then you have the same problem that the other established accounts had.

Then look at how much of the budget goes to auto and discovery campaigns. If it is well above 10 to 15%, then the top of the search is likely getting too little. And finally, make sure you have the negatives in place and that you are running branded and generic traffic in separate campaigns.

If your account hasn’t had a structural audit in a while, we’ll run one as part of our free marketing plan. We also back our work with a 60-day guarantee: if we don’t improve your performance, you don’t pay.