Amazon TACoS benchmarks: what 14 accounts we've managed actually ran (156 months of data)

Search for a TACoS benchmark and the same band comes back: 10% to 15% is healthy. One benchmark table puts the “2026 Platform Average” for TACoS at “10% to 15%”, with under 10% as outperforming and over 20% as needing attention¹. Another says “A healthy account-level TACOS is 10-15% of total Amazon revenue"², and explains that it took the band from the first page and blended its other figures from several reports, “not a single survey"². Neither page says how many accounts, which marketplaces or which months the TACoS band was measured on.

We hold the monthly sales and advertising figures for the accounts we’ve managed, so we measured it. This page sets out what we found, how we worked it out and what we left out, so you can judge it for yourself. If you want to set a target for your own brand rather than compare against other people’s, start with our guide to a good TACoS; it shows how to work out a ceiling from your margin.

What we measured

TACoS is ad spend divided by total sales. For each account and each month we took two figures from the sheet we keep for that account:

• Ad spend, from Amazon Ads.
• Ordered Product Sales, the total sales figure in Seller Central’s Business Reports, which counts every sale, whether or not an ad was clicked.

That gave 14 accounts and 156 account-months, between November 2024 and April 2026. Accounts are in the sample for between 4 and 17 months, depending on when we started keeping the sheet and how long we ran the account. The sheets for four accounts stop before September 2025, and three of those four are the three highest figures here: those three were launches, on short stretches of data. Two sheets show their figures in dollars; TACoS is a ratio, so currency doesn’t change it. One sheet labels sales in pounds and its ad figures in dollars; we treated that as a labelling slip, because the sheet’s own TACoS sums treat them as one currency.

We worked out each account’s median month, so that one unusual month can’t drag an account’s figure, and then compared the 14 accounts with each other. Every account counts once, whatever its size.

What we left out, and why

A script applied the same rules to every account and printed each month it removed. It took out 22 account-months for five reasons:

• Months with no ad spend recorded (12). Nine were one account’s months where the sheet records no ad spend. TACoS is meaningless without ads.
• Months where ad spend was under a tenth of the account’s usual level (3). Ads were effectively paused, so the month says nothing about running ads.
• A month whose sales figure was clearly wrong (1). Sales divided by units gave an average price of about a fifth of that account’s usual price.
• Months below £1,000 of sales, or from an account left with fewer than four usable months (5). That removed one small account entirely.
• A month whose ad figures were an exact repeat of the month before (1). Ad spend and ad sales matched to the penny, which points to a copying slip.

We also left out four sheets that weren’t usable: one that was an exact copy of another account’s sheet, one second copy of an account that disagreed with its main sheet, and two that were empty.

The result: a median of 9.8%, and a wide spread

Seven of the 14 accounts ran under 10%, three ran between 10% and 15%, one ran between 15% and 20%, and three ran above 20%. Seven accounts sit in the range the published tables label outperforming, three in the range labelled average, and three above the 20% level flagged as a concern¹². So most of our accounts were not in the “average” range, and most of those outside it were below it.

The three accounts above 20% were launches. A launch buys its first sales, reviews and rank, so ads cost far more than the product can carry for long. We treat that as a temporary phase with a budget and an end date, not as an acceptable level long term. Leave those three out and the median falls from 9.8% to 8.3%, the middle half runs 6.1% to 11.5%, and accounts run from 2.6% to 18.9%.

What separates a low TACoS from a high one

TACoS can be split into two parts: how efficient the ads are, and how much of the business runs through them.

Grouping the accounts by their median TACoS:

The two parts differ by about the same amount. Between the low group and the high group, the median ad share is about 1.8 times higher (32% to 58%) and so is the median ACoS (21% to 38%). The pattern isn’t the same all the way up: ad share rises through all three groups (32%, 47%, 58%), while ACoS is similar in the two lower groups (21% and 23%) and only stands out above 15%. Fourteen accounts can’t show what caused the difference. In our experience, ad share falls as rank, reviews and repeat buying grow, and ACoS rises with click prices and weak listings, but that is our reading, not something this data measures.

We don’t publish a TACoS figure by category, because fourteen accounts is too few to split by category honestly.

Why a good TACoS depends on the business

These 14 accounts can’t give you a figure for your category, but the type of business changes what you can afford. In our experience, three things matter most:

• Margin. The more of each sale is left after costs, the more you can spend on ads and still make a profit. Electricals tend to run on thinner margins, so they need a lower TACoS.
• Repeat buying. If customers come back without another ad, the first sale can cost more, because it isn’t the only sale. That is one reason supplements can carry a higher TACoS.
• Order value. A larger order leaves more pounds to pay for each click, even at the same percentage margin.

This is the rough guide we use in our Profit X-Ray for a product past its launch. It is drawn from published agency ranges and checked against the accounts we’ve managed. It isn’t an Amazon figure, and it isn’t measured from the 14 accounts above.

One published guide’s table puts electronics at 7-11% at maturity, health and household at 12-18% and beauty at 14-20% (its summary gives 7-13% for electronics and 14-20% for supplements and beauty), and says launches can run 25-40%, sometimes up to 50%, in the first one to three months².

The same TACoS can leave room to spend more in one business and be too high in another:

So use the tables on this page to see whether your number is unusual, not whether it is good. Our guide to a good TACoS shows how to set your own ceiling.

How much TACoS moves from month to month

For each account we measured how far TACoS moved from one month to the next, and took the median.

• Nine of the 14 accounts moved by less than 1.5 points a month. For them, a two-point jump in one month is worth looking into.
• The four accounts above 15% moved by 2.4 to 16 points. One month on its own told us very little about them; three months did.

If your TACoS is above 15%, track a rolling three-month figure rather than reacting to each month.

What happens in November and December

We compared November and December with the other months, using only the 7 accounts with a year or more of data. In the shorter samples, November and December are often the account’s first months, so a comparison would mostly show how an account started, not the season.

• TACoS was lower in those months in 4 accounts and higher in 3.
• In 6 of the 7 the gap was under 2 points.

So we found no clear Christmas effect in either direction. Seven accounts is too few to rule one out.

Did TACoS fall over time?

We compared each account’s first three months in the sample with its last three, using the 11 accounts with at least six months, so the two periods don’t overlap. Pooling each period’s ad spend and sales, TACoS fell in 5 accounts and rose in 6. Total sales grew in all 5 where TACoS fell, and in 4 of the 6 where it rose. Using each period’s median month instead, one account flips, giving 6 and 5, so the split is close to even either way. So a rising TACoS did not by itself mean trouble, which fits what our guide to a good TACoS says about reading the trend. These periods also fall in different seasons, so treat this as a rough read.

All 14 accounts

Accounts are lettered in order of median TACoS. We don’t name them, their categories or their sales.

Ad share and ACoS are each account’s median month, measured where the sheet recorded ad-attributed sales below total sales.

What this data can’t tell you

• It is 14 accounts from one agency. They are accounts we’ve managed, not a random sample of Amazon sellers.
• It mixes stages. The three accounts above 20% were launches; that comes from our knowledge of those accounts, not from the sheets. We haven’t labelled the stage of the others.
• It mixes categories and sizes, from small accounts to some of the largest we run.
• We did not remove VAT from sales. We used Ordered Product Sales as our sheets hold it, while Amazon Ads reports ad-attributed sales without VAT. Where the sales figure includes UK VAT at 20%, TACoS and ad share read up to a sixth lower than on a VAT-free basis. Compare your own figure on the same basis.
• Organic sales are an estimate. Amazon credits a sale to an ad only when a shopper clicked and bought within its attribution window, so “not credited to an ad” isn’t quite the same as “would have happened without ads”.

A benchmark is useful for one thing: telling you whether a number is unusual. In our data, 10 of the 14 accounts ran between about 5% and 19%, and the highest three were launches. Whether your own figure is good depends on your profit: your margin, your type of product and whether more of each month’s sales are arriving without an ad. This data doesn’t measure your margin.