TACoS, total advertising cost of sale, is advertising spend divided by total sales, ads and organic together.
Picture the quarterly report that ends an agency relationship. Over three months, ACoS has fallen from 38% to 22%. Every slide says so. What no slide says is that total monthly revenue fell over the same period, because the way ACoS was lowered was to pull spend off every discovery campaign and concentrate it on the brand name. The agency did exactly what it was asked to do. The number it was asked to move was the wrong number. (The figures in that scenario are invented to show the mechanism. The mechanism is not invented at all.)
This kind of thing is rarely dishonest. ACoS is the metric Amazon puts in front of you, it is the one every tool reports on, and it is genuinely useful for the question it answers. The problem is that people use it to answer a different question, and it cannot.
What each number is actually measuring
ACoS is advertising spend divided by advertising-attributed sales. It answers: of the sales my ads got credit for, what proportion did the ads cost? That is a real question with a real use. It is the right lens on a single keyword, a single ad group, a single targeting decision.
TACoS — total advertising cost of sale — is advertising spend divided by total sales, ads and organic together. It answers a different question: what proportion of everything this business sold went on buying advertising?
The difference is the denominator, and the denominator is the whole argument. ACoS has a denominator you can shrink or grow by changing what you target. TACoS has a denominator that is simply the business. You cannot make TACoS look better by moving spend onto traffic you were already getting, because that traffic is already in the denominator. That is the entire reason the number is worth computing.
The branded-keyword trap, in numbers
Take a seller doing £40,000 a month in total sales, spending £8,000 on ads, with £24,000 of that £40,000 attributed to advertising. ACoS is 33%. TACoS is 20%.
Now someone is told to get ACoS down. They cut the discovery campaigns and the non-brand exact terms with mid-range efficiency, and push the remaining budget onto the brand name and a handful of terms the listing already ranks well on organically. Spend drops to £5,000. Attributed sales drop to £20,000, so ACoS is now 25% — a good-looking improvement. But total sales have fallen to £33,000, because the traffic that used to come from the terms nobody is bidding on any more has gone. TACoS is now 15%, which also looks like an improvement.
So TACoS is not magic either. It fell too. What it did do is force the third number into the conversation: revenue went from £40,000 to £33,000. You cannot read TACoS without revenue beside it, and that constraint is the point of it. Nobody has ever presented a slide showing ACoS down and revenue down on the same axis, because ACoS lets you leave revenue off the page. TACoS does not, because revenue is inside it.
Those figures are an illustration built to make the mechanism visible, not a case study. The shape is what matters, not the specific pounds.
How to compute it from your own reports
TACoS is not a native metric in Seller Central or in the advertising console. You build it from two figures that live in different places, which is most of the reason people do not bother.
- The numerator: total advertising spend. Not just Sponsored Products. If you are running Sponsored Brands, Sponsored Display or DSP against the same catalogue, all of it belongs in the numerator, or the number is not a total.
- The denominator: total sales for the same period and the same marketplace. For sellers, that is ordered product sales from the Business Reports section of Seller Central. For vendors, it is shipped or ordered revenue from your Vendor Central reporting — and you should decide which of the two you are using and stay on it, because they are different numbers and the gap between them is not small.
Divide one by the other, or let the TACoS calculator do it. That is the whole calculation. There are three things worth knowing about it before you start quoting the result to people.
The dates do not line up perfectly. Ad spend is recorded on the day the click happened. Ad-attributed sales are credited back to the click date, within an attribution window. Total sales in Business Reports sit on the order date. Over a week these differences are visible; over a month they mostly wash out. Use monthly figures for anything you are going to make a decision on, and treat a weekly TACoS as a trend line rather than a measurement.
Returns are not in it. Ordered product sales reflects what was ordered, not what was kept. In a category with a high return rate, TACoS understates what advertising is really costing you as a share of money you get to keep. If your returns run materially above the rest of your catalogue, compute a second version against net sales and use that one internally.
Compute it per marketplace and per product family, not just per account. An account-level TACoS on a business selling four product lines across three marketplaces is an average of things that have nothing to do with each other. One launching product at a deliberately high TACoS and three mature ones will produce an account number that describes none of them.
There is no good TACoS, only a good trend
People want a benchmark. We are not going to give you one, because a target TACoS is a function of your gross margin, your stage and your category, and any single figure quoted as an industry standard is a figure someone made up. What we can give you is the three shapes we look for, which is what we use internally when reviewing an account.
At launch, TACoS should be high and you should have decided in advance how high. You are buying rank and reviews, and almost all sales are paid sales, so TACoS and ACoS will be close together. The thing to watch is not the level but whether the gap between them is opening. When organic sales start arriving, ACoS stays roughly where it is and TACoS falls away from it. That divergence is the first evidence the launch is working, and it is visible well before the revenue line makes it obvious.
On a scaling product, expect TACoS to be flat or drifting down while revenue climbs. Flat TACoS with rising revenue means advertising is buying growth proportionally — you are spending more because you are selling more, which is the deal you signed up for. Falling TACoS with rising revenue means organic is compounding underneath the ads, which is the best outcome available.
On a mature listing, TACoS should be low, stable and boring. A mature product whose TACoS starts creeping up month on month while revenue holds flat is telling you something specific: you are paying more to stand still. That usually means a competitor has entered, your organic rank has slipped, or your conversion rate has fallen — and the fix for all three is on the listing and the offer, not in the bid file.
When rising TACoS is the right answer
TACoS going up is not automatically bad, and treating it as automatically bad is the failure mode of people who have just discovered the metric. If TACoS rises from 14% to 18% while revenue rises 40%, you have bought growth. Whether that was a good trade depends entirely on your gross margin: at a 55% margin it very likely was, at a 22% margin it very likely was not. This is a judgement call about your own economics, not a rule anyone can hand you.
The version that is genuinely a problem is TACoS rising while revenue is flat or falling. That is the only combination that always needs investigating, and it is the one an ACoS-only report will hide from you completely.
There is one more reason to trust the total over the attributed figure. When you run Sponsored Products, Sponsored Brands and Sponsored Display against the same products, a single order can be credited in more than one of those reports. Add the attributed-sales columns together and you can produce a number larger than what you actually sold. Total sales cannot do that. It is one order, counted once.
What to do on Monday morning
- Open a spreadsheet with four columns: month, total ad spend, total sales, TACoS. Nothing else.
- Fill in the last twelve months from your advertising reports and your Business Reports. It is two exports and about ten minutes.
- Look at the two columns together. Find every month where TACoS improved and revenue did not. Those months are where somebody optimised the wrong thing, and they are usually the most instructive rows in the sheet.
- Split the sheet by marketplace if you sell on more than one, and by product family if one line is at a different stage from the rest.
- Put TACoS and revenue next to each other in whatever weekly report you already read. Never on their own, never in a separate deck, and never a month behind.
Keep ACoS. You need it — it is still the right tool for judging a keyword, a placement or a single bid change, and TACoS is far too blunt for any of those. What TACoS is for is the question ACoS structurally cannot answer: is the advertising making this business bigger, or just making the reports look better?