What Is a Good TACoS on Amazon? Lower Isn't Always Better

A brand told me their TACoS was running around 9% and said it the way you'd report a good result. Their agency had it trending down quarter on quarter, and everyone involved was pleased with it.

Then we talked about what they wanted the channel to do, which was reach new customers, because the existing ones were buying about as much as they were going to.

A 9% TACoS is a fine number for a brand protecting margin on customers it already has, and it's the wrong number for a brand trying to win trial. Both sides of the table agreed within a few minutes: to grow the customer base, that figure needs to be nearer 15% to 25%.

Nobody had made a mistake. The metric had been optimized in a direction nobody had checked against the goal.

What does TACoS measure?

Total advertising cost of sales: total ad spend divided by total sales on the channel, including sales your ads had nothing to do with.

That last part is what makes it useful. ACoS compares ad spend to ad-attributed sales, so it tells you how efficiently the ads bought the sales the ads bought, which is a closed loop. TACoS opens the loop. If your organic sales grow while ad spend holds, TACoS falls. If your ads are the only thing keeping sales alive, TACoS stays stubbornly high whatever your ACoS says.

That's why we look at TACoS and not ACoS when we want to know how the channel is doing rather than how a campaign is doing.

How do you calculate TACoS?

Take total ad spend for a period, divide it by total sales for the same period, ads and organic together, and multiply by 100.

A worked example. Say a month's ad spend was $30,000, the ads were credited with $120,000 in sales, and the account sold $400,000 in total. ACoS is 30,000 divided by 120,000, or 25%. TACoS is 30,000 divided by 400,000, or 7.5%. Same spend, same month, and the two numbers describe different things: the ACoS says each ad-attributed sale cost a quarter of its price, and the TACoS says advertising cost seven and a half cents of every dollar the channel brought in.

You'll usually build it yourself. Pull spend from the Amazon Ads console and total sales from Business Reports in Seller Central, for exactly the same date range. Include every ad type you run, Sponsored Brands, Sponsored Display and any programmatic spend along with Sponsored Products, or the number comes out flattering. Then do it per ASIN as well as for the account, for the same reason we build margin per SKU.

There's a practical argument for the total-sales denominator, too. Ad-attributed sales depend on the attribution window you pull. When I connected an AI tool to Seller Central, it gave me different sales numbers depending on whether it read a 7-day or a 14-day window: one question, two answers. Total sales don't move with the window, and that makes TACoS the harder number to argue with.

Why isn't a lower TACoS always better?

Because TACoS falls for two very different reasons and the number doesn't tell you which.

It falls when your organic position improves, so you're getting sales you no longer have to pay for. That's the good one, and it's what a well-run account produces over time.

It also falls when you stop spending on anything hard. Non-branded search, new audiences, competitor placements, the top of the funnel. Cut all of that and keep an always-on branded campaign, and your TACoS drops beautifully while your advertising does nothing except collect people who were already coming to find you.

Branded search converts at roughly 90% to 95%, so it produces excellent-looking reports. A brand harvesting its own demand and a brand earning new demand can show you the same efficiency number, and only one of them is growing.

The five-minute test for whether anyone new is buying

Here's the diagnostic I'd run before I looked at another advertising report.

Pull your catalog by unit volume and find your cheapest entry product, the one a curious new customer would try first. Compare its volume to your flagship.

On a legacy brand I looked at, the low-price entry SKU was doing a small fraction of the flagship's volume. That's backwards. A brand acquiring new customers sells a lot of the cheap thing, because trial is a small purchase, and the flagship is what people buy on the second and third order.

When the entry product is quiet and the flagship is busy, almost no trial is happening. The revenue is loyal customers repeat-buying, which is a real business and also a ceiling. It'll hold for as long as those customers stay and no longer.

Then check it properly with the new-to-brand metrics in your advertising reports and the repeat purchase behavior report in Brand Analytics. The catalog test tells you where to look. Those reports tell you the size of it.

New customers against churn

Our head brand manager frames it in the way I'd want every brand to hear, and he arrived at it independently of anything above: new customers have to outpace the ones you're losing, and repeat is the name of the game in CPG.

So when an account reports twenty-five new customers a week, the number on its own means nothing. Twenty-five new against fifteen lost is a business compounding. Twenty-five new against forty lost is a business shrinking while its dashboard shows growth, because the top-line revenue can hold for months on the back of existing subscribers before the gap surfaces.

Amazon gives you enough to run this. Customer counts week over week, repeat purchase rate, and Subscribe & Save subscriber counts and trend. If your subscriber count is flat and your new customer count is modest, you're replacing people rather than adding them.

Josh Grillo, who runs brand management here, finds a version of this on a lot of audits of subscription-heavy brands. The account is still winning new-to-brand customers every week and shrinking anyway, because more people are lapsing than arriving. If a large Subscribe & Save coupon stacked on the standard subscription discount isn't bringing people in, price isn't what's holding subscriptions back. And when subscribers are worth several times what one-time buyers are, and most of the sales come from reorders, the repeat engine is worth feeding even when it looks like it's starving.

That's the moment a higher TACoS becomes the right call, and it's worth being explicit about why. Acquiring a customer costs more than harvesting one. If the job is trial, you're buying customers at a price that looks inefficient on this month's report and is justified by what they buy next. That only works if you know your repeat rate, which is the connection between these two metrics that most brands never make.

What conversion rate belongs beside it

While you're in the reports, check conversion, because a TACoS problem often turns out to be a listing problem. We've set out the benchmarks in the piece on Amazon listings: roughly 5% is the floor rather than the goal, and a healthy listing does about double that.

The connection to this article is the sequence. If your listing converts at 3%, raising TACoS to chase trial is buying traffic a broken page will waste. Fix the page, then spend.

How do TACoS, ACoS and ROAS fit together?

ACoS and ROAS are the same ratio turned upside down, both measured on ad-attributed sales, and TACoS is the one measured on the whole channel; none of the three means much until you hold it against your margin.

ROAS is ad-attributed sales divided by ad spend, and a 4x ROAS is the same thing as a 25% ACoS. The useful version of either is the break-even. An ad-attributed sale covers its own advertising only when ACoS sits below your pre-ad gross margin. Put the other way up, ROAS has to beat one divided by that margin. At a 40% margin, break-even ROAS is 2.5. I put it to brands this way: if you have a 10% margin, even if I'm getting you a 5-to-1 return on ads, you're losing money.

That math is what tells you whether there's room to move. A brand at a 40% margin running ROAS around 3.5 is clearing the cost of its campaigns with margin to spare, and it can afford to spend more on finding new customers. TACoS then tells you what all that spending adds up to across the channel, and that's the number the parent company's P&L will see.

So what's a good TACoS for my brand?

Set it from the job and then check it against your margin.

If you're protecting profit on an established catalog with strong organic rank, single digits is reasonable and falling TACoS is a sign the flywheel is working.

If you're trying to win new customers, expect 15% to 25% while you do it, and expect it to come back down as the customers you bought start reordering without an ad in front of them.

If you're launching, it'll be higher than either and you shouldn't be alarmed by that, as long as you've decided in advance how long the launch window runs and what the product needs to be doing at the end of it.

When I walked a manufacturer's operations team through this, the range I gave for a brand pushing growth was about 12% to 15%, and I told them anything under 10% can mean the account has pulled back too far. For a brand recovering from a decline, Josh's range is 15%, maybe 20% depending on margins. Those overlap the low end of the 15% to 25% above, and your margin decides where you belong in it.

You'll find tables online that give a good TACoS by category. I'd treat them with the same caution I give to category conversion benchmarks: a category median includes every neglected listing in the category, and no table knows your margin.

Then run the margin check, because none of those ranges mean anything against a broken unit economic. Your pre-ad gross margin sets what you can afford to spend. At a 30% margin, a 25% TACoS leaves you very little and had better be buying customers who come back. At a 50% margin, the same spend is comfortable. Build the number per SKU, because a blended catalog average hides the loser inside the winner, and the loser is usually the one you're advertising hardest.

What does a rising TACoS tell you?

A rising TACoS is either a decision you made or a symptom, and you should be able to say which within a minute of looking.

The decisions are easy to recognize: a launch, a push for new customers, a promotion window you planned for. If none of those is happening and the number is climbing, the cause is usually one of a few things we find again and again.

Organic rank slipping is the most common. Rank on Amazon follows sales velocity, and Josh estimated about 40 units a day to hold a page-one spot in one small niche we looked at. That gives you a target you can count. When organic sales fall away, the same ad spend is propping up a smaller total, and TACoS rises even though nothing in the ad account changed.

Stockouts are the second. A product starts losing organic and sponsored rank as soon as it stocks out, winning the position back can take up to 90 days, and you pay advertising to rebuild what the gap cost you.

Third, losing the Buy Box. If a reseller is holding the Buy Box on your listings, your total sales shrink while your spend holds and TACoS rises. Your ad reports also undercount what the campaigns did, and bids get cut on keywords that were working.

And then plain waste: an ad group so far past break-even that Josh's advice is to turn it off immediately.

Read the trend beside ACoS. If both are rising and you aren't launching, more spend is buying less. If spend and ACoS both hold steady while TACoS climbs, the ads are performing about as well as they were and the organic sales underneath them are shrinking. Look at rank, stock and the Buy Box before you touch the campaigns.

How do you raise TACoS without wasting the money?

Cut the bloat first, then put the extra spend on campaigns that find new customers, because a higher TACoS target is a budget decision and gives nobody license to spend anywhere.

On an account like that, Josh's advice is to spend more, and in the same breath to point at the worst ad groups and say not there. The account needs more spend on discovery and new customers and less on the campaigns that are burning money.

The structure we build follows from that. Auto campaigns do the research, because they're a cheaper way to find converting search terms than testing on exact match. Once a term has converted enough to trust, we negate it in the auto campaign and move it into an exact-match campaign with its own daily budget. Around that sit non-branded terms, competitor terms, competitor product pages, Sponsored Display retargeting of shoppers who viewed or added to cart and left, and your own product pages defended with your other products. A shopper who lands on one of your listings should see more of you there, and not a competitor. Branded search gets an always-on, exact-match, low-spend campaign, because you'll win your own terms more cheaply than a competitor can, and defending them is the cheapest ad money in the account. I build it from the bottom up, starting with the point-of-purchase campaigns, and add programmatic only once the core PPC is performing.

Two cautions. Don't raise budgets by relaunching campaigns: campaigns build up history, and a copy won't perform like the original it replaced. Accounts that have changed hands often carry several versions of the same Sponsored Products campaign, launched, paused and relaunched with inconsistent bids. And fix the listing before you spend. A coffee brand we took on wanted its catalog reworked before it trusted us with ads; it grew three months in a row with no ads running, and adding ads roughly doubled it after. Listing work like that lowers TACoS without touching the ad account. Before anyone takes credit for a falling number, find out what moved it.

When a low TACoS is exactly right

I don't want to leave the impression that efficiency is the enemy.

A brand with a healthy repeat rate, strong organic rank on its core terms and a catalog it isn't trying to extend should be running lean. That's the flywheel doing its job: you built the position, you're not paying to re-buy it, and the margin comes home.

If you want to know whether your low number is earned, there's a test for it. Pick a non-branded term where you hold strong share, deliberately scale ad spend back on it, and watch the search query performance report: impressions, click share, add-to-cart share and purchase share. If your share holds, that spend was redundant. Being number one in your category makes the test harder, because there's less room to cut before you start giving up impression share.

The failure isn't a low number. It's a low number nobody chose, on an account where the goal is growth and the advertising stopped doing anything difficult. When your TACoS falls, ask which of the two reasons produced it. The report won't tell you.

Does TACoS work the same way on Instacart?

Expect it to stay higher on Instacart, because Instacart ads don't build organic rank the way Amazon ads do.

On Amazon, advertising buys sales velocity, velocity earns organic rank, and over time a growing share of sales arrives without an ad. That's how a healthy Amazon TACoS comes down. Instacart doesn't reward ad-driven sales with rank in the same way, and the ad share of sales stays high. On the Instacart accounts we run, the large majority of sales come through ads, and the spend can't be tapered the way it can on Amazon. Judge Instacart against its own baseline, on new-to-brand and repeat customers, and don't hold it to an Amazon TACoS target it was never going to reach.

The bottom line

TACoS is the rare Amazon metric that rewards being read rather than being optimized. The number goes down for a good reason and a bad reason, and telling them apart takes about ten minutes in reports you already have.

Do the entry-SKU check this week. If the cheap product is quiet, your advertising is collecting rather than growing, and no amount of efficiency will change that.

If you want a read on which of the two your account is doing, send us your ASINs and we'll look at the branded and non-branded split, the new-to-brand trend and the conversion rates, and tell you what we see. No obligation either way.

cta_free_trial

Read more blogs

How to Increase Amazon Sales Without Buying Them: The Amazon Profit System
  • Amazon

How to Increase Amazon Sales Without Buying Them: The Amazon Profit System

How to Increase Amazon Sales Without Buying Them: The Amazon Profit System
Flat for Two Years, Then +60%: A Specialty Ingredient Brand's Amazon Turnaround
  • Amazon
  • Case Study
  • Natural CPG

Flat for Two Years, Then +60%: A Specialty Ingredient Brand's Amazon Turnaround

Flat for Two Years, Then +60%: A Specialty Ingredient Brand's Amazon Turnaround
From Break-Even Ads to Roughly 10x Profit: A Natural Snack Brand's Amazon Turnaround
  • Case Study
  • Natural CPG
  • Amazon

From Break-Even Ads to Roughly 10x Profit: A Natural Snack Brand's Amazon Turnaround

From Break-Even Ads to Roughly 10x Profit: A Natural Snack Brand's Amazon Turnaround

Find out what your Amazon channel really earns

Apply for a free trial of our work, or chat with our team. Either way you'll leave knowing your true profit, which is more than your current reports can tell you. And if we work together and miss the targets we set, the sprint to fix it is on us.