What Is a Good ROAS on Meta and TikTok Ads?

Every brand asks what a good ROAS is, and the honest answer annoys people: the benchmark is not the point. Two brands can both report the same return on Meta, and one is funding payroll while the other is quietly liquidating inventory at a loss. The difference is not in the ad account. It is in their cost structures, and only one of them has done the arithmetic.

The benchmarks below are worth knowing, because they tell you whether your account is normal. They will not tell you whether it is profitable.

What is a good ROAS on Meta ads?

A good ROAS on Meta in 2026 is 2.87x for the average ecommerce brand, according to Hawky.ai’s 2026 ROAS benchmarks by industry. The same report puts the ecommerce median at 2.04x, which is the more useful of the two figures: half of all ecommerce advertisers sit below a 2:1 return, and the average is dragged up by the tail above it. The 2026 Meta ecommerce compilations put the bottom quartile under 1.8x and top performers at 4x to 6x.

Category matters more than most brands expect. According to the same Hawky.ai dataset, beauty and personal care averages 4.52x on Meta while healthcare sits near 2.30x, at the bottom of a by-industry range that runs up to about 8x in legal services. That is a twofold spread inside ecommerce alone, and it has almost nothing to do with media buying skill. High-margin, high-repeat, visually demonstrable products convert on paid social in ways that considered, regulated or infrequent purchases do not. If you sell in a hard category, a 2.4x may be a genuinely good account and a 4x may be unreachable at any spend.

Campaign type moves the number nearly as much. Prospecting typically returns between 1.8x and 3.2x while retargeting exceeds 8x in some accounts, per Superscale’s 2026 ROAS benchmark compilation, which sounds like an argument for shifting budget to retargeting until you notice what retargeting is: showing ads to people who already found you. Its ROAS is high because the attribution is generous, not because it is where growth comes from. An account that improves its blended ROAS by starving prospecting is an account that will shrink next quarter.

One structural note. Automated campaign structures, Advantage+ Shopping in particular, are widely reported to outperform manual ones, and the specific multiples quoted for that gap vary enough between sources that none of them are worth repeating here. Test the automated structure against your own manual one and trust that number instead, because accounts adopting it earliest skew toward better-resourced advertisers and the reported gap carries that bias.

What is a good ROAS on TikTok ads?

A good ROAS on TikTok in 2026 is around 2.5x for ecommerce, against a platform average of 2.21x based on full-year 2025 data in Triple Whale’s TikTok benchmarks, which drifted down slightly year over year as the auction got more competitive.

The category spread on TikTok is wide in the same way it is on Meta. Beauty and skincare averages around 4.8x on Spark Ads, food and beverage 4.1x, health and supplements 3.6x, fashion 3.2x and home and lifestyle 2.9x, per Influee’s 2026 TikTok benchmark compilation. Impulse-priced, visually demonstrable products do well. Considered purchases do less well, and no amount of budget changes that ordering.

The comparison between the platforms is less reliable than the numbers suggest, and it is worth being explicit about why. TikTok’s default attribution settings are shorter and its signal on cross-device journeys is weaker, so a purchase that Meta would claim often lands in TikTok’s account as organic or direct. In practice TikTok understates its own contribution more often than it overstates it, which is the opposite of the usual assumption.

That is why brands running both platforms should judge them on blended performance, total revenue over total spend, and on what happens to blended numbers when one platform is paused. A full comparison of the two platforms on cost and creative requirements is a better basis for deciding where to start than either platform’s reported ROAS.

What ROAS do you actually need to break even?

Your break-even ROAS is 1 divided by your contribution margin, and until you calculate it, every benchmark on this page is decoration.

Contribution margin is the share of each sale you keep after cost of goods, payment processing, shipping and expected refunds. Not gross margin. Triple Whale’s break-even ROAS guide works the example: a product with 35% COGS, 3% payment processing and 8% shipping has 46% variable costs, so a 54% contribution margin and a break-even ROAS of 1.85x.

The arithmetic is unforgiving in both directions.

Contribution margin	Break-even ROAS	Profit at 3x ROAS, per $1 of ad spend
30%	3.33x	lose 10 cents
40%	2.50x	make 20 cents
50%	2.00x	make 50 cents
60%	1.67x	make 80 cents

Read the first row carefully, because it is the one that catches people. A brand with a 30% contribution margin hitting 3x is above the ecommerce average reported by Hawky.ai, would be described as performing well by most agencies, and is losing money on every order.

To target profit rather than break-even, the formula extends, according to the same Triple Whale guide: target ROAS equals 1 divided by (contribution margin minus your desired margin). A 54% contribution margin and a 20% target margin gives 2.94x. Calculate both numbers before you set a single campaign target, and recalculate whenever your shipping costs or discount rate move.

What ROAS should you target if customers buy again?

If customers buy again, you should stop targeting ROAS on the first order and target a payback period instead, because first-order ROAS below break-even is the intended state of a repeat-purchase business rather than a failure of one.

The metric that replaces it is CAC payback: how many months of contribution from an average new customer it takes to earn back what you paid to acquire them. Published 2026 bands differ by business model rather than by industry. Marketplaces recover acquisition in roughly one to three months, subscription brands in three to nine, and pure DTC in six to twelve, according to Eightx’s 2026 payback benchmarks. The mechanism is order frequency rather than cheaper media: a subscription brand gets its second contribution event in 30 days instead of 12 months, which compresses payback by 40 to 60% at an identical acquisition cost.

Payback is the metric that actually constrains decisions, because it is a cash constraint before it is a profitability one. Work the arithmetic on a $20 million brand putting 25% of revenue into acquisition, which is $5 million a year. At a nine-month payback, roughly $3.75 million is sitting in customers who have not yet repaid, per Eightx’s framing of the same benchmark. At three months, it is $1.25 million. Nothing about the ad account changed between those two numbers and the second brand can self-fund growth the first one has to raise for.

Business model	Typical CAC payback	What first-order ROAS can legitimately be
Marketplace	1 to 3 months	At or above break-even, because there is little to recover later
Subscription	3 to 9 months	Below break-even by design, often well below
Pure DTC, low repeat	6 to 12 months	Close to break-even, because the second order may never arrive

Is LTV to CAC a better target than ROAS?

LTV to CAC is a better diagnostic than ROAS and a worse target than payback, because one of its two inputs is a forecast you control the assumptions on.

The ratio everyone quotes, LTV to CAC, is worth calculating and worth distrusting as a target. Three to one is the number that circulates, and the published 2026 benchmarks disagree about it in ways that matter: Eightx puts the cross-vertical median near 3.4 to 1 and the spread from about 2.1 to 1 in consumer electronics up to 5.2 to 1 in luxury, while Foundry CRO’s 2026 benchmarks call 1.5 to 1 through 3 to 1 the healthy band for DTC on thinner margins. Two credible sources on the same metric, and no agreement on the pass mark. Use it to compare yourself against last quarter, not against a number from a blog, and remember that every LTV input is a forecast while every CAC input is a receipt.

One more trap sits in the metric brands reach for next. MER, the marketing efficiency ratio, is defined by Triple Whale as total revenue divided by total ad spend, which produces a figure like 4.0. The same company reports a median MER of 41% across its customers in 2025, which is the inverted convention, ad spend as a share of revenue. Both are in use and they point in opposite directions, so a MER quoted without its direction is not a number. Whichever way you calculate it, split new customer revenue out from the total: blended MER that includes repeat orders from email and SMS will look healthy for months after acquisition has stopped working.

Why doesn’t your reported ROAS match your bank account?

Reported ROAS and actual profit diverge because platform ROAS measures attributed revenue, not incremental revenue, and those are different quantities.

Attributed revenue includes everyone who saw an ad and bought, including the customers who were going to buy anyway. Retargeting is the clean example: it reports strong numbers because it is credited with purchases from people who already knew you. Some of that is real influence. Much of it is bookkeeping.

Three habits close the gap. Watch blended ROAS as the primary number, meaning total revenue divided by total advertising spend across all platforms. It is coarse and it does not let anyone hide inside a channel. Run holdout tests occasionally, pausing a campaign or a geography and watching what happens to total revenue rather than to attributed revenue. And track new customer acquisition cost separately, because an account whose ROAS is drifting up while new customer count drifts down is not improving. It is harvesting.

The awkward consequence is that a genuinely healthy growth account often shows a lower reported ROAS than a stagnant one, since prospecting dilutes the average that retargeting inflates. Brands that optimise the dashboard number rather than the business tend to discover this about two quarters later, when the retargeting pool runs dry.

Should you set a ROAS target in the campaign?

Set a ROAS target only when you have enough conversion volume to support one, and set it slightly above break-even rather than at your ambition.

A target return goal is a constraint on delivery. Raising it tells the system to bid only where it predicts a high return, which narrows the pool of impressions it will buy. Efficiency usually improves a little and volume usually falls a lot, and the brands most tempted to raise the target are the ones who can least afford the volume loss. Setting a five-times goal on an account averaging half that does not produce a five-times account. It produces an account that spends a third of its budget.

Volume is the other constraint, and it is the harder one. Value-based bidding needs conversion history to predict from, and an ad set that cannot reach roughly 50 conversions in 7 days never leaves the learning phase, so its predictions are guesses. Below that threshold a target return goal makes delivery worse rather than more profitable. Run the cheaper optimisation until the volume exists.

One habit worth building. A target set once is wrong within a quarter, because shipping costs, discount depth and return rates all move the break-even underneath it, and nobody gets an alert when they do. Recalculate the contribution margin every time your cost of goods or your promotional calendar changes, and move the campaign target with it.

What moves ROAS more than anything else?

Creative moves ROAS more than targeting, bidding or campaign structure, and by 2026 it is not close, because the platforms have absorbed most of the other levers.

The mechanism is in the distribution of outcomes. Motion’s Creative Benchmarks 2026, covering 578,750 creatives and $1.29 billion in Meta spend, found that about 5% of ads become winners, defined as spending at least ten times the account median and at least $500. Account performance is therefore set by how many winners you find, and how many you find is set by how many concepts you put into the auction. An advertiser testing four ads a week surfaces roughly 0.2 winners a week at that rate. One testing eighteen surfaces roughly 0.9. Motion also found the hit rate is worse for smaller advertisers, about 3.8% below $10,000 a month, which means low-spend accounts need proportionally more concepts, not fewer.

Which reframes the ROAS question. Asking how to improve a 2.2x by adjusting bids is asking the wrong question, because bid strategy operates on the ads you have. Testing volume operates on which ads you get to have.

The second lever is creative quality at the top of the funnel, measurable before revenue arrives. Hook rate on Meta and TikTok tells you within a day whether an asset earns attention, long before the ROAS figure has enough conversions to mean anything. Reading early signals correctly is how you avoid killing a good ad at 40 conversions or funding a bad one for three weeks.

Everything else, meaning audience layering, placements, manual bid caps, is worth roughly a few percent. Creative is worth multiples.

So what ROAS should you actually aim for?

Aim for your break-even ROAS plus the margin you need, and treat every published benchmark as context rather than a target. Break-even is 1 divided by your contribution margin, and the target that earns a profit is 1 divided by (contribution margin minus your desired margin), both per Triple Whale’s break-even guide. A 54% contribution margin and a 20% profit target gives 2.94x. That number is yours. It has no relationship to the 2.87x Meta ecommerce average from Hawky.ai or the 2.21x TikTok average from Triple Whale, and the fact that it may land near one of them is a coincidence.

The practice worth dropping is benchmark-chasing, and it is close to universal. Brands set a target because a competitor or a report quoted it, then read a shortfall as a media buying failure. Most of the time the target was arbitrary and the account was solvent, or the target was cleared and the account was losing money. Both mistakes come from the same habit of importing someone else’s number in place of doing the arithmetic on your own.

Where this stops applying: if you are running a first-purchase-loss model on a subscription or high-repeat product, first-order ROAS is meant to be below break-even and judging it against these benchmarks will make you cut spend on a working account. In that case the number to track is CAC payback in months, which subscription brands typically clear in three to nine per Eightx’s 2026 benchmarks, and the ROAS figure on the first order stops being a target at all.