There is no good ROAS. There is a figure that clears your gross margin and a figure that does not, and those two are different for every business reading this.
Shopify, the largest authority in ecommerce, declines to answer the question at all. Its own page on the subject says an acceptable return depends on your profit margins, operating costs and advertising goals, then stops. No number follows.
That silence is honest. It is also unhelpful when you are trying to decide whether 2.4x is a result or a problem.
Read on if somebody has asked you what target to set and you were about to go looking for an industry average.
Nobody will give you a good ROAS number, including Shopify
The biggest name in ecommerce software has published a page on this and refuses to put a figure on it.
Shopify’s stated position is that an acceptable return depends on your profit margins, your operating costs and your advertising goals. All three are correct. None of them is a number.
That refusal is right, and it also explains why the results pages above this one are stuffed with invented tables. Somebody has to answer the question, the honest answer is arithmetic rather than a benchmark, and a market of fabricated tables grew into the gap.
So a good ROAS is not a fact you can look up. It is a calculation you run once, then re-check every quarter when your costs move.
Why the industry table is the wrong place to look
Category averages answer a question nobody actually has.
Two businesses inside one category can run at 20% and 50% gross margin, sell baskets of £40 and £400, and buy demand on completely different channels. An average across that pair describes neither of them.
Then there is the sourcing problem. Most category tables circulating online carry no sample size, no date range, no geography and no stated method, and several trace back to the same undisclosed upstream source.
Apply a simpler test before you quote one. If a table cannot tell you how many accounts it covers, and across what period, it cannot tell you anything reliable about yours either.
Your margin sets the floor, and nothing else does
One division produces the only number that matters here.
Break-even is one divided by your gross margin. Across my ecommerce clients that margin runs from 20% to 50%, which puts break-even somewhere between 5.0x and 2.0x.
Read that range again. The floor moves by a factor of two and a half across perfectly ordinary businesses, which is wider than the gap between any two rows in any industry table you will find.
So a 3.0x is a healthy result at 50% margin and a losing one at 20%. Same advertising, same platform, same quarter. Only the cost of goods differs, and cost of goods is exactly the thing nobody publishes.
What a good ROAS looks like at 20% and at 50% margin
Put a real basket behind the arithmetic and it stops being abstract.
Shopify reported an average order value of $114.70 across its full merchant base over Black Friday Cyber Monday 2025, published on 2 December 2025. Use that as your basket.
At 50% margin each order contributes $57.35, break-even lands at 2.0x, and a good ROAS starts somewhere above it. At 20% margin the identical order contributes $22.94, break-even lands at 5.0x, and a respectable-looking 3.0x is losing $15.29 on every order shipped.
Nothing in that paragraph is industry-specific. Two shops selling identical products at identical prices reach opposite verdicts, purely because their cost of goods differs.
The progression that made one account defensible
One live account of mine shows why a single headline figure means so little.
A US flag and banner ecommerce client, Meta, June to August 2026. Count every campaign, awareness included, and the account returns 1.25. Count only the six sales ad sets and it returns 1.58. Drop the single ad set that was testing a neighbouring state, which failed on fulfilment rather than creative, and the same three months return 1.99.
Nothing inside the account changed between those three numbers. Only the question being asked changed.
I will not tell you which of the three is the good one, because that client has never disclosed gross margin to me. What I can say is that arguing about 1.58 against a published benchmark would have wasted everybody’s afternoon.
Channel mix changes a good ROAS more than industry does
Where the money goes moves this number further than what you sell.
My EU fashion client returned 4.7x across 1,715 orders in the full year 2022. That came from Meta and Google running together, with the majority of sales arriving through Google Search and Shopping. My read is that a Meta-only view of the same account would have reported a much lower figure, because search harvests demand that social created.
So a good ROAS on Meta and a good ROAS on Google Search are not the same target, even inside one business.
Split your expectations by channel before you split them by industry. The first split describes something real. The second mostly describes whoever assembled the table.
One published median, and the caveat it comes with
If you want a single figure to hold loosely, this is the least bad one available.
Triple Whale reported a median return of 2.04 for 2024 across the brands using its platform, with Meta accounting for 68.3% of the ad spend it tracked. The wider report covers $18.4 billion in spend across more than 33,000 brands.
Now say the caveat out loud, because the pages quoting this figure never do. That 2.04 carries no stated sample size or date range on the page where it appears, and the platform’s own customers are not a random sample of ecommerce.
Hold it as a sanity check instead of a target. If your account reports 6.0x, something is probably being counted twice. I manage one account reporting 44.37, and that is a tracking fault rather than a triumph.
Setting a good ROAS target you can actually defend
Four inputs, in this order, with no industry table anywhere among them.
Start with gross margin and get it in writing. Convert it into break-even. Add whatever profit you want on top, which is a business decision rather than a marketing one. Then check the platform can actually learn against the target you set.
That last step catches people out. Google states that target ROAS bidding on Search and Shopping wants 15 conversions in the past 30 days, that Demand Gen needs 50 in 35 days, and that you should wait four weeks or three conversion cycles after a reporting change before judging anything.
Below those volumes, a target ROAS is a wish. Set a cost target instead, gather conversions, and come back to the ratio once the account can carry it.
What to change this week
Three things.
Get gross margin in writing from whoever owns the profit and loss, then divide one by it. That single figure is your break-even, and a profitable ROAS is anything meaningfully above it.
Then rebuild reporting so one account can be read three ways: all campaigns, sales campaigns only, and sales campaigns minus anything currently under test. Mine read 1.25, 1.58 and 1.99 across the same three months.
Then take the industry benchmark out of your reporting deck. It has been answering a question about somebody else’s business since the day it went in.
A good cost per lead is decided by your close rate runs the identical argument on the lead generation side, the two acquisition figures I can date gives the volume and the dates behind the accounts quoted here, and revenue divided by spend is not money in the bank covers why the ratio and the profit pull apart. Basket size and checkout friction move this number as much as bidding does, which is why website work belongs in the same conversation. To have a target set from your margin rather than a table, book a teardown.