What is a good ROAS for a new product?

Start from the return at which the product itself breaks even, worked out from its own margin rather than the store average. A new product often has a different cost of goods, a different weight to ship and a different price point from your bestseller, so borrowing the store's usual target will point you at the wrong answer.

Then accept that the opening period will come in below that, and decide in advance how far below is acceptable. This is not a failure of the campaign. It is what buying an unknown product costs. There is no review count on the page, no repeat buyers, no existing customers who already trust the item, and no history for the bidding system to learn from. All four of those improve the number later, and none of them exists in week one.

So the useful launch question is not what return is good. It is how much you are willing to pay to find out whether this product sells, and what you will do with the answer. Treat the first stretch of spend as the cost of information, price it, and cap it. That framing stops the two common failures: killing a product at day four, and letting a losing product run for a quarter because nobody set a line.

Four things make a new product's reported return look worse than the product deserves, and knowing which are at work stops you from drawing the wrong conclusion.

Bidding has no history. Automated bidding on a new item is guessing until it has seen a run of conversions, so early cost per order is high and unstable. This is the one that fixes itself, given volume.

The page has nothing to reassure a stranger. No reviews, no photos from customers, no questions answered by other buyers. The same traffic that converts on your established product will convert worse here, and the gap is the missing proof rather than the ad.

The measurement window is too short. A considered purchase can take days or weeks between the click and the order, so a launch measured on the first seven days always understates itself. Compare a fixed window against the same window, not a young campaign against a mature one.

And the audience is cold. Launch spend usually reaches people who have never heard of the product, while your existing campaigns quietly harvest people who already know you. Those are not the same job and should not carry the same target.

Write down three numbers before the campaign turns on: how much you are prepared to spend in total, how long you will run before judging, and what result would make you stop.

Size the budget by conversions rather than by dollars. You need enough orders in the window for the result to mean anything, and a handful is not enough. Work backwards: take a plausible conversion rate for the page, take your cost per click, and see what spend would produce enough orders to read. If that number is larger than you want to commit, the honest options are to narrow the launch to one product and one audience, or to wait until the product has some organic sales and reviews behind it.

For the length, two to four weeks suits most stores, longer if the purchase is considered or your order volume is small. Whatever you choose, hold it. Changing the creative, the target and the budget in the same week guarantees you learn nothing.

The stop line should be a return, not a feeling. Something like: if the fortnight after learning settles below break-even and the add-to-cart rate is also weak, we stop and fix the page before spending again.

In the first two weeks, the page metrics are more informative than the return, because they tell you what is broken while there is still time to fix it.

If clicks are cheap and add-to-cart is close to your store's normal rate, but orders do not follow, the problem is after the product page. Look at shipping cost, delivery time, stock messaging and the checkout.

If clicks are cheap and add-to-cart is far below your usual, the product page is not doing its job for a stranger. Missing sizing, no scale in the photos, an unclear description of what the product actually is, or a price that reads high without context. Fix the page, then judge the ads.

If clicks are expensive and few, the ad or the targeting is not matching demand. For search and shopping, check that the title uses the words people type rather than your internal name. For social, the creative is the variable.

And if everything looks fine but nobody buys at any price, that is a real answer too. Some products do not sell, and finding that out for the price of a small test is a good outcome rather than a bad one.

Graduate the product once it produces orders steadily at or above break-even, without needing its own protected budget. Until then, keep it separate.

The reason to keep it separate is simple. Inside a shared campaign, a new product competes with proven ones for the same money, and the bidding will always prefer the item it already knows how to sell. The new product gets a trickle of impressions, never gathers data, and gets blamed for a result it was never funded to produce. Giving it a ring-fenced budget for the launch window is how you get an honest test.

When it graduates, fold it into the campaign that matches its margin band rather than the one that matches its category, and let it compete on the same terms as everything else.

At the same time, look at what the launch has taught you about the second order. If early buyers of this product come back, or buy something else, the return you can afford on it is higher than the first-order arithmetic suggests. Check that from your order data after a few months rather than assuming it on day one.