What is a good ROAS for ecommerce?

Return on ad spend is revenue divided by ad spend. On its own it says nothing about profit, because it uses revenue, and revenue is not money you keep. What you keep is gross margin, so gross margin is what decides whether a given return is good.

The rule of thumb is simple. Divide one by your gross margin expressed as a decimal and you get the return at which ads pay for themselves before overheads. A store keeping half of every dollar breaks even at a return of 2. A store keeping a quarter breaks even at 4. A store keeping three quarters breaks even at about 1.3.

That is why quoted figures are useless without context. A return of 3 is comfortable for one store and a slow loss for another, and both statements can be true on the same day in the same category.

Use a full margin, not a rough one. Take the selling price, then take out the cost of goods, payment processing, packaging, the shipping you actually pay rather than the shipping you charge, expected returns and any per order app fees. What is left is the number that matters. Most stores that discover their campaigns were unprofitable discover it here, because the first four deductions were counted and the last three were not.

Once you have that figure, a good return is any number above it that also leaves enough to pay for the parts of the business ads do not touch.

Three versions of the number exist and people argue because they are quoting different ones.

Campaign return tells you whether a particular campaign is worth its budget. It is the most useful figure for deciding where the next dollar goes, and the least reliable as a picture of the business, because every platform counts conversions it can see and none of them can see the whole path.

Channel return tells you whether a channel is carrying its weight. It is worth watching over months rather than days, and worth comparing only against the same channel's own history.

Blended return is total revenue divided by total ad spend across everything. It cannot be inflated by attribution, because it does not use attribution at all. It is the honest number for a board meeting and the wrong number for deciding a bid.

Use all three, for different jobs. When the platforms report a healthy return and the blended number is falling, the platforms are claiming credit for orders that would have happened anyway. When the platforms look weak and the blended number holds, the tracking is undercounting and the campaigns are doing more than they get credit for.

That gap between reported and blended is a diagnosis in itself. Track it as a ratio month to month and it tells you how much to trust the dashboards.

Stores that only ever accept a high return end up small, because the highest returns come from traffic that was already going to buy. Brand searches, retargeting and email pull in people who had decided. Growth comes from paying more for people who had not.

So the honest question is not how high the return is, but how much you can afford to pay for a first order given what a customer is worth over time. If a customer typically buys twice more, the first order can run near break-even and the relationship still pays. If nobody ever comes back, the first order has to carry everything.

This is why the repeat purchase rate belongs in an advertising conversation. It sets the ceiling on what you can pay to acquire a customer, which sets the return you can accept, which sets how fast you can grow.

Be careful with the logic though. Lifetime value used as an excuse for a poor first order return only works if you can show the second order in your own data, by cohort, not in theory. Look at customers acquired six months ago and count how many bought again. That figure, not an optimistic projection, is the one to plan with.

Separate new and returning customers in reporting where the platform allows it. A campaign at a return of 2 that brings first time buyers may be worth more than one at 6 that mostly re sells to your existing list.

Bidding down is the obvious move and usually the weakest one. Lower bids buy fewer clicks and cheaper positions, which often means the same cost per order with less volume. There are four better places to look, in order of how quickly they pay.

Raise the conversion rate on the pages the ads land on. Every point of improvement lifts the return on all of the traffic you already pay for, and costs nothing extra per click.

Raise the order value. A sensible bundle, a free shipping line set just above the typical order, or one relevant add-on at the cart lifts revenue per click without touching the auction. Return on ad spend rises because the numerator moved.

Cut the products that cannot carry a click. Some items simply do not have the margin to be advertised, and moving that budget to items that do is the fastest change in the account.

Then, and only then, work on the campaigns. Negative keywords for searches that never buy, tighter product groups, better titles so the match is closer to the intent. That work is worth doing, but it works better on a store that has already fixed the three things above it.

If all four have been done and the number still sits below break-even, the channel may not fit the product at that price. Say so early rather than spending another quarter proving it.