Return on ad spend is a useful dial and a terrible destination. It answers one narrow question, how much revenue a platform believes it produced for what you paid it, and stores keep promoting it to a role it cannot fill: the number that says whether the business is working.
The trouble is that it can move for reasons that have nothing to do with whether you made money. Cut spend and it usually rises, because you keep the cheap conversions and drop the expensive ones. Sell more of your low-margin range and it rises while your profit falls. Let a campaign buy your own brand searches and it rises while your total orders sit exactly where they were.
We still look at it every week. We just never let it be the number that decides anything on its own. A ratio with revenue on top and one channel's spend on the bottom is missing the two things that determine whether the store survives: what the goods cost you, and everything else you spent.
Revenue on the top line is not money you keep
The first structural problem is that a return figure is calculated on revenue, and revenue is not yours. Out of it come the cost of goods, the payment fees, the pick and pack, the shipping you subsidised and the refunds you will process next month.
Two stores can report the same return and be in completely different situations. Say both spend $10,000 and report $40,000 in revenue. One sells at a gross margin of six in ten, so the ads produced $24,000 of gross profit and cost $10,000. The other sells at a gross margin of two in ten, so the same campaign produced $8,000 of gross profit against $10,000 of cost, and every extra dollar spent makes the situation worse.
That is why the first thing we work out for any store is break-even return, which is simply one divided by your gross margin. At six in ten margin, break-even sits near 1.7. At two in ten, it sits at 5. The same reported figure is a success in one store and a slow bleed in the other, and no benchmark you read anywhere can tell you which is which.
Once you know your break-even, the reported figure becomes readable. Until then it is a number without a scale.
If you want a better return figure this month, cut your budget. It works every time, and it shrinks the business.
The denominator is only part of what you spend
The second structural problem is the bottom of the fraction. It contains the media cost of one platform. It does not contain the agency or staff cost, the creative production, the apps, the tools, the discounts you funded, or the spend on every other channel.
This matters most when stores compare channels. A channel with a small budget and a strong ratio looks like the winner and gets more money, right up to the point where the extra money finds nothing left to buy at that price. Meanwhile the channel doing the heavy lifting on new customers reports a weaker ratio because that work is genuinely more expensive.
It also hides fixed costs. A store paying for photography, an agency, and a stack of subscriptions has a real cost of marketing that no platform ratio will ever show. If you make decisions on the ratio alone, you can grow the reported figure every quarter and still watch the profit line flatten.
This is the gap that a whole-business measure fills. Total revenue divided by total marketing cost has no attribution in it, cannot be gamed by a platform, and matches something a bookkeeper would recognise.
Attribution makes the number a claim, not a fact
Every platform reports the conversions it believes it caused, using its own window and its own rules. Those beliefs overlap. Add up the revenue claimed by your ad platforms and your email tool and the total will usually exceed what your store actually took, sometimes considerably.
So a reported return is not an observation. It is a claim made by an interested party. That is not an accusation of dishonesty, it is a description of how attribution works when several channels touch the same buyer and each one counts the sale.
The practical response is not to abandon platform reporting. It is to demote it. Use platform figures to compare a campaign against itself over time, where the measurement bias is at least consistent. Use your store's own order data to decide what the business earned.
And when a platform reports a large improvement that your order count does not reflect, believe the order count. Store revenue is the number your bank agrees with.
The number rises fastest when you stop growing
There is one behaviour that gives the whole thing away. If you want a better return figure this month, cut your budget. Keep only the searches and audiences that were converting most cheaply and the ratio will improve, reliably, every time.
It will also shrink the business. You have stopped paying for the harder, more expensive customers, which are the ones that make a store bigger next year rather than more efficient this month.
This is the trap in setting a return target as a company goal. The goal is achievable by doing less, and organisations are very good at achieving goals in the cheapest available way. A team asked for a higher ratio will deliver it, and the delivery will look like a contracting ad account.
We would rather set the target on profit at a volume. Grow orders while keeping contribution per order above a floor you set from your margin. That goal cannot be achieved by retreating, and it forces the conversation onto the things that actually create room to spend: better conversion, higher order value, more repeat purchases.
What we watch instead
Four things, every month, in this order.
Total revenue against total marketing cost, for the whole business. This is the ratio that cannot be gamed by moving budget between platforms.
Contribution per order, meaning revenue minus cost of goods, fees, shipping and the acquisition cost. If this is positive and holding while orders grow, the account is healthy no matter what any platform reports.
New customers acquired, separated from returning ones. This is the number that tells you whether you bought growth or bought back people you already had, and it is the single most useful split to add to your reporting.
Break-even return by product group, so that spend decisions have a floor that reflects what each part of the catalogue can actually carry.
Campaign return still has a job in all of this. It is the right dial for deciding whether one campaign should get more of the budget than another, this week, at the same target. That is a real job. It is just a smaller one than the seat it usually occupies.