You outspend everyone. You can't see what worked.

Mattress retail is one of the most heavily advertised categories per square foot of showroom in the country, and one of the least measurable, because the product is something almost nobody will buy without lying on it first.

So the spend is optimized against clicks, form fills and store-locator taps, none of which are revenue. Two campaigns with identical cost per lead can differ by a factor of two in what they actually sell, and nothing in the ad platform will ever tell you.

Every delivered mattress has a phone number attached to it, because somebody had to arrange a delivery window. That is the match key, and it already exists.

Matched against ad platform lead exports or a call tracker, the floor sales resolve to channels. Sales with no traceable origin stay in Direct / Unknown rather than being redistributed — which matters in a category where the honest answer often is that a third of business is brand and word of mouth.

Most retail categories can rescue a poor acquisition cost with repeat purchase. Mattress retail cannot. The customer buys, sleeps on it for eight or ten years, and by the time they need another one the store may not exist, the salesperson certainly does not, and no report anybody built will connect the two transactions. Every cent of acquisition cost has to be justified by a single sale, and there is no second order coming along later to fix a bad channel decision.

What does repeat is the household rather than the person. A guest room bed two years later, a child's first proper mattress, a parent moving in and needing one. Those purchases carry the same phone number often enough to matter, and they credit the source that produced the original customer, but they are nowhere near frequent enough to rescue a channel that acquires badly. They are a bonus sitting on top of a decision that had to be right the first time it was made.

The practical conclusion is that the ranking has to be built on money rather than on leads, and built from a long enough period to be stable. A single month of mattress sales in a category this seasonal and this promotion-driven is close to noise. A year, compared against the previous year, is a measurement. Both can be loaded at once because nothing needed to be recording at the time, which is the only reason a store can have that comparison in its first week.

A mattress retailer's margin does not live in the mattress. It lives in the adjustable base, the protector, the pillows, the frame and the delivery package sold alongside it, and the attachment rate varies enormously from customer to customer. Somebody who came in for an advertised price point and left with exactly that is a completely different transaction from somebody who came in on the same advert and left with an adjustable set, even though both of them count as one sale.

Because attachment is decided on the floor, it is tempting to treat it as a sales-performance question with nothing to do with marketing. The report usually disagrees. Sources differ systematically in the customer they send, and a channel advertising a headline price will produce headline-price customers no matter who is serving them. Ranking channels on the total ticket, accessories included, is what makes that visible rather than leaving it as a standing argument between the sales floor and the marketing budget.

Ranking on margin rather than revenue takes it a step further, because revenue and profit rank differently here. The promotional set that moves volume can earn less per sale than a mid-range bed with a base attached, and a channel measured on revenue alone gets credited for the wrong kind of month. The ranking runs on whatever amount each row carries, so exporting margin as the amount ranks channels on profit — chosen once and kept to, because the report adds up every row in the workspace and the two should not be mixed. A store that advertises heavily on price owes itself that ranking before it decides where next quarter's budget goes.

A large share of this category's advertising is funded, at least in part, by manufacturers. Co-operative money arrives with conditions about what is shown and how, and with reporting obligations that are usually satisfied by spend and impression figures. What almost never accompanies it is a defensible account of which units the advertising actually sold, because the store has no way of producing one from a click report. So the money is spent, acquitted and renewed on evidence describing effort rather than outcome.

Adding a brand column to the sales export, mapped as Product / service, closes part of that loop. The report then shows each manufacturer's revenue over the period the co-operative funding covered and how much of it traces to a channel, drawn from the store's own delivered orders rather than from a platform's version of events. It does not break each brand down by channel, but showing that a named share of a brand's sales was produced by advertising is a considerably stronger position in a negotiation than an impression count. It is built from a file the store already runs every month for its own accounting, so nothing additional has to be collected.

It also protects the store from a subtler problem. Co-operative conditions tend to push advertising toward whichever brand is funding it, which means the media plan slowly drifts away from whatever the floor actually sells best. Seeing revenue by brand and by channel together shows when that drift has gone too far, and gives the retailer an evidence-based reason to argue for different terms rather than accepting the plan that arrived attached to the check. That conversation is difficult to have without numbers and reasonably straightforward with them in hand.