The orders that don't close in the cart.

For retailers selling anything configured, fitted, delivered or expensive, a meaningful share of orders never goes through the cart. The customer researches online and then calls, emails a quote request, or walks in.

That revenue is not small and it is not random — it skews toward the largest orders, because the bigger the purchase the more likely someone wants to speak to a person first. Measuring only what the pixel sees systematically undervalues whichever channels produce considered purchases.

Keep your pixel and your on-site analytics. They do their job well for the checkout, and CloseRev makes no attempt to replicate them.

What it adds is the other half: an export of phone, quote and in-store orders matched against your ad and call data, so the channel report finally covers all of the revenue rather than the convenient part of it.

Open three advertising dashboards for the same month and add the conversions together. The total will exceed the orders in your own system, often substantially. Each platform counts a sale it believes it influenced, using its own rules about what qualifies and how long the credit lasts, and none of them is under any obligation to reconcile with the others. That is not dishonesty, it is three separate measurements of overlapping populations presented as though they were one ledger. The retailer is left holding three answers and one bank statement.

Consent choices, cross-device shopping and the plain fact that people research on a phone and buy on a laptop widen the gap further. Whatever the causes, the practical effect is identical: the platform figures cannot be summed, cannot be compared like for like, and cannot settle an argument about where the next dollar of budget should go, because each one is answering the question from inside its own boundary. A number only its own author can compute is not a shared measurement of anything.

Matching your own order export against the lead and click exports from each platform produces one ledger instead of three. Every order is counted once, against one source, and the total ties back to what you actually sold in the period. Orders that cannot be traced stay in Direct / Unknown rather than being divided up to improve the attributed share. That single constraint is why the resulting numbers can go in front of a finance director without a paragraph of explanation attached.

Gross revenue by channel is the wrong ranking for most retailers, and it is usually the only one they have. Two channels can produce identical revenue while one sells full-price stock to people who keep it and the other sells discounted stock to people who send half of it back. After returns and after the cost of goods those are not the same business at all, and the difference between them is frequently larger than anything a bid adjustment could ever deliver.

Both corrections happen in the export rather than in a project. Export net of returns, because the report counts the amount each row carries and does not subtract a returns column. Profit works the same way: the ranking runs on whatever amount you export, so a workspace that exports margin as the amount ranks channels on profit instead of revenue. Choose one and keep to it, because the report adds up every row in the workspace and margin and revenue should not be mixed. A margin ranking can disagree with a revenue ranking, sometimes right at the top, and the channel that looked best on revenue can turn out to be the most expensive way you have of moving stock.

Category, mapped as Product / service, is the breakdown worth adding, because a channel does not have a returns rate; the products it sells do. The report then shows revenue by category with how much of each traces to a channel, which turns an unflattering total into a question about which lines are selling and which are being advertised. It does not rank channels within each category, so it is a starting point rather than a verdict on any one channel.

Acquisition cost only means something against what a customer is eventually worth, and for anything consumable, replaceable or collectable the first order is a poor guide to that. The customer who buys once and the customer who buys four times a year for three years cost the same to acquire and are worth nothing like the same amount. Ranking channels on first-order revenue therefore rewards whichever source is cheapest at producing a single transaction, which is rarely the source producing customers worth having.

Because the match is on the person, every later order credits the channel that produced them originally, with no window in between and no cookie that needed to survive. Export two or three years of orders rather than one month and the ranking rearranges itself. Sources that looked expensive rise because their customers came back; sources that looked efficient fall because they were buying one-off purchases from people shopping on price alone. That reordering is usually larger than any optimization available inside an ad account.

Subscriptions and replenishment make the effect starker still. A recurring order credited every month to whatever campaign was running that month makes recent advertising look remarkable and the acquisition campaign look pointless, which is precisely backwards. Attaching the whole stream to the source that produced the subscriber is the only reading that describes what happened, and it is what lets a retailer justify spending more to acquire a customer than the first order is worth on its own. That decision is impossible to justify from a report built on single transactions.