“Most of our sales have no customer attached.”
Then the report shows a large Direct / Unknown and tells you the truth about what is measurable. Emailed receipts are usually the cheapest way to change that.
For Square
Export your Square transactions, upload the inquiries beside them, and find out which marketing produced the people who walked in.
CloseRev reads a Square export of transactions — the customer's email or phone where it was captured, the amount and the date — and matches it against the calls, forms and ad clicks that brought that person in. A card-present sale is the least attributable event in marketing: there is no click, no session and no referrer, only somebody standing at a counter. Sales with no matching lead are reported as Direct / Unknown rather than assigned to whatever was running.
Page updated September 24, 2026.
The gap
You advertise to bring people through the door. They come through the door. Nothing you own connects those two sentences.
What was sold, to whom, and for how much.
The click, the keyword, the call, and what each one cost.
The file
Three things carry the match: who, how much, and when. Anything else is optional and only changes how the report can be sliced.
On the customer record attached to the transaction. This is the constraint that decides everything on this page: a sale with no customer attached cannot be matched at all.
What was taken, net of refunds. Tips are a judgment — include them consistently or exclude them consistently.
When the payment was taken. For Square Invoices, the paid date rather than the sent date.
Which site took the sale. For anybody running more than one, this is the column that turns an average into a decision.
Step by step
Written for somebody with Square open in the next tab. Report names vary by edition, so each step says what to look for.
One row per sale with a contact detail, the amount and the date. A transactions export without a customer column is the common case and it cannot be matched — check this before anything else.
This is the single number that decides whether the exercise works. If a quarter of sales carry an email or phone, that is your ceiling, and the report will say so rather than pretend otherwise.
Call tracking, the booking widget, the contact form, local search leads — in one lead file, with a contact detail and where it came from on every row.
Somebody who searched on Tuesday may come in on Saturday, or the Saturday after. Give the sales file more room than the inquiry file.
The join runs on the email and the phone, normalized. A repeat sale carries the same contact and credits the channel that first brought the customer in.
What comes back
Revenue by channel, the count of sales behind each figure, and an honest bucket for the ones nobody could trace. Sample figures, from the worked example on the Med Spas & Dermatology page — not from a Square account.
| Channel | Share | Sales | Revenue |
|---|---|---|---|
| Meta Ads | 96 | $78,200 | |
| Google Ads | 61 | $55,200 | |
| Email marketing | 44 | $32,200 | |
| Direct / Unknown | 79 | $64,400 |
Unmatched sales stay in Direct / Unknown. They are never spread across the paid channels to make the total look better.
The argument
Everything digital advertising measures depends on the purchase happening in a browser. A card-present sale has none of it: no click to attribute, no session to attach, no referrer, no landing page.
So businesses that sell in person get told, in effect, that their advertising produces nothing — or they get a proxy like store visits, which is modeled, cannot be reconciled with anything, and never appears on a bank statement.
Matching a payment export against an inquiry file is the only reading built on two things that actually happened: somebody inquired, and somebody paid. It is coarse and it is real, which is the right trade for a question that currently has no answer at all.
Square will take a payment without knowing anything about who paid. Many businesses run for years that way, and their transaction export is a list of amounts with no people attached.
Where a customer is attached — because they were emailed a receipt, booked an appointment, joined a loyalty scheme or were invoiced — the sale can be matched. Where they are not, it cannot, and no amount of cleverness changes that.
So the honest first step is counting. If the attachment rate is low, the report will show a large Direct / Unknown, and the useful conclusion is about the till process rather than about the marketing. Emailed receipts alone can move this more than any campaign change would.
Square's booking and invoicing products capture a contact detail as a matter of course. A salon appointment, a trades invoice or a quote sent by email arrives with an email address attached and matches cleanly.
An anonymous counter sale does not. That means the matched share is not evenly distributed across a business's revenue, and reading it without knowing that will mislead — the bookable side of the business will look like it is where all the marketing works.
Where the export distinguishes the two, read them apart. It is the difference between a report about your marketing and a report about your data capture.
Multi-site operators buy marketing centrally and read it centrally, and the blended cost per customer is the number that reaches the owner.
The same campaign performs very differently across sites — catchment, competition, footfall and the staff at the counter all differ — and the average describes none of them. It is entirely normal for a channel to be the cheapest customer at one site and the most expensive at another.
Where the transaction export carries a location, map it as Location and the report cuts revenue by site; the matched-records export pairs site and channel on every sale, which is where a channel's site-by-site cost shows.
Fair questions
Then the report shows a large Direct / Unknown and tells you the truth about what is measurable. Emailed receipts are usually the cheapest way to change that.
Those describe who bought. They do not describe which ad produced them, because Square never saw the ad.
Then read the bookable and invoiced side of the business, which already captures contact details, and take the walk-in revenue as unmeasured rather than misattributed.
No. Nothing is installed at the till and no credential to your account exists; the report is built from a file you downloaded.
Transactions with a customer contact detail, the amount and the date.
They cannot be matched and are reported as Direct / Unknown. The share of sales carrying a customer is the ceiling on the whole exercise.
Yes, and better than counter sales — a booking captures a contact detail as a matter of course.
Yes, and it matches cleanly. Use the paid date rather than the sent date.
Either, consistently. Mixing them across a file makes one location or one period look better for no real reason.
Include them. A channel's revenue should fall when its customers are refunded.
Yes, where the export carries one — and for a multi-site business that is usually the point of doing this.
Yes. A repeat sale carries the same contact, so it credits the channel that first brought the customer in.
Further than the inquiry file. People search days or weeks before they come in.
A contact detail, an amount and a date, plus a location if you include one. No card data, no line items, no loyalty history — and card details could not be included even if you wanted them, because they never leave Square. Encrypted in transit and at rest and deleted with the import.
Revenue by channel for sales that had no click, the share matched, and everything unmatched reported as unmatched.
By trade
What the report looks like once the export is in, written for each one.
Other systems
Running more than one system, or comparing? The method is the same and the columns are not.
Square and the other product names and logos on this page belong to their owners and are shown to identify the software a file comes from. CloseRev is not affiliated with or endorsed by them, and connects to none of them: it reads a file you export.
From the blog
The export is the part of an attribution project that everyone treats as trivial and that determines whether the whole thing works. Here is the field list, the date-range trap, the deduplication rule, and the six columns to leave out on purpose.
Read the articleEvery attribution vendor quotes a match rate and none of them define it the same way. Here is what the number actually measures, what range is honest for a mid-market business, and the three ways the figure gets inflated before it reaches your slide.
Read the articleOne tracks individuals and cannot see causation. The other infers causation and cannot see individuals. Vendors sell them as competitors, which is convenient for whoever is selling. Here is what each can and cannot tell you, and what a mid-market business should do instead of buying either.
Read the articleStart today
Nothing to install in Square, no API key, and no need to have been tracking anything until now. Last year works as well as this month.