How it works
Datarithms takes your ad spend from Google, Meta, TikTok and LinkedIn and your payments from Stripe or HubSpot, ties each paying customer to the last ad they clicked, and ranks your campaigns by what one paying customer costs.
The rules are fixed and written down here, so you can check every number in your table yourself.
Which ad gets the credit for a customer?
The last ad click before the customer paid, if that click happened within 90 days. One customer counts once, for one campaign. Views without a click never get credit, and platforms do not get to count the same customer twice.
| Rule | Value | What it means |
|---|---|---|
| Matching window | 90 days | Only an ad click in the 90 days before payment can earn the credit. |
| Refund window | 14 days | A customer refunded within 14 days of paying is not counted. |
| Fewest customers for a verdict | 30 | Below 30 paying customers a campaign gets no position and no verdict. |
| Re-check | 28 days | Every change you approve is measured again four weeks later. |
Plain-terms comparison with platform reporting: last click vs platform attribution.
What is a paying customer?
A paying customer is a person or company whose payment reached you and was not refunded within 14 days. A trial, a sign-up, a form fill or a booked call is not a paying customer. If you sell through a sales pipeline, a closed-won deal in HubSpot can stand in for the payment.
What happens to customers you cannot match?
They are listed as unmatched, with their count, every week. A customer can be unmatched because they never clicked an ad, because the click was more than 90 days earlier, or because the click left no trace we can read. We never spread unmatched customers across campaigns to make the table look complete.
Why is there no verdict below 30 paying customers?
With only a handful of customers, one more or one fewer changes the cost per customer a lot, so a ranking would mostly show luck. Below 30 we show the counts and say "too few to tell". The reasoning is in how many customers you need before judging a campaign.
Who changes the ads?
You do, or you approve each change before anyone makes it. The weekly table proposes keep, cut or move budget, with the numbers behind each proposal. Nothing changes in your ad accounts without your approval, and each approved change is re-checked against paying customers after 28 days.
How a review session runs: the weekly ad budget review.
What do you need to send today?
Datarithms is in early access. Today we prepare your weekly table together with you, by hand, from export files you choose to send. Direct read-only connections to the ad platforms are being built and need each platform's approval before they can be offered.
Ad spend
A campaign cost report by day from each ad platform you use. See the platform pages.
Sign-ups with their source
Your sign-up or lead records with the ad click identifier or UTM tags captured when the person arrived.
Payments
A Stripe payments export, including refunds, or closed-won deals from HubSpot.
Step by step: how to match Stripe payments to ad clicks with exports.
Related pages
- How to calculate cost per paying customerSpend divided by customers who paid and stayed. How to count both sides honestly.
- Why platform conversions add up to more customers than you haveWhy Google, Meta, TikTok and LinkedIn together report more customers than you have.
- Ad platformsWhat Google Ads, Meta Ads, TikTok Ads and LinkedIn Ads report, how each can mislead, and which export Datarithms needs from each today.
- Questions and answersStraight answers about Datarithms: is it finished, what access it needs, whether it changes your ads, what it costs, and how the numbers are made.
Want your own table?
Tell us which ad platforms you use and roughly what you spend. We reply by email within two working days.
Early access: we prepare your weekly table with you from your own exports while the direct connections are built.