Most analytics stop at the click. They tell you a campaign drove sessions, then go quiet about whether those sessions ever became money. Revenue attribution closes that loop: it puts a source next to every sale, so you can rank channels by what they earned instead of what they sent. The usual assumption is that this needs a cookie that follows the customer from the ad to the receipt. It does not. What it needs is one short-lived id that travels with the checkout and comes back with the payment. This is the how-to: how it works without cookies, which visit gets the credit, what happens to renewals and refunds, and where the method stops. For what revenue attribution is and which model to pick, start with the [revenue attribution guide](/blog/revenue-attribution).
What does revenue attribution actually have to do?
Strip attribution down and it answers one question per sale: which visit led here, and how did that visit begin? Everything else, revenue by source, by campaign, by referrer, by channel, is a sum over those answers.
A cookie is one way to carry the answer across pages and days. It is not the only way, and it comes with costs you pay in data: consent banners, browsers that cap cookie lifetimes and blockers that strip them all cut the trail before the payment arrives. A method that never writes to the device does not depend on a cookie surviving until the payment.
Vero writes nothing to the visitor’s device. No cookie, no localStorage, no stored identifier, and the IP is never stored. Visitors are counted with a salted SHA-256 hash that rotates daily. The link between a visit and a sale is made somewhere else entirely: inside the payment.
The visit id rides the checkout
Every time the tracker boots on a page, it holds an ephemeral session id in memory. When the customer goes to pay, that id is handed to the payment provider as part of the checkout, in a field the provider already offers for exactly this kind of reference.
On Stripe, it is the client_reference_id of a Payment Link, or a metadata.vero field, set on the checkout or on its subscription. On Lemon Squeezy, the checkout link carries it as checkout[custom][vero], and the webhook returns it in meta.custom_data.vero. On Paddle, you open Paddle.js with customData.vero, and the webhook returns it in data.custom_data.vero. For Stripe Payment Links and Lemon Squeezy checkout links, the tracker’s checkout module adds the id on the click, so the link itself does the work.
When the payment completes, the provider sends a signed webhook to Vero. Vero checks the signature, reads the amount and the id, and links the sale to the visit that id names, all on the server. An id counts only if it names a visit of that site, so a stray or copied value attributes nothing. If you run your own checkout, your server posts the sale with the id to the ingest endpoint instead, and the rest is the same.
This post covers the idea. The step-by-step guide to connecting a provider covers the setup, one provider at a time.
Why doesn’t the checkout page get the credit?
A sale almost never happens on the page that introduced the customer. Someone lands from a link, reads two pages, opens pricing and clicks buy. The tracker mints a new session id on every script boot, so the page where they clicked buy is a different boot from the page where they arrived, and it carries no campaign of its own.
Worse, its referrer is your own site. A referrer on the site itself, or on one of its subdomains, counts as no referrer, so that boot reads as direct. Crediting each payment to its own boot would score almost every sale as direct and leave your real sources with nothing.
So Vero takes two hops. First, the payment identifies a visitor: the boot it names carries the same daily visitor hash as the other pages that person loaded. Second, the visitor plus the gap of inactivity identifies the visit that boot belongs to. The first event of that visit carries the source: its utm_source, else the host of its referrer, else direct.
That is first-touch attribution, and it is the same rule the traffic reports use. A visit is counted under one source on the Sources screen, and the money it made is counted under that same source on the revenue screen, so money and traffic never disagree on what a source is.
What counts as a sale
A sale is what was charged, without tax, less refunds, before fees. Vero takes the tax out using the figure each provider sends, and it never subtracts a fee, because it does not know what the provider charged you. Read it as what your customers paid you for the product, not what landed in your bank.
Refunds come off the sale they belong to, so they come off that sale’s source too. A channel that sends buyers who ask for their money back ends up scored on what it actually kept, not on the sales it once looked good for. Disputes, also called chargebacks, are not deducted: Vero’s revenue webhooks handle refunds only, so a payment you lose in a dispute still counts until you refund it.
A few things never become revenue at all. A sale that charged nothing, such as a trial’s first invoice or a free order, is not stored. Test payments are acknowledged and not counted. A webhook whose signature is more than 5 minutes off is refused, so an old payload cannot be replayed into your numbers.
Each sale stores its visit’s source, campaign, referrer, channel, landing page and first-touch time when it is received. A sale from March keeps the source it had in March, whatever date range you read it in.
Renewals inherit the first payment’s source
A subscription renews without anyone visiting your site. The renewal webhook arrives with no visit id, because there was no visit. Left alone, every renewal would land with no source, and the channels that bring long-lived customers would look exactly like the ones that bring one-month churners.
So a sale without a visit id inherits the source of that customer’s first attributed payment. The customer who came from a newsletter in January is still credited to the newsletter when month six is billed. Recurring revenue stays with the source that earned it, and refunds of a renewal come off that same source.
A worked example: the demo week
The public Vero demo is a product demo, not live data: a fictional SaaS in its launch week, with simulated numbers. The week brings a Hacker News front page on Tuesday and Product Hunt on Thursday. Over 7 days it records €18,423 of revenue from 187 transactions, an average order of €98.52. The launch week analytics post walks through how to read that kind of week.
By source, news.ycombinator.com is on top with €7,124 from 72 payments. Next is twitter with €4,821 from 49, then producthunt €1,845, google €1,783 from brand_search ads, google.com €1,000 from organic search, github.com €968, newsletter €482 and chatgpt €400. They add up to the week’s €18,423.
Put visits next to the money and the picture changes. news.ycombinator.com sent 3,420 visits for its €7,124. twitter sent 1,245 visits, fewer than half as many, and still earned €4,821. producthunt sent 1,632 visits, more than twitter, and earned €1,845. The biggest traffic spike was not the best audience, and a visit from twitter was worth more than a visit from either launch site. That is the case for ranking sources by revenue instead of visits.
Two details show the rules at work. Direct is the largest line of visits in the demo, 3,535, yet no sale is credited to direct, because a sale goes to how its visit began, not to the checkout page. And newsletter keeps its €482 even though an email client sends no referrer: its links carry a utm_source, and the source is read from that first.
Where this method stops
The visitor hash rotates daily. That is what keeps it from identifying anyone across days, and it is also a limit: a sale is credited to a visit from the same day. Someone who reads your post on Monday and buys on Friday is credited to how Friday’s visit began.
A sale with no visit on record has no source. That covers a checkout opened from a link that never carried the id, a payment made on a device the tracker never saw, and a first payment with nothing to inherit from. The sale still counts in your revenue, it just sits outside every source.
An ad blocker can stop the tracking script. When it does, there is no session id to hand to the checkout, and that customer’s sale arrives with no source. Nothing about avoiding cookies changes that.
First touch is one model, not the truth. It credits how the visit that ended in a purchase began. It does not split credit across earlier visits, and it does not see an ad someone saw and never clicked. Read revenue by source as a ranking of where paying visits started, then subtract what each source cost before deciding where the next euro goes.
Unsure of a term? Read the glossary

Vero is built by one person in Spain. I read every waitlist entry and pick the beta accounts myself.