The manifesto: affiliate numbers you can verify

"Affiliate marketing you can verify." The line sits at the top of our front page, and it was not chosen because it sounds good. It is the requirement the rest of the platform is built to: every number in the dashboard must be checkable by the person it is about. This article is the manifesto behind the line – what we promise, how you can verify it, and what we deliberately do not promise.

The background is our own experience. Before we built Affilyflow, we ran affiliate sites ourselves, and the experience was the same again and again: numbers you have to take on faith. A dashboard shows a number of clicks and a number of sales, but there is no path from the number back to reality. Was the rate changed after your click? Was a sale rejected, and if so, why? How many clicks were filtered out as bots? As an affiliate you can rarely answer any of these – you can only choose to believe the screen.

The trust problem is a control problem#

Affiliate marketing has a built-in imbalance: one party measures, and the other is measured. The store and the network see all the data, the affiliate sees a summary. That is why every affiliate should ask an affiliate network four questions – including us:

• Can the commission rate be changed retroactively, after a click has been recorded?
• Can a sale be rejected without a reason and without a deadline?
• Can you see the clicks and conversions that were filtered out – or only those that counted?
• Do you know what the store pays, and what the network takes?

The rest of the article is our answer to those four questions, with reference to how the system is actually put together. If a network cannot answer them concretely, the answer in practice is no. We have set out the prices and terms of the other networks used by Danish web shops, with sources, in Partner-ads, Adtraction, Awin or Affilyflow.

The commission is locked the moment someone clicks#

When a visitor clicks an affiliate link, we copy the program's current rate and commission type into the click row itself in the database. If the click later becomes a sale, it is settled at the rate on the click – not at the program's terms on the day the sale happens. We call that rate lock, and it means a rate change can never hit traffic you have already sent.

The attribution window is also decided on our server from the click's age – the browser cannot lie about when the click happened. The store chooses its own cookie window between 14 and 90 days, and it is shown openly on the program page, so you know the term before you send the first click. The whole chain from click to conversion is described on the tracking page.

Public metrics – with a minimum threshold#

Our program pages show real metrics from real data: clicks, sales and conversion rate. But they are only shown once the program has at least 25 clicks and 3 sales. The threshold is not something we hide behind – it exists because small numbers mislead more than they inform.

A program with 2 clicks and 1 sale has a conversion rate of 50 %. The number is true, and it is still worthless – it says nothing about the program and everything about the sample size. Showing it would be advertising dressed up as data. A minimum threshold is the more honest choice: below it we say "not enough data yet" instead of displaying a percentage without substance.

The same principle applies the other way: a documented zero is anti-advertising, not transparency. A new program with 0 sales is not a bad program – it is a program without history. The threshold protects both sides of the network from hasty conclusions drawn on too little data.

The event log: including the rejected attempts#

Transparency that only shows successes is not transparency. That is why we log every single tracking event – including those that did not become a sale. A click filtered out by the bot filter is logged in a separate bot log and shown as a count on the store's Analytics. A conversion stopped by dedup because the order ID had already been counted, and an attempt without a valid click ID, are in the event log, which the store and our support use for troubleshooting. What each status means is in the docs under status codes. As an affiliate you do not see them in your own dashboard yet – if you ask us, we can look them up.

The point is simple: the question "where did my sale go?" must always be answerable with data. A system that only stores what succeeded cannot answer that question – it can only ask you to trust the filter. Our filter leaves a trail.

The maths: the shown rate to you, the fee in the store's price#

A classic source of muddy numbers in affiliate marketing is the economics between the three parties. Our model is one line: the store sets a total price per sale, our network fee is 20 % of it, and the affiliate gets the other 80 % – exactly the rate the program shows. There is no subscription – if nothing is sold, the store pays nothing.

Commission is calculated on the order value excluding VAT, so the basis is the same amount the store itself earns on – via the Shopify app, however, the value of the goods without shipping, where it is not yet confirmed whether VAT is deducted. The whole money side – balance, approval and payout – is described on the payments page.

Try it yourself: verify a sale

What we do not promise#

A manifesto about verifiable numbers must also contain the limits. We do not promise to capture 100 % of sales. Attribution rests on a click ID that travels with the visitor – if it is blocked or lost, a sale can be lost. Those are the conditions of all click-based tracking, and a number you can verify is worth more than a promise nobody can check.

Nor do we do cross-device tracking or fingerprinting. That is not a gap we are working on – it is a choice. Tracking that follows a person across devices without a click to hang it on cannot be verified by either party and brings a privacy problem we do not want to own. Our attribution is simple enough to be explained on one page, and that is the whole point.

We only promise automation where it exists: if an order from the WooCommerce plugin is refunded or cancelled, or a Stripe payment is refunded or disputed, the commission is rejected automatically. On Shopify and the other platforms, a refund requires the store to reject the sale manually – within the same approval period and with the same requirement for a reason. How the Shopify integration is built, and why it does not see refunds, is explained in affiliate marketing for Shopify stores. The difference is stated here because it exists in the system.

All of the above boils down to one principle: the order is code first, words afterwards. We do not write a promise on the site until the mechanism behind it exists and can be demonstrated – and if we cannot build a guarantee, we state the limit openly instead of dressing it up. That is not a marketing strategy. It is the only way a two-sided network can keep both sides when one party measures and the other is measured.

If you want to see the principles in action, the two ways in are become an affiliate and start an affiliate program – the whole engine room is described on the platform pages, and the article from click to payout follows a single sale all the way. If you have a question the numbers do not answer, write to us. If something cannot be verified, we are the ones with a problem – not you.