How to start an affiliate program for your SaaS

Most guides on this subject open by selling the channel. This one opens with the cases where it fails, because they are predictable and because discovering them after three months of setup is expensive.

• Your product is hard to describe in one sentence. An affiliate has to explain it to their audience in a paragraph. If it takes a demo call, they will not attempt it.
• Your customers churn in month two. A recurring commission on a customer who leaves immediately is worth nothing to the affiliate, and they will notice within a quarter.
• Your price is very low. A 20% commission on €9/month is €1.80. No one builds content for that, however good your product is.
• You cannot answer "why you". An affiliate is putting their own credibility on the line. If they cannot tell their audience why your tool rather than the obvious alternative, they will pick something else to promote.

Everything else is configuration. These four shape what the program becomes, and changing them later is possible but costly in trust.

1. Percentage or flat amount. A percentage scales with your prices and rewards affiliates who bring bigger customers. A flat amount is easier to communicate and suits a single-price product. Percentage is the norm on SaaS, for the simple reason that it keeps working when you change your pricing.

2. One-off or recurring.This is the decision that changes the affiliate's behaviour most, because it changes what they are being paid for — a signup, or a customer who stays. The dedicated guide works through the maths and the cases that lower the total.

3. The cookie window. How long after a click a sale still counts. Thirty days is the common default; a product with a long evaluation cycle deserves more, and setting it too short quietly loses sales the affiliate did generate.

4. The approval window before payout. The delay between a commission being earned and becoming payable, so a refund can still arrive. Too short and you pay on revenue you will lose; too long and affiliates wait for money they have earned.

Two lines, and only one of them is fixed. Commissions only leave on sales you have collected: they cannot put you in trouble, and they rise with the revenue that produced them. The tool is a fixed cost that lands whether the program works or not — the only real financial risk of this channel in its first months.

The market's rough shape: established tools start around $49/month, with an affiliate revenue cap that pushes you to the next tier — usually somewhere between $5,000 and $10,000 a month. Two of them offer a permanent free plan, capped as well. Our comparison pageslist each one's public pricing, with the date it was checked.

Three questions beat a feature list when choosing:

• What you bill on. If it is not Stripe, the field narrows sharply — several tools know nothing else.
• What triggers the next tier. Rarely the number of affiliates; almost always the affiliate revenue tracked. Read that cap before the headline price.
• What is taken on top of the subscription. Some take a percentage of affiliate revenue, others a fee on payouts. Neither is disqualifying, but both belong in the total.

The commission calculator works out the commissions side over twelve months from your own price and rate. The subscription and any fees add on top of that.

This is where most people expect the difficulty, and it is not there. Three things have to be true: a click has to be remembered, a sale has to be recognised, and the two have to be joined.

• A tracking snippet on your site, which remembers which affiliate link a visitor arrived through. One script tag.
• A webhook on your billing, which says when someone paid. On Stripe that is one endpoint to add.
• A portal for the affiliates, where they get their link and watch their numbers. Nothing to build — it comes with the tool.

The Stripe integration page covers the detail, including the route that needs no backend code at all and the trap that silently stops recurring commissions if it is handled naively.

More common than it sounds, and it carries a trap that costs real money. You switch tools: the new one knows none of your existing subscribers. Their renewals keep arriving, but with no attribution — no cookie, because the script was not there, and no initial sale, because it happened before you arrived.

The consequence is that the affiliates who brought those customers stop being paid on the day you migrate, and nothing anywhere reports an error. They notice before you do.

What to check before moving: that the tool can reattach an already-running subscription to its affiliate from whatever evidence still exists — metadata left on the subscription, or a tracked coupon applied to it. Whatever cannot be proved has to be decided by hand. And tell the affiliates concerned: a commission that vanishes without explanation is the fastest way to lose the ones who produce.

Tracking is solved. Recruiting is not, and it is where programs die. A tool cannot make someone want to recommend your product — it can only make it easy once they do.

Three sources, in the order most founders actually use them:

• Your existing customers. The shortest path: they already use the product and can describe it honestly. Ask the ones who wrote to say thank you.
• People already writing about your category. Newsletter authors, YouTubers, community moderators. Approach them with a specific reason they should care, not a template.
• A marketplace, where affiliates are already looking for products to promote. That reverses the direction of the effort — instead of finding them, you become findable. Ours is open to browse, and the affiliate side explains what they see.

On that second point, the message matters more than the list. What a creator wants to know fits in three lines: why them specifically — quote something they published, not their "niche" — what their audience gets out of it, and what they earn, as a real amount rather than an abstract percentage. "20% recurring" means nothing; "around €120 per customer over a year" is understood in a second. Give free access too: nobody recommends a product they have not opened.

The timeline matters because most people who give up do so on false expectations rather than on bad results.

• Week 1 — setup. Script, webhook, written rules, and one test sale carried all the way through to a commission. This is the short part.
• Weeks 2 to 4 — the first affiliates, who almost always come from your own customers. The goal is one affiliate who publishes something, not ten who sign up.
• Month 2 — the first clicks, and often no sales. That is normal and not a signal. Look at EPC and conversion rate rather than the total.
• Month 3 — the first payout. This is where the machinery is really tested: threshold, approval delay, invoice, missing bank details.

What separates a program that will work from one that will not is rarely visible before the fourth month, and never in the number of sign-ups. The useful signal is repetition: an affiliate who publishes a second time without being asked.

Commissions accumulate, cross a minimum threshold, and go out as a payout. That part is mechanical. What surprises people is the paperwork: an affiliate is a supplier, and a supplier issues an invoice.

Across borders that invoice has a VAT treatment that depends on both countries and on whether each side holds a VAT number. The dedicated guide covers the four situations. It is worth reading before your first payout rather than after, because the numbering has to be continuous from the first invoice.

Two categories of problem, and they are not solved the same way. Fraud is technical: someone buying through their own link, fake traffic, coupons posted where they should not be. Those are detected and blocked automatically.

The second is a conflict of interest, and no algorithm settles it: an affiliate bidding on your brand name captures customers who were already coming. The answer is explicit program rules, stated before someone joins, plus the ability to suspend or block. Suspending should stop new clicks and hold payouts without erasing what was legitimately earned — those are different levers and they should stay different.

The number of affiliates is the vanity metric of this channel. A program with two hundred registered affiliates and four active ones is a program with four affiliates.

• Active affiliates — how many produced at least one click this month. This is the number that tells you whether the program is alive.
• Revenue per affiliate — almost always concentrated. Knowing which two people carry the program tells you who to talk to.
• Retention of referred customers — compared to your other channels. If affiliate customers churn faster, something in the promise is off.
• Total commission cost against your other acquisition cost — the only comparison that decides whether to keep going.

• A rate set too low out of caution. It costs nothing and produces nothing, and you conclude the channel does not work.
• Opening and then leaving it alone. Affiliates who hear nothing for three months stop promoting. A short update every month is the whole job.
• No written rules. The day someone bids on your brand, you have no ground to stand on unless it was written before they joined.
• Paying late. This is the fastest way to lose the affiliates who work. Whatever the threshold, honour it on time.
• Recruiting anyone. An affiliate whose audience has nothing to do with your product brings traffic that never converts, and costs you the time you spend looking at it.