How to Start an Affiliate Program (Decide These Six Things Before You Pick the Software)

An affiliate program is the only marketing channel where you pay after it works. No ad budget burned on clicks that didn't convert; a commission exists because a sale does. That's why everyone wants one, and it's also why half the programs I've seen are ghost towns: "pay after it works" sounds like free money, so people switch the feature on, pick a commission number off a competitor's site, and wait. Nothing happens. The software was never the hard part.

I've been running affiliate programs on my own products since the 2000s: close to 100 affiliates on a self-hosted Post Affiliate Pro install, a two-tier program on EzyGold, about $250,000 sold through JVZoo, and programs on Lemon Squeezy and Systeme.io for my current software. This is the guide I'd hand myself before the first one.

Do you need an affiliate program at all?

Here's the test, with a number: has your product sold at least 50 copies to strangers, meaning people who aren't your list, your friends, or your launch spike? If not, don't start a program yet.

Affiliates amplify a product that already converts. They don't fix one that doesn't. An affiliate sends you traffic, the traffic doesn't buy, the affiliate leaves, and now your pitch to the next affiliate is a program with visible zeroes in it. I learned this with money: years ago I put products on JVZoo, WarriorPlus, PayDotCom and the Warrior Forum expecting the resident affiliates to pick them up, and they mostly didn't, because affiliates follow proof, not listings. One digital product of mine did sell well on JVZoo, and it happened only after the product caught traction on Facebook first. The affiliates showed up for a product that was already moving. Nobody joins a queue with no one in it; that's social proof working against you until you have some.

The other requirement is margin. If paying out 30% of a sale makes you wince, your price is wrong or the channel is wrong. Work that out before anyone promotes you.

The six decisions that matter more than the tool

Every tool in the directory will ask you the same questions in its setup screens, and the defaults are nobody's friend. Decide these on paper first.

1. The commission rate, anchored to lifetime value. The going rates: 20% to 30% recurring for SaaS, 30% to 50% for courses and digital products (no marginal cost, so you're splitting profit, not revenue), and single digits to about 20% for physical goods depending on margin. The arithmetic that matters isn't "what does this sale cost me," it's "what is a referred customer worth over their lifetime, and what share of that buys an affiliate's sustained attention." A generous rate on a product that retains is cheaper than a stingy rate plus the recruiting you'll have to do to replace bored affiliates.

2. Recurring or first-payment-only. Recurring commissions turn your affiliates into stakeholders in your churn rate; one-time commissions buy a burst of promotion and no loyalty. For subscriptions, recurring for 12 months is the SaaS norm you'll be compared against. Cap it in time, not because affiliates don't deserve more, but because "forever" liabilities make your business harder to sell later.

3. The cookie window. 30 days is standard, 60 to 90 reads as generous, and affiliates do check. Last-click attribution is the default everywhere. Whatever you pick, write it where affiliates can see it; an unpublished cookie window, like Kartra's, is the kind of thing that ends up on a site like mine.

4. Payout schedule, minimum, and the refund buffer. Pay monthly, set the minimum low ($25 to $50; a $100 minimum tells small affiliates not to bother), and hold commissions for your refund window plus a few days before they're payable. That delay is standard and nobody resents it if you state it up front. Decide who pays the PayPal fees too. ClickFunnels makes the affiliate pay them; affiliates notice.

5. Terms that protect you. No self-referrals, no bidding on your brand name in ads, coupon rules (whether coupon and cashback sites can join at all), and the right to void commissions for fraud. Write these before launch. Adding rules after someone's exploited the gap means voiding commissions you already announced, which is how programs end up with a reputation.

6. The tax and paperwork question. US-based and paying US affiliates over $600? That's a 1099 problem. Most tools don't handle it (FirstPromoter does on its top tier, Reditus and Refersion have managed options). Know whose job it is before January, not in January.

Now pick the room

The tool question is really a "where does your checkout live" question, and the directory is organised the same way:

• Shopify or WooCommerce: GoAffPro free, UpPromote cheap, Social Snowball if you want customers auto-enrolled as affiliates at checkout.
• SaaS on Stripe: Rewardful, Tolt, FirstPromoter, or Affonso at €15 while you're small. The standalone software category compares them.
• Courses and funnels: the platform you're on probably has a built-in program. ThriveCart's and Kajabi's are genuinely usable; others gate the feature behind expensive plans, and each listing does that arithmetic.
• Launch-world digital products: JVZoo, WarriorPlus and ClickBank come with affiliates already in the building, and my listings explain why that's less of an advantage than it sounds.
• Want to own the data outright: a self-hosted install like iDevAffiliate. You become the IT department, which is a fair trade some of the time.
• Doing $20,000+ a month in affiliate sales already: that's when networks enter the picture, and the DTC networks guide is the full treatment.

One distinction that saves people a wrong purchase: if the people promoting you will be your existing customers sharing a link for a discount or credit, that's a referral program, not an affiliate program, and tools like ReferralCandy and Referral Rock are built for it. Affiliates are marketers paid in cash; advocates are customers thanked in kind. Mixing up the two is how you end up paying cash commissions to people who would have recommended you for free.

Whatever you pick, read its lock-in score before you commit. You're not choosing a tool, you're choosing the container your affiliate relationships will live in, and some containers don't open.

Launch with ten affiliates, not a signup page

The program launch that works is embarrassingly manual. Make a list: customers who already rave about you, peers with adjacent audiences, anyone who has ever linked to you or mentioned you unprompted (search your brand name), and the newsletters and YouTube channels your customers actually read and watch. Email them one at a time, personally, with the rate, the cookie window, and one piece of proof the product converts. Ten active affiliates recruited by hand beat four hundred signups from a directory, because the four hundred were collecting logins, and the ten said yes to a person.

This is the oldest play in the category. When Amazon launched Associates in July 1996, it didn't wait for bookish website owners to discover a signup form; the pitch was put in front of people who were already writing about books. The program that invented the industry was recruited, not posted.

Give them something to work with

The day an affiliate says yes, they should receive: their link, two or three email swipes and social posts in your actual voice (they'll edit them, but a blank page kills momentum), a one-paragraph description of who buys and why, your best-converting angle, and the payment terms restated in plain words. Then pay the first commissions fast and visibly. The first payout is the moment an affiliate decides you're real; nothing in your onboarding sequence does what a $40 PayPal notification does.

How I'd decide

Under 50 sales to strangers: don't start a program. Fix conversion first, and spend the energy getting proof.

Selling on Shopify: GoAffPro free today, revisit when you outgrow it.

SaaS on Stripe, bootstrapped: Affonso or Tolt now, FirstPromoter or Rewardful when the program is paying for itself.

Courses on an all-in-one platform: use the built-in if you're already on the plan that includes it; the listings show which plans those are. Don't upgrade a plan just for affiliates without doing the math against a $49 standalone.

Everything decided but no time to wire it: the six decisions above, the tool setup, the terms and the onboarding page are exactly what the setup service delivers. You bring the product and the proof; I'll build the program around them.

One more thing

Your affiliate program is a product, and its customers are affiliates. It has an offer (the commission), an onboarding, a retention problem and competitors (every other program those affiliates could spend the same hour promoting). The merchants who win treat recruitment like sales and payouts like customer service. The ones who treat the program as a feature they switched on are the ghost towns from the first paragraph, and their software subscription renews either way. Start when you have proof, decide the six things on paper, pick the container with the exit in mind, and go recruit your first ten by hand.

Published November 6, 2026. Commission and cookie norms are drawn from the pricing and terms research behind the directory listings, verified on the dates shown on each. Amazon Associates launch date (July 1996) is public record. Program-running history is the author's own: Post Affiliate Pro, EzyGold, JVZoo, WarriorPlus, Lemon Squeezy and Systeme.io, from the 2000s to now. Some links on this page are affiliate links.