One flat commission
Pick a single recurring rate so payouts are trivial to calculate and explain.
How-to — — by Mahmoud Zalt
How to build an affiliate program without a manager: a small program, AI Employees for recruitment and payouts and reporting, plus a clean weekly founder routine.
Most solo founders are pitched an affiliate manager before they have ten paying customers, which is the wrong order of operations. A real affiliate manager costs a meaningful chunk of monthly revenue and only pays back once you have dozens of active partners pushing volume. Before that point, the manager is mostly polishing assets, chasing applications, and answering the same three questions over email, which is exactly the kind of work that does not need a human. The smarter early move is to keep the program tiny, let AI Employees handle the repeatable layer, and protect your time for the parts of partner relationships that genuinely require a human voice. A solo founder who runs the program themselves for a few months also learns which partner types convert, which is information no hire can give you on day one.
The smallest viable program has five moving parts and nothing more. You pick one commission rate, one tracking method, one payout cadence, one application form, and one welcome pack. That is it. Resist the urge to ship tiers, leaderboards, contests, or seasonal promos in version one, because each of those is a separate workflow that needs eyes on it later. The goal of the first version is not optics, it is to prove that real partners can find you, sign up without friction, share a link that tracks correctly, and get paid on time. Once those five pieces work for ten partners across a full month, you have earned the right to add tiers and bonuses. Before then, complexity is your enemy.
Pick a single recurring rate so payouts are trivial to calculate and explain.
Use a UTM plus a discount code per partner. No custom dashboards yet.
Monthly on a fixed date by Stripe, Wise, or PayPal. Predictability builds trust.
Five fields max, gating on real audience or use case, not a marketing essay.
A short doc with link, code, two assets, and a single ask for the first promo.
Yes, with the right division of labor between AI and the founder. AI Employees are excellent at the structured, repeatable layer of an affiliate program, which is most of the work in the first year. They can sort applications, write personalized welcome notes, reconcile sales to partners, draft payout summaries, and produce a weekly digest that lands in your inbox every Monday. What AI should not do alone is approve a borderline applicant, send the one-off thank you that wins a partner over, or negotiate a custom deal with a creator who can move real volume. Treat the AI as your operations layer and yourself as the relationship layer, and the program quietly compounds while you keep building the product.
The first two months are mostly tuning. You read the AI's triage scores against your own gut, correct the rubric where you disagree, and trim the welcome template until it sounds like you in three sentences. After that, your weekly touch becomes a quick scan of the digest, one round of approvals, and three or four personal notes to the partners who actually moved volume. The cadence is slow on purpose. Partner relationships compound when they feel consistent, not when they feel busy, and AI Employees are unusually good at being consistent.
Once the operational loop is humming, the real growth lever shifts from process to relationships. The partners who are quietly sending three or four sales a month are the ones worth investing a personal hour into, because turning them into ten-sales-a-month partners is cheaper than recruiting brand new ones. The next two sections cover how to keep the top tier engaged without daily babysitting, and what a sustainable weekly routine looks like once the program runs itself.
High touch and high frequency are different things. The partners who matter want to feel seen and to know the product is moving forward, not to receive ten emails a week. The pattern that works for solo founders is a steady drumbeat of small, useful, predictable touches that an AI Employee can prepare and you can approve in minutes. A monthly performance note with their actual numbers, an early heads up on the next product release, an occasional ask for a quote, and a yearly thank you with a real gift. Four moments a year, executed well, beats forty emails that feel like a CRM blast. The AI prepares, the founder sends, and the relationship deepens on a schedule the program can actually keep.
AI drafts a personal one-paragraph note with the partner's clicks, conversions, and earnings.
Top partners see new features one week before launch, with copy and assets ready to share.
Quarterly, AI asks for a one-line quote and offers a feature on the partner page in return.
One physical thank you a year for partners over a threshold. AI tracks the list and reminds you.
A clean weekly routine is the difference between a program that runs and a program that drifts. The version that has worked for me as a solo founder fits inside one hour a week, sits on Monday morning, and ends with three personal notes sent before lunch. AI does the heavy lifting around it: pulls the numbers on Sunday night, triages new applications, reconciles the prior week, and queues drafts. You open one tab, scan one digest, and approve or rewrite. The discipline is to never let the routine slip a week, because partners forgive small payouts and they forgive shy outreach, but they do not forgive silence. The schedule, kept religiously, is the program.
For SaaS, 20 to 30 percent recurring for the first year is the common range, dropping to 10 to 15 percent in year two. For one-off purchases, 10 to 20 percent flat is normal. Pick a rate you can pay forever without renegotiating, because the worst affiliate move is cutting commissions once partners are committed.
Yes, with a clear rubric. Give the AI Employee three to five criteria (audience overlap, content quality, traffic source, previous affiliate history), have it score every applicant, and surface only the borderline cases for your decision. The clear yes and clear no decisions can be fully automated once the rubric is calibrated.
Generally no. Caps signal distrust and chase away the best partners, who are the ones who would have earned the most. If you must cap to protect margins on a launch, cap by campaign rather than by partner, and tell partners up front before they invest the work.
Three guards cover most cases: no self-referral on the same email or IP, a thirty day cookie cool-off before commission counts, and AI flags on unusual conversion patterns (a flood of cheap accounts, refunds spiking on one code). Manual review on flags, no public accusations, quiet removal with the earned balance honored.
When the program crosses roughly thirty active partners producing meaningful monthly volume, when negotiation and custom deals become weekly work, and when your own time on the program exceeds five hours a week despite AI handling the loop. Below that bar, a manager is an expense, not a multiplier.
If you want a wider playbook for how AI Employees fit into the rest of the growth stack (content, outbound, lifecycle, and retention next to the affiliate program), the companion read walks through the lean stack I run on my own business as a solo founder. It is the operating manual the affiliate program plugs into, so the partner channel does not become a silo.
The honest framing is simple. An affiliate program does not fail because you skipped hiring a manager. It fails because nobody runs the boring middle: applications stack up unanswered, payouts slip, top partners stop hearing from you, and the energy dies. AI Employees solve that middle layer cheaply and consistently, which is the layer that breaks first when a founder is wearing seven hats. You still need taste on who to approve, voice on how to thank, and judgement on when to bend rules for a partner who can move real volume. Keep those three on your plate, push everything else into the AI loop, and you will have a program that ships value to partners every week without becoming the job you most resent. That is the version that survives long enough to deserve a human manager later.