How to Launch an Affiliate Program: A Growth Marketer's First 90 Days
Turn this article into takeaways for your work.
Each assistant summarizes the article only for you and suggests best practices for your work.
Search for "affiliate marketing" and nearly every result is written for someone who wants to earn commissions: pick a niche, join a network, start a blog. That's the wrong side of the table for you. You're the growth marketer who has been told to build the program, which makes you the merchant, the one who pays.
The merchant's job is different, and so are the mistakes. Nobody asks you to build a following. They ask you to produce customers at a cost leadership can defend, without a compliance problem, without a fraud problem, and without a program that quietly pays people for sales that would have happened anyway.
This guide is a 90-day plan for that job. It covers the economics you settle first, the commission model choices, tracking and attribution, the legal disclosure rules that land on you, the recruiting motion, and what to put in front of leadership at day 90. It's written for B2B SaaS and ecommerce, where the mechanics are the same and the economics differ.
Key Facts
- You carry the legal risk. The FTC states that "your company is ultimately responsible for what others do on your behalf," and expects advertisers to train and monitor their affiliates.
- Disclosure has to sit next to the recommendation. The FTC says the closer a disclosure is to the endorsement, the better, and that one buried in the middle or end of a post is "easier to miss."
- Vague labels don't count. The FTC says "affiliate link" alone may not be adequate and "commissionable link" is "probably not a clear disclosure."
- Affiliate links should be marked
sponsored. Google's guidance on outbound links says paid placements should use thesponsoredlink value. - Thin affiliate pages are a search risk. Google's spam policies treat copied merchant descriptions as thin affiliation, so recruiting for quality protects your brand as well as your budget.
- Day 30 decides the program. Economics, terms and the tooling decision are made in the first month. Everything after that is execution.
First, Decide If This Is Affiliate, Referral or Reseller
Before you open a spreadsheet, check that "affiliate" is the right channel. The three get lumped together and they behave very differently. An affiliate promotes you to an audience they own and gets paid per outcome. A referral partner introduces someone they know personally. A reseller buys and sells under their own commercial terms and owns the customer relationship.
The distinction matters because it changes who you recruit, what you pay, and how you attribute. Our breakdown of referral vs affiliate vs reseller is the quick way to settle it. If your product needs a consultative sale with implementation, a reseller or referral motion usually fits better. If your buyer researches alone and converts on a pricing page, affiliate fits.
If you're in ecommerce, the case is more settled, and the channel mechanics are covered in building an affiliate marketing program. For B2B, the next section is where most plans quietly fail.
Days 1 to 30: Economics, Terms and the Tooling Decision
Work out what you can afford to pay
Start from your own numbers, not from what competitors seem to pay. Three inputs matter:
- Contribution margin per customer. For SaaS that's lifetime gross profit, not first-month revenue. For ecommerce it's first-order margin after returns and shipping, plus an honest repeat-purchase estimate.
- Your current blended acquisition cost. The affiliate channel has to beat or match your marginal paid channel, not your average one.
- Payback tolerance. Ask finance how many months of payback they'll accept. A program that pays out immediately on a customer who takes 14 months to recoup is a cash decision, not a marketing one.
Then do the arithmetic the other way round. If your allowable acquisition cost for a customer is X, your maximum commission is X minus whatever you spend on tooling, partner management and the fraud you'll inevitably pay for. Most first-time programs forget the last two lines.
We won't print a "standard" commission rate here. Rates vary by category, price point and sales cycle, and any single figure would be a guess dressed up as a benchmark. Your allowable-cost math is the benchmark that matters.
Choose a commission model on purpose
There are four models you'll realistically weigh. Each rewards a different behavior and invites a different kind of abuse.
| Model | What you pay for | Works best when | Main risk |
|---|---|---|---|
| One-time percentage of first sale | A single conversion | Ecommerce, low-consideration purchases | Affiliates optimize for volume, not retention |
| Flat bounty per customer | A qualified outcome | Fixed-price products, free trial to paid | Easy to game with low-quality sign-ups |
| Recurring percentage | Revenue over a set period | Subscription products | Higher long-term cost; needs clear end terms |
| Tiered or performance-based | Higher rate at higher volume or quality | Mature program with proven partners | Complexity, and tier fights over borderline cases |
For a new B2B SaaS program, a time-limited recurring share or a bounty on a paid, activated account is usually the defensible starting point, because both tie payment to a customer who stuck. Paying on trial sign-ups is the most common way to lose money in month two. If you want a fuller view of incentive design across partner types, partner incentives goes through the options.
Whatever you choose, pay on a delay. A validation window before payout (long enough to cover refunds, chargebacks and trial conversion) is the cheapest fraud control you'll ever install.
Write the terms before you recruit
Your program terms are the document you'll point to when something goes wrong, so write them while everyone is still friendly. At minimum:
- Commission definition: what counts as a qualified conversion, the validation window, and what's excluded (refunds, cancellations, self-referrals, existing customers).
- Attribution rules: cookie or lookback window, how conflicts between affiliates are resolved, and whether last click wins.
- Permitted and prohibited promotion: brand-term bidding, coupon sites, trademark use, email sending, incentivized traffic, adult or misleading content.
- Disclosure requirement: affiliates must disclose the relationship clearly (more on that below).
- Your right to terminate and claw back commissions for violations.
A lightweight partner agreement template covers most of this. Have counsel review it once, because the claw-back and disclosure clauses are the ones that get tested.
Decide how you'll track and what you'll buy
This is the tooling decision, and it's mostly a tracking decision. Ask three questions:
- Can it track the conversion you actually care about? For SaaS that's a paid, activated account, which often happens days or weeks after the click, in your billing system rather than on a thank-you page. Confirm the tool can accept server-side or billing-event conversions, not just a browser pixel.
- Can it handle your payout reality? Currencies, tax forms, approval holds, reversals.
- Does it connect to your stack? CRM, billing, and your analytics. The growth marketer tech stack guide covers where this tool sits among the rest.
There are three broad categories: affiliate networks (the network brings a partner marketplace and handles payments), standalone affiliate platforms (you bring your own partners), and partner-management platforms built for B2B SaaS. A network gives you reach fast and takes a cut; a standalone platform gives you control and leaves recruiting to you. Our best affiliate marketing software roundup compares the field, our guide to choosing affiliate marketing software walks through the evaluation criteria and pricing models, and we've kept pricing out of this article on purpose because it changes too often to belong in a playbook.
One caution on tracking, because it's a trap for growth teams: don't let the affiliate platform become your source of truth for revenue. Reconcile its conversions against billing every month. Platforms report what they were told, and what they were told can be wrong.
Day-30 checkpoint
By the end of the first month you should have, in writing: the allowable commission, the model, the validation window, the signed-off terms, the tool decision, and the one conversion event you'll pay on. If any of those is still "TBD," don't start recruiting. Recruiting before terms exist is how you end up with six partners on six different verbal deals.
Days 31 to 60: Recruit and Enable the First Partners
Start with ten, not a thousand
The temptation is to open the gates and let the network fill itself. Don't. Early volume from unvetted partners mostly gives you noise, and noise is expensive to clean up. Aim for a first cohort of ten to fifteen partners you've personally looked at.
Where do they come from? Work in this order:
- Your own customers and champions. People who already use and like the product and have an audience: consultants, newsletter writers, community moderators.
- Sites that already rank for your buying-intent queries. Comparison and review publishers who already write about your category. Check their content before approaching. Google's spam policies describe pages that copy merchant descriptions without adding value as thin affiliation, and that's the traffic you don't want attached to your brand.
- Adjacent tool vendors and agencies with a complementary product. These sit closer to a referral or technology-partner motion, so decide upfront whether they belong in the affiliate program or a separate track.
- Network marketplace partners, if you chose a network, after you've seen the first three groups perform.
Make the offer easy to say yes to
A good partner has many programs to choose from, so your pitch should answer their three questions fast: what do I earn, how reliably do you pay, and what will my audience get? Give them a one-page summary with the commission, validation window, payout schedule, and a plain-language description of who the product suits and who it doesn't.
Then enable them properly. This is where most programs under-invest, and it's the same lesson partner teams learn everywhere: a partner you don't onboard becomes a partner who never sends anything. The principles in partner onboarding apply directly. The practical kit is small:
- A product brief with accurate positioning, approved claims, and the claims partners must not make.
- Tracked links, and a deep-link option so a partner can send readers to the right page rather than only your homepage.
- A few tested assets: logos, screenshots, a demo account or free-trial extension for reviewers.
- A named contact who answers within a working day.
Train on disclosure before the first post goes live
This is the part that lands on you, not on them. The FTC's position is that an advertiser is ultimately responsible for what others do on its behalf, and that outsourcing doesn't relieve you of responsibility. It also expects "reasonable programs in place to train and monitor members of their network," including periodically searching for what your affiliates say about you.
Build that into onboarding, not an afterthought:
- Tell partners to disclose the relationship clearly, close to the recommendation. The FTC says the closer a disclosure is to the recommendation, the better, and gives "I get commissions for purchases made through links in this post" as a sample of plain wording.
- Rule out vague labels. The FTC says "affiliate link" alone may not be adequate and "commissionable link" is "probably not a clear disclosure."
- On social posts, put the disclosure at the start, not buried at the end. The FTC says disclosures in the middle or at the end of a post are easier to miss.
- Ask partners publishing on the web to mark affiliate links with
rel="sponsored". Google's guidance says paid placements should carry thesponsoredvalue.
Put the disclosure requirement in your terms, send a one-page summary with the welcome email, and keep a record that you did. If a regulator or journalist ever asks, "we trained them and we checked" is a much better answer than "we didn't know." This isn't legal advice, so have counsel confirm the wording for the markets you sell in. The FTC's endorsement guides Q&A is the primary source and worth reading in full once.
Set up the dashboard before the first click
By day 60 you want a working view of the funnel by partner: clicks, sign-ups, qualified conversions, reversals, and revenue. Check it with a test purchase from a tracked link before the first partner goes live. Then do it again from a different browser and device, because cross-device gaps are where tracking breaks first.
Days 61 to 90: Measure, Control Quality and Iterate
Measure incrementality, not just attribution
Here's the uncomfortable question every affiliate program owner should ask: would these customers have bought anyway? Coupon and loyalty publishers often intercept a customer who was already at your checkout, collect a commission, and look like top performers in a last-click report.
You don't need a research team to test for this. Three practical checks:
- Compare partner types. If coupon-site conversions have a much shorter time-from-click-to-purchase and a lower average order value than review-site conversions, you're likely paying for intercepted demand.
- Check the path. Look at what the customer did before the affiliate click. Direct and brand-search visits immediately before a coupon click are a warning sign.
- Run a holdout. Pause one partner or partner segment for a few weeks, against a comparable control, and watch whether total conversions actually fall. It follows the same logic as any test, and growth experiment design covers the sample-size reasoning so you don't read noise as a result.
For the attribution mechanics behind this, partner attribution explains why last-click credit and true contribution diverge. Then decide, with eyes open, whether you reward intercepting behavior or restrict it in the terms.
Build a fraud and quality control routine
Fraud in an affiliate program isn't exotic. It's mostly a handful of repeat patterns, and you should write a check for each one.
| Pattern | What it looks like | Control |
|---|---|---|
| Cookie stuffing or forced clicks | Conversions with no plausible visit or content engagement | Review click-to-conversion patterns; require partner site review |
| Brand-term bidding | Partner ads appearing on your own brand queries | Prohibit in terms; search for it monthly |
| Self-referral or fake sign-ups | Same payment method, device or IP across "customers" | Match billing details; validation window before payout |
| Incentivized or misleading traffic | Sign-ups that never activate | Pay on activated or paid accounts, not sign-ups |
| Content that breaks the rules | Missing disclosure, false claims, copied merchant text | Periodic content audit; terminate and claw back |
Run the review on a fixed schedule, weekly for the first quarter, then monthly. Each review should answer the same four questions per partner: is the traffic real, is the content compliant, do the customers activate, and do they stay? A partner who fails any of the four gets a warning, a rate change, or removal, and the decision gets written down.
This also feeds the FTC expectation discussed earlier. A calendar entry for a monthly search of "what are our affiliates saying" is evidence of a reasonable monitoring program.
Rank partners and iterate
By day 75 you have enough data to separate your partners into three groups. A short scorecard works well here. Keep it short:
- Earners: high activation, good retention, compliant content. Give them better terms, early access, co-created assets, and a direct line to you.
- Maybes: traffic but weak conversion or activation. Send feedback, better creative or a more specific landing page, and give them one cycle to improve.
- Drags: low quality, rule-breaking, or zero activity. Remove them. A quiet program with ten good partners beats one with a hundred noisy ones.
Then pick one or two changes and test them, rather than redesigning the program. Typical levers include the validation window, the landing page partners send traffic to, the commission step for top performers, and the offer to the end customer. Change one thing at a time so you know what moved the result.
What to Report to Leadership at Day 90
Your CFO and CEO don't want a clicks report. They want to know whether to keep funding the program and what it will do next. A one-page summary built around these lines does the job:
| Metric | Why leadership cares |
|---|---|
| Program-sourced revenue and customers, reconciled to billing | Proves the numbers aren't just platform-reported |
| Fully loaded cost per customer (commissions, fees, tooling, your time) | Compares fairly to other channels |
| Payback period and early retention versus other channels | Shows customer quality, not just volume |
| Estimated incremental share, from your holdout or path checks | Answers "would they have bought anyway?" |
| Reversals, rejected conversions and removed partners | Shows the control routine is working |
| Compliance status: terms signed, disclosure training done, audits run | Shows legal exposure is managed |
| Next-quarter plan and what you need | Turns the report into a decision |
Be straightforward about what you don't know yet. Ninety days is enough to prove the plumbing works and to spot early quality signals. It isn't enough to prove the channel's long-term ceiling, especially for B2B with long sales cycles. Say so, and name the date you'll have a better answer. Leadership trusts a marketer who states the limits of the data far more than one who rounds up. If you're newer to the role and want to frame this alongside the rest of your plan, the first 30-60-90 days for a growth marketer is a good companion, and growth metrics helps you pick the activation event to pay on.
The 90-Day Plan on One Page
| Window | Focus | Done when |
|---|---|---|
| Days 1-30 | Allowable commission, model, terms, tracking, tooling | Written terms, one payable conversion event, tool chosen and tested |
| Days 31-60 | Recruit 10-15 vetted partners, enable, train on disclosure | Partners live with tracked links, a dashboard, and disclosure training logged |
| Days 61-90 | Incrementality checks, fraud routine, partner ranking, first iteration | Reconciled report, partners ranked, one change tested, leadership update delivered |
Common Ways First Programs Go Wrong
- Paying on sign-ups. Fast volume, slow regret. Pay on activated or paid accounts.
- Skipping the validation window. Refunds and fake conversions come out of your budget.
- Opening to everyone on day one. You inherit the worst traffic first.
- Treating the platform report as truth. Reconcile to billing every month.
- Assuming disclosure is the affiliate's problem. Per the FTC, it's yours as well.
- No owner. A program run as a side project decays. Someone needs a named hour a week, at minimum.
What to Do Next
If you're starting tomorrow, do three things this week. Calculate your allowable commission with finance. Pick the single conversion event you'll pay on. And draft the one-page terms, including the disclosure clause. Everything else in the plan follows from those three decisions, and none of them need a tool purchase.

On this page
- Key Facts
- First, Decide If This Is Affiliate, Referral or Reseller
- Days 1 to 30: Economics, Terms and the Tooling Decision
- Work out what you can afford to pay
- Choose a commission model on purpose
- Write the terms before you recruit
- Decide how you'll track and what you'll buy
- Day-30 checkpoint
- Days 31 to 60: Recruit and Enable the First Partners
- Start with ten, not a thousand
- Make the offer easy to say yes to
- Train on disclosure before the first post goes live
- Set up the dashboard before the first click
- Days 61 to 90: Measure, Control Quality and Iterate
- Measure incrementality, not just attribution
- Build a fraud and quality control routine
- Rank partners and iterate
- What to Report to Leadership at Day 90
- The 90-Day Plan on One Page
- Common Ways First Programs Go Wrong
- What to Do Next