What is referral marketing?
Referral marketing gives customers a clear way to recommend a product to people who may have the same problem. The introduction can be informal or supported by a link, an account credit, or another stated benefit. A program makes the eligibility, attribution, and reward rules explicit so customers know what to expect.
Shopify's comparison distinguishes customer referrals from affiliate relationships that often involve publishers or partners promoting to an audience. Real programs can overlap, so define the relationship and rules in your own offer. This guide focuses on existing SaaS customers introducing suitable peers.
If you want a publisher commission program, use our micro-SaaS affiliate guide. Here, the practical job is to help a satisfied customer make a useful introduction and then keep the reward promise accurate.
Start with a referral moment that makes sense
Choose a moment when a customer has received a useful result and can reasonably recognize someone with a similar need. For a fictional reporting SaaS, that might be after a successful reporting cycle and a positive support conversation. It does not need to interrupt every session with a request to invite contacts.
Ask whether the customer can explain the product accurately. A share page should state who it helps, the relevant limitations, and the benefit attached to the referral. Give people wording they can adapt rather than requiring exaggerated claims or promises about another person's results.
Let the referring customer choose whom to contact. Do not upload an address book or send invitations to their contacts without the appropriate permission and a clear process. The downloadable worksheet plans the program; it does not send messages, create rewards, or collect personal information.
Work through an original SaaS referral offer
The fictional ReportDesk offer gives the referring account a twenty-dollar account credit and the referred account a ten-dollar credit after a qualifying first paid invoice has remained valid for thirty days. Credits are not cash, and the example caps referrer credits at sixty dollars per billing month. These values illustrate a design choice; they are not recommended reward levels.
Eligibility requires an active existing customer making an introduction to a genuinely new, separate customer account. The referred account must have an eligible referral recorded before its first paid invoice. The program owner reviews the qualifying event before issuing either credit.
Write the conditions on the offer page before the customer shares it: who qualifies, how attribution works, the review period, credit limits, and where questions go. If a person receives an incentive for recommending the product, make that relationship clear in the suggested message and offer.
| Decision | Example rule |
|---|---|
| Qualifying customer | A new separate account with an eligible first paid invoice |
| Attribution | First valid referral recorded before the invoice; manual review for disputes |
| Reward timing | After a thirty-day review period; not on signup alone |
| Referrer benefit | Twenty-dollar account credit, capped at sixty dollars per billing month |
| Referred benefit | Ten-dollar account credit after qualification |
| Excluded behavior | Self-referrals, duplicate account creation, and misleading claims |
Check reward economics before promising the benefit
Count both sides of the offer. In the example, the total promised credit is thirty dollars per qualified referral before review and support effort. A ledger that records only the referrer's twenty dollars understates the offer. Credits affect billing differently from cash payments, so use the treatment appropriate to your billing records.
Estimate what the customer actually contributes over a relevant period after service costs, then compare that with the reward and operating work. Keep uncertain retention and future usage assumptions visible. Do not turn a speculative lifetime value into permission to offer an unlimited incentive.
The customer acquisition cost guide explains consistent cost boundaries. For a referral pilot, separately record earned rewards, issued credits, support effort, and qualified new customers. This lets you inspect the program without claiming that a referred customer would never have arrived through another route.
Make attribution and exception handling inspectable
Use a stable referral identifier and a server-recorded attribution event. A share-link click, account creation, paid invoice, review completion, and issued reward are different states. Record timestamps and the applicable rule version. A browser cookie alone may not explain a cross-device signup or a later dispute.
Choose a deterministic duplicate rule and document when a human can override it. Our fictional program uses the first valid referral recorded before the qualifying invoice. A second introduction is marked duplicate rather than creating another reward. An account referring itself is ineligible, and an ambiguous relationship is held for review.
Handle refunds, reversals, and failed invoices according to the published terms. Before the review period ends, the example leaves the reward pending. Do not silently invent a new clawback rule after a credit has been issued; define the process and explain corrections to the affected customer.
Use the review ledger to follow a referral to completion
The original CSV contains five fictional referrals. Two qualify, one remains pending, one is a duplicate, and one is an excluded self-referral. The two qualified rows earn forty dollars in referrer credits and twenty dollars in referred-customer credits, for sixty dollars total. No other row earns a credit in the worked example.
One qualified pair has been issued and the other remains earned but unissued. Earned credit totals and issued credit totals therefore differ: sixty dollars earned, thirty dollars issued. Use separate columns so a queue awaiting action is visible. The synthetic account IDs do not identify real customers.
Blank rows let you adapt the ledger. Review access and retention before entering real records, and keep only what is needed to establish qualification and fulfill the offer. The file is a planning resource, not a payment processor or an automated fraud decision.
Run a bounded pilot and review customer quality
Start with a small eligible audience and an owner who can review exceptions. Track invitations or shares only if you can define them reliably; do not confuse opening a share page with a recommendation that actually reached someone. Follow qualified accounts into the product outcome you care about.
Compare activation and later retention using consistent observation windows and appropriate context. Referred accounts may differ from other accounts before they join, so an observed difference is not automatically caused by the referral program. Watch support burden, misleading claims, and concentration in a few referrers.
Use customer success to establish whether the new customer reaches the intended result. Keep the offer stable during the pilot or version changes clearly. The worksheet includes the offer, qualification, review states, economics assumptions, exception rules, and a stop decision. Source and resources checked October 27, 2026.