They work when customers already recommend you and you are removing friction. They fail when an incentive is doing work the product should be doing.
A referral programme tries to manufacture word of mouth. It works when customers already recommend you and you are removing friction. It fails when you are using an incentive to create a recommendation that would not otherwise happen.
If nobody refers you now, an incentive will not change that. It is a multiplier on something that has to already exist.
Before building anything, answer one question: has any customer referred someone in the last three months without being asked?
If yes, a programme removes friction from something already happening and is likely to pay back. If no, the honest reading is that your product does not currently produce recommendations, and an incentive is a weak substitute for that.
The reason recommendations do not happen is usually one of three things: the product is not visibly good enough yet, the customer has no natural occasion to mention it, or recommending it carries a social cost. Only the first is fixed by time, and reading fit at small scale covers assessing it.
Each step is cheap, and each tells you whether the next is worth building.
Step 1 — just ask. Email ten customers who have used the product for three months or more: is there anyone you would introduce me to? No incentive, no link. This tests whether the goodwill exists.
Step 2 — add a link. If the ask worked, give people something to send. A simple shareable link with no tracking beyond a source tag. Most referral value at solo scale comes from this step, not from incentives.
Step 3 — add an incentive. Only if steps one and two produced referrals and you want more. Anything earlier is paying for something you have not proved exists.
| Structure | Effect | Best for |
|---|---|---|
| Referrer only | Feels transactional; can embarrass the referrer | Rarely the right choice |
| Referred only | Gives the referrer something to offer | Products with a social cost to recommending |
| Both sides | Highest conversion, clearest to explain | Most solo SaaS |
A month free on each side is the simplest structure that works and costs you margin only on customers you would not otherwise have. Cash rewards introduce accounting and attract the wrong behaviour.
Avoid permanent discounts as a referral reward. A recurring reduction for every referral compounds into a customer paying almost nothing while generating full support load — the same reasoning that makes discounting a poor cancellation save.
Make the reward claimable without you doing anything manually. A referral programme that requires you to apply credits by hand becomes a support task that scales with its own success, which is the wrong shape for a one-person business.
Three conditions together. You have enough customers that a small percentage referring produces meaningful volume — roughly $5k MRR or fifty-plus customers. Referrals are already happening unprompted. And you can automate the reward end to end.
Below that, the same hours spent on activation or churn produce more. A referral programme at twenty customers can only ever produce a handful of signups, and it costs the same to build as one at two hundred.
The $1k to $10k playbook covers what to prioritise at the stage below this, and retention versus acquisition covers where the effort pays best.
Before a programme is worth building, one practice is available immediately and outperforms it at small scale: ask every happy customer, personally, once.
When someone says something positive — in an email, a support thread, a cancellation survey they did not complete — that is the moment. One line: is there anyone else who has this problem?
It does not scale past about a hundred customers, which is exactly why it is available to you now. The customer interview guide covers keeping those conversations useful rather than transactional.
Not below roughly $5k MRR or fifty customers. A programme is a multiplier on recommendations that already happen — if nobody currently refers you unprompted, an incentive will not create that, and the same hours spent on activation or churn produce more.
Ask whether any customer has referred someone in the last three months without being prompted. If yes, a programme removes friction from something already happening. If no, the product does not currently produce recommendations.
A reward on both sides, typically a month free each. It converts better than one-sided rewards and is easiest to explain. Referrer-only rewards feel transactional and can embarrass the person recommending you.
No. A recurring reduction for every referral compounds into a customer paying almost nothing while generating full support load. Use a one-off reward such as a free month instead.
Asking every happy customer directly, once, at the moment they say something positive. It does not scale past about a hundred customers, which is precisely why it is available to you at small scale and why it outperforms an automated programme there.
Tell Marcus your customer count and whether anyone has referred you unprompted. You get a straight answer on whether to build it.
Try GhostCoach free →14-day free trial · cancel anytime · 30-day money-back on Lifetime