Is referral right for this business
Referral works where three conditions hold. Customers are already pleased, so a recommendation costs them nothing in credibility. The product is easy to explain to a friend in a sentence. And the customer's friends look like customers: people who live nearby for a local service, who share the hobby, who work in the same kind of business. Where those hold, a programme turns occasional recommendations into regular ones and makes them traceable. Where they do not, a reward cannot rescue the product, and a programme can do harm by paying people to recommend something they would not recommend on its own merits.
The flywheel framework treats a referral programme as a force that speeds the wheel and a cumbersome referral flow as the friction that cancels it. The growth loops framework adds a warning: a referral programme is often a small loop presented as a large one, with incentive cost and fraud as its weak points. The guide to growth loops in practice shows how to map the loop and find its leaks, and for service businesses the leak is usually the ask. Where the people recommending a subscription product are publishers paid by commission rather than customers, the guide to affiliate programmes for subscription businesses covers that arrangement instead, and the comparison of referral and affiliate programmes helps decide between rewarding customers and paying publishers when both routes are open. Where the recommender is a neighbouring business rather than a customer, the guide to partnering with nearby businesses covers choosing the partner, the written terms and how each side measures what it gets.
| Signal | Points towards referral | Points away from it |
|---|---|---|
| What customers say | They recommend the business unprompted, and say so in feedback | Satisfaction is middling or falling |
| Who new customers are | Many already arrive through a recommendation | Almost all arrive through paid channels |
| The product | Easy to explain and to try | Hard to explain, or embarrassing to mention |
| The customer's network | Friends and colleagues are likely buyers | The customer is unusual among the people they know |
| The economics | A margin that can fund a reward for both sides | A margin too thin to pay for anything |
Before you start
- Know the economics. The contribution margin on a first order, and the repeat value that can be relied on, set what a new customer is worth; the gap between that and the current cost of winning a customer (CACCost to acquire one new customer.) shows how much room a reward has.
- Know how customers arrive today. A how-did-you-hear question at checkout or on enquiry forms gives a baseline for recommendations before the programme exists, which is what measurement will be compared against.
- Have a consented way to talk to customers. The programme is announced to existing customers, so the email list and its consent records need to be in order; the list building and consent guide covers both.
- Choose the mechanism. Check whether the e-commerce, booking or subscription platform already has a referral feature or add-on, and make sure whatever you choose can issue unique links and codes, hold a reward until its conditions are met and export the data.
The steps
Confirm customers already recommend you
Look for evidence before building anything: what customers say in surveys and reviews, the share of new customers who say a friend recommended the business, and whether recommendation comes up unprompted in interviews. The voice of the customer guide explains the surveys that ask whether customers would recommend the business, and how to read the open-text answers that say why. If the evidence is weak, fix the experience first; a reward on top of a mediocre product buys a few referrals and some disappointed friends.
Decide who is rewarded and with what
Choose between a reward for the referrer, a welcome for the friend, or both. A welcome for the friend makes the recommendation easier to pass on, because the customer is offering something rather than asking a favour. A reward for the referrer gives a reason to remember. Then choose the form: account credit, which costs the business less than its face value and brings the customer back to spend it; a free product or service; or a donation to a cause, which suits customers who would feel awkward being paid to recommend. Cash is the easiest to understand and the most attractive to people gaming the scheme.
Set the reward from the margin
Work out the allowable cost per new customer: what a customer is worth in margin on the first order plus the repeat value the business can rely on (LTVTotal value a customer brings over time.), less the profit the business wants to keep. Add up everything a referral costs: both rewards, any discount on the friend's first order and the platform's fees. That total must sit under the allowable cost, with room for the rewards paid on customers who would have come anyway. The CPA calculator does the arithmetic once the inputs are known. Compare the result with what paid channels cost per customer, but set the ceiling from the margin, not from what those channels happen to cost.
Choose the moment to ask
Ask just after a success point in the journey, when the customer has had the result they came for: a delivery that arrived well, a problem solved, a milestone reached, a repeat order placed. The lifecycle guide places the advocacy request at the moment satisfaction is likely to be highest. Ask in the channel where that moment happens, such as the confirmation page, the follow-up email or the app, and repeat the invitation at later success points rather than once at sign-up, when the customer has nothing to recommend yet. Where the confirmation page is the place, the guide to confirmation and thank-you pages explains how a referral request fits beside the confirmation itself as the one next step offered.
Build a simple sharing flow
Give each customer a personal link or code and let them share it from their own email, messaging and social accounts, with a short message they can edit. As general practice, the business does not collect friends' contact details in order to message them itself: that would mean the business contacting people who never agreed to hear from it, which is the question the UK Information Commissioner's Office's direct marketing guidance addresses when it explains when consent is needed for electronic marketing. Once the friend arrives, any marketing the business sends them follows the consent they give, as the list building guide describes. Rules differ by market, and nothing here is legal advice. Keep the flow to as few steps as possible and test it on a phone.
Write terms and fraud checks
Publish short terms: who can refer, what counts as a successful referral, when and how rewards are paid, any limit on rewards, when credit expires, and that the business can end or change the programme. Then build the checks. The affiliate partner and fraud guide lists the patterns that apply to referrals too: self-referral, where people refer themselves through a second account, and codes published on voucher and deal sites they were never meant for. Match new accounts against existing addresses, payment details and devices; pay the referrer's reward only after the friend's order is complete and past any return or cancellation period; and treat a sudden spike from one code as a question, not good news.
Handle disclosure for public posts
Many customers will share privately, but some will post their link publicly on social media, in forums or in reviews. When someone posts about a business and receives a reward for it, their audience should be able to tell. The FTC's endorsement guidance explains that a material connectionA relationship between an endorser and a brand, such as a fee, free product, commission or employment, that endorsement guidance expects to be disclosed. between an endorser and a brand should be disclosed clearly and conspicuously, and the influencer contracts and disclosure guide sets out how that and the UK regulators' approach apply in practice. Give customers suggested wording that says plainly that they receive a reward, and never ask them to hide it. Keep the referral reward entirely separate from reviews, and never make a reward depend on a review: the reviews and social proof guide sets out how regulators treat incentives for reviews, and the local reviews guide notes that review platforms can prohibit them altogether.
Tag and track every referral
Every personal link carries the referrer's identifier, and adding UTMURL tags identifying traffic source. parameters for the channel the customer shared through shows whether referrals travel by email, messaging or social posts. Record the referral code at checkout or sign-up, so that a friend who types the address instead of clicking is still counted, and keep asking how new customers heard about the business, because some recommendations will never pass through a link. Store the referrer, the friend, the date, the reward status and any fraud flag together, so that every referral can be traced from share to reward.
Measure customers beyond those who would have come anyway
Some referred customers would have arrived without the programme, because their friend would have recommended the business regardless. Count the programme's effect, not its activity. Compare the share of new customers arriving by recommendation before and after launch; where the platform allows, show the programme to only part of the customer base for a period and compare new-customer numbers from each group. Then compare referred customers with other new customers on Repeat Purchase RateShare of customers who placed more than one order in the period., average order value and returns, because a referral is worth more only if the customer it brings is worth at least as much. The cost per extra customer, not the cost per referral, is the figure to set against the paid channels. Referred customers can recommend the business in turn, and how a growth loop compounds shows what those later rounds add and why they arrive in later cycles rather than in the month a programme launches.
Review and adjust
Review the programme on a fixed cycle. If customers share but friends do not convert, the welcome or the landing page is the problem. If few customers share, the ask comes at the wrong moment or the reward means little to them. If referred customers churn quickly, the reward is attracting the wrong friends, often because it is too generous on the first order. Change one element at a time, so the effect of each change can be read, and retire the programme if it does not pay for itself after a fair trial. When the open question is who the reward should go to, the referral reward test sets out a design that splits customers between reward structures and reads the margin each brings per invited customer.
A laundry service launches referrals (illustrative)
Sudlark Laundry rewards both sides
Sudlark Laundry, an invented laundry collection service, earns a contribution of about £9 per collection after cleaning and driver costs. Customers who stay average 11 collections in their first year, but many leave sooner, so it treats £60 of first-year margin as the value it can rely on and sets an allowable cost per new customer of £35. Paid social costs it about £42 per new customer.
The design gives the referrer and the friend a £10 credit each, paid once the friend's second collection is complete. Platform fees add about £2 per referral, so a successful referral costs £22, leaving room for credit paid on friends who would have joined anyway. The invitation appears after a customer's third collection, by which point complaints have usually been resolved and the routine has settled. Each customer gets a personal link and a short message to edit and send from their own phone.
| Check | Rule |
|---|---|
| Self-referral | Block any friend whose collection address or card matches an existing account |
| Code sharing | Codes valid only through the personal link page; codes found on deal sites deactivated |
| Reward timing | Credit released after the friend's second completed collection |
| Public posts | Suggested wording: "I get £10 credit if you sign up through my link" |
Over six months, 14% of eligible customers share their link and 290 friends complete a second collection. Before launch, 11% of new customers already named a friend as the reason they joined; afterwards the figure is 24%, so the business attributes a little over half of the referred customers to the programme. The 290 referrals cost £6,380, which works out at about £40 per extra customer, above its £35 allowable cost. It then compares groups at the 90-day mark: referred customers show a repeat rate of 68% against 52% for other new customers, so it raises the value it relies on for a referred customer and, with it, the allowable cost for referrals to £45, which the programme now sits under. The address check blocks 31 attempted self-referrals in the first month.
Common mistakes
- Launching a programme to fix a weak product. Rewards cannot make people recommend something they would not recommend anyway.
- Setting the reward by copying competitors. The reward has to fit under the business's own allowable cost per new customer.
- Asking at sign-up. A customer with no experience of the product has nothing to recommend yet.
- Collecting friends' contact details. Let customers share from their own accounts rather than messaging people who never agreed to hear from the business.
- Paying the reward on the first click or order. Self-referrals and quick cancellations collect rewards that bring in nobody.
- Leaving disclosure to chance. Customers who post a reward link publicly need simple wording that says so.
- Counting every referred customer as new business. Some would have come through ordinary word of mouth; measure the extra, not the total.
- Never changing the programme. Review it on a cycle and change one element at a time.
