Referral Programs:
Design for Quality Growth

Referral programs encourage existing customers to introduce new customers through tracked links or codes, with rewards for qualifying actions; Salesforce’s guide identifies sharing, tracking, and incentives as core program elements.

The central design choices are who can participate, what counts as a successful referral, who receives a reward, and when that reward becomes available. This guide covers customer referral programs: their mechanics, reward structures, real examples, implementation, and measurement.

In Shopify’s comparison, referrals typically come from existing customers sharing with their personal contacts; affiliates are promotional partners, such as publishers or creators, who earn commissions. Read the participation and payment terms to identify the arrangement.

How referral programs work#

The process outlined in Salesforce’s guide has four stages:

1. Identify. An eligible customer receives a unique referral link or code.
2. Invite. The customer shares the offer with another person.
3. Qualify. The recipient completes the required action through the tracked route.
4. Reward. The business verifies eligibility and issues the promised benefit.

The qualifying action is part of the offer. Account creation, product activation, and a completed purchase are different milestones; the terms must specify which one earns the reward.

Common referral program types#

These reward designs can be combined:

Specify the reward schedule and spending cap for each chosen structure.

Dropbox: storage rewards for both participants#

Under Dropbox’s published referral rules, a Basic account earns 500 MB for a successful referral, up to 16 GB, and the invited friend receives 500 MB. The completion requirements include account creation through the referral link, app installation, desktop-app login, and email verification.

This is a two-sided program that provides a product benefit. Its completion requirements also show why a referral offer needs a qualifying event separate from the invitation.

Rakuten: bonuses tied to qualifying purchases#

Rakuten’s referral terms provide for referrer and new-member bonuses whose amounts and spending requirements depend on the applicable offer. A qualified referral must be a new member who joins through a trackable referral method and makes purchases that earn Cash Back within 90 days, subject to those terms.

Here, registration alone does not complete the referral. The program connects the reward to purchasing activity and publishes eligibility and distribution restrictions. These are examples of different mechanics, rather than performance benchmarks for another business.

What benefits can referral programs deliver?#

Evidence of customer value exists, but the results depend on the program. A 2011 study by Schmitt, Skiera, and Van den Bulte followed approximately 10,000 German bank customers for almost three years and found higher retention and at least 16% higher average value for referred customers than comparable nonreferred customers, with differences across customer segments.

That finding is an observed result at one bank, not a universal forecast. A June 2024 working paper by Rubli and Tudon, based on randomized interventions at a financial-technology platform in Mexico, found that incentives approximately doubled the likelihood of making a referral. The authors also reported lower first-year app engagement among customers referred by incentivized users than among those referred by control-group users.

These studies support measuring acquisition and subsequent customer behavior separately. A program evaluation should report how many customers qualified, what they cost to acquire, and whether they continued to use or buy the product. Higher referral activity alone does not answer the customer-value question.

How to design a customer referral program#

Define eligibility and the qualifying action#

Write a precise definition of a new customer before choosing the reward. Specify whether previous customers, employees, household members, or existing account holders qualify. State any geographic or age restrictions and the deadline for completing the required action.

Choose a qualifying event that matches the program’s objective. For paid customer acquisition, use a verified purchase or payment milestone. For adoption, define the activation requirement explicitly. Set reward approval after the relevant cancellation or refund checks when those checks affect eligibility.

Publish five details together: who can refer, who can be referred, what must happen, what each participant receives, and when the benefit becomes available. Keep the short invitation consistent with the full terms.

Choose the reward and its recipient#

Decide the recipient, form, and amount separately. Evaluate cash, store credit, discounts, additional usage, and product benefits against the budget and redemption rules. Specify expiration, minimum spending, redemption limits, and any restrictions on combining offers.

The Rubli and Tudon experiment found stronger responses to rewards for the referrer than to rewards for the invited person. This supports testing the allocation rather than assuming an equal split is optimal everywhere.

Reward fit is another testable choice. Hu and Zhang’s 2021 research found an aggregate benefit from matching practical rewards to practical products and experiential rewards to experiential products. The authors caution that their Chinese participant samples limit generalization across cultures.

Compare reward variants within the same planned cost ceiling, and evaluate qualified customers and retention alongside participation. Record the expected cost of delivering or redeeming each benefit; its advertised value is a separate budgeting input.

Set the budget from customer contribution#

Use a defined evaluation period and an explicit acquisition-cost limit. Build the reward budget after accounting for software, payment processing, administration, customer support, and losses from invalid rewards.

For planning, define:

Maximum reward budget per customer = allowed acquisition cost per customer − other program costs allocated per customer.

Set the allowed acquisition cost from expected customer contribution and the required payback period. Contribution means revenue after the costs included in the business’s chosen contribution calculation; document those inclusions so the budget can be reviewed consistently.

Include benefits for both participants. Count a discount either as reduced revenue in contribution or as a program cost, and avoid deducting it twice. For credits and product benefits, document assumptions about redemption and fulfillment. Use a total spending cap as well as a per-referral limit.

These are planning definitions, not recommended reward amounts. The appropriate budget requires the business’s own customer-value and cost data.

Make the offer easy to find and share#

Shopify’s referral marketing guide describes trackable links and promotion through website navigation, customer emails, and post-purchase touchpoints. Use a dedicated program page that explains the benefit, qualification requirements, timing, and material limits.

Place a copyable link or code beside the explanation. Keep the recipient’s landing page consistent with the invitation, including any deadline and purchase requirement. Make the program available through customer account pages and relevant purchase communications without requiring customers to search for the terms.

Schedule invitations after a completed customer experience and test their timing. Measure qualification and retained customer value for each placement, rather than judging placement solely by links copied.

Choose tracking and reward-management tools#

Evaluate software against the published rules. Require a reliable association between the referrer and recipient, records of qualifying actions, duplicate checks, and a history of reward decisions. Include refund handling, spending limits, customer-facing status, and data export in the requirements.

Show the remaining requirement when a reward is pending. Dropbox’s status documentation distinguishes invitations, installation, email verification, completion, and ineligibility. Use equally specific statuses for the chosen qualification process.

Test the actual purchase or activation flow before promotion, including mobile access and code entry. A small program can begin with controlled manual review; recurring volume may justify automated approval and reconciliation. Define who resolves missing-credit claims in either arrangement.

Referral terms, disclosure, and abuse controls#

Attribution rules should state which referral receives credit, how long eligibility lasts, how links and manually entered codes interact, and whether one recipient can generate more than one reward. Decide how refunds, cancellations, payment failures, and suspected duplicate accounts affect approval or reversal.

For a documented example, Rakuten prohibits self-referrals, fake accounts, purchases made on another person’s behalf, and specified unsolicited distribution practices. Its terms also permit identifying duplicates and withholding or canceling rewards under the program rules.

Set review rules for these behaviors, record the reason for a decision, and provide a way to report an error. Use spending caps and approval checks before issuing high-cost rewards. Document the data needed for eligibility and attribution, with access and retention rules appropriate to the program.

Disclosure is a separate requirement from reward eligibility. In the United States, FTC guidance calls for clear and conspicuous disclosure of an unexpected material connection that affects the evaluation of an endorsement, including valuable incentives. The FTC also warns that a disclosure hidden behind a link may be inadequate and that a personalized discount code may not explain a financial relationship.

Build the incentive disclosure into promotional sharing text, close to the recommendation, and review it in the channel where it appears. FTC guidance also requires honest endorsements and forbids using them to convey claims the advertiser could not legally make. Apply the relevant local advertising, privacy, and messaging requirements before launch; the cited guidance concerns the United States.

How to measure referral program performance#

Use a funnel that separates eligible customers, customers shown the offer, participating referrers, unique referred prospects, qualified new customers, approved rewards, and retained customers. Choose one reporting period and define each event before comparing results.

The following definitions provide a practical measurement framework:

Keep attributed and incremental results separate. Attribution identifies credited customers; incrementality estimates how many additional customers the offer produced compared with its absence.

For an incrementality test, use randomized offers among eligible referrers, adapting Rubli and Tudon’s experimental approach. Measure qualifying customers per assigned referrer, including those making no referrals, and record possible overlap between groups.

Compare referred and nonreferred cohorts over the same elapsed time after acquisition, using consistent definitions of retention and contribution. Treat this comparison as a quality assessment; differences between customer groups alone do not establish the program’s causal effect.

Change one major design variable per test where possible. Set the observation period, cost ceiling, and review triggers before the test begins. Review complaints, rejected rewards, refunds, and customer activity alongside acquisition totals.

Launch with clear acceptance criteria#

Begin with a limited rollout and a documented spending cap. Confirm that the invitation and terms agree, tracking reaches the qualifying event, rewards reconcile with completed actions, and support can explain each status. Complete checks for duplicate credit, expired offers, and reversals before expanding distribution.

Expand after the program meets its predefined customer-quality and cost targets. Keep reward promises and limits visible throughout the customer experience, and reassess the qualifying event when acquired customers fail to reach the intended value milestone.