A referral is a data handoff between an advocate and a potential buyer. The system works when value proof, timing, incentive logic, and ownership are explicit.
Outcome, milestone, or avoided risk that the customer can describe.
Recent use, first win, renewal, expansion, or resolved risk.
A named recipient, clear benefit, consent, and an accountable owner.
Select a node to inspect the requirement, evidence, and failure mode at that stage. The centerline is deliberate: no gate should be skipped because the channel appears easy.
Trigger the referral ask only after the customer can point to a real result, milestone, or avoided problem. Satisfaction language is weaker than evidence of value.
outcome → fit → permissionRecord the business or personal change the customer experienced. A testimonial adjective is not a measurement; identify the task completed, risk reduced, time saved, or result made possible.
Recent and intensive use gives the customer a more accurate picture of fit. Weak activation, long inactivity, unresolved support, or payment trouble should suppress the ask.
The advocate should be able to describe a similar person or team with a similar constraint. “Anyone who needs this” is not a routing rule.
Make the ask easy to decline. Capture whether the customer wants to share, which channel they prefer, and whether they consent to the business following up with the recipient.
Failure mode: a company asks every account because it has a customer list. The result is a large activity number with no relationship between customer outcome and referral quality. The fix is not a louder prompt. The fix is an eligibility layer built from product usage, outcome evidence, relationship health, and fit.
A useful trigger explains why the request is relevant now. First value, renewal, expansion, successful recovery, or recent intensive use can each create a different window; none should become an automatic calendar ritual.
Write the customer event that makes the result vivid and the referral relevant.
Allow enough time for the customer to use the outcome, but do not wait until the memory is stale.
Set a suppression period and stop after a decline. Frequency is an experience constraint.
Compare qualified conversion and customer quality by event, segment, and time since use.
Measurement note: one longitudinal platform study found referral activity was associated with recent and intensive usage, but the percentages came from a specific category and should not be treated as a universal benchmark. Use the direction of the finding—fresh value matters—as a test hypothesis in your own business.
The reward architecture should reduce the recipient’s risk, recognize the advocate’s effort, and leave enough contribution to fund delivery. Design from the end state backward.
| Architecture | Why it can work | What to monitor |
|---|---|---|
| Recipient-first | The advocate can frame the introduction as useful help; uptake friction is reduced. | Qualified conversion, reward redemption, low-fit volume, and post-purchase quality. |
| Sender-first | The advocate receives a visible benefit for effort and may become more active. | Share quality, reputational friction, abuse, and whether the reward overwhelms contribution. |
| Two-sided | Both people can understand the exchange and the recipient receives a reason to act. | Eligibility, duplicate claims, reward liability, and incentive stacking. |
| Non-cash | Access, service credit, training, or a useful upgrade can preserve offer congruence. | Internal delivery cost, capacity, perceived value, and effect on retention. |
Design rule: hold the message and eligibility constant while testing incentive architecture. If the reward, copy, timing, and qualification all change, the result cannot teach you which mechanism moved behavior.
Store advocate identity, recipient identity, consent, context, source, timestamp, campaign variant, and existing-lead status.
Assign a named owner and response SLA. Give the owner enough context to continue the conversation without interrogating the recipient.
Separate a completed introduction from a qualified opportunity. Track fit, urgency, authority, capacity, and the actual problem the recipient wants solved.
Record conversion, reward trigger, contribution after reward, retention, and the advocate’s future health. Feed the result back into eligibility and timing.
Weekly control panel: eligible customers, asks delivered, initiation rate, completed introductions, response time, qualification rate, close rate, reward cost, contribution, retention, and referrer retention. A channel is predictable when the variance is explainable.
Ask after a credible value event, while the result is recent enough to describe. Test first value, renewal, expansion, recovery, and recent intensive use as separate triggers. Avoid a universal calendar rule.
Name the customer’s result, describe the type of person who may benefit, make the recipient’s benefit clear, and give the advocate an easy decline. Do not make the customer perform a vague favor for the company.
Pay at the economically meaningful event you can verify. A qualified payment, completed onboarding milestone, or retained period may be more protective than a click or an unqualified form submission.
Compare referred and nonreferred cohorts using the same definitions and acquisition window. Review conversion, contribution after reward, retention, expansion, support load, and time to value.
Define the value proof, timing window, incentive test, and process owner before sending the next referral request.