Referral Funnel Metrics
Compare industry sharing, invitation conversion, acquisition, and time-to-first-referral metrics, aligning vendor denominators to build a referral forecast.
Written and reviewed by Eric Han · Co-founder & CTO, reopt
Key takeaways
- Vendors use different sharing denominators: Extole uses promotion clicks; ReferralCandy uses prompt exposures. Treating either as participation among all customers greatly overstates results.
- ReferralCandy's top-25% sharing-action threshold is 4.64%; its top-10% threshold is 13.38%.
- Median time to first referral sale is 14 days; 68% of measurable programs reach it within a month.
- Referrals concentrate among a few people: the 1.1% of referrers making at least 11 referrals generate 30.2% of referred customers.
- We found no Korean-company invitation participation or invite-to-signup benchmarks. The nearest evidence is Toss's published sharing and event-participation data.
Industry Benchmarks
Extole publishes industry medians. Share rate = shares / promotion clicks; signup rate = signups / shared-link clicks, including engagement such as email entry or lead submission; conversion rate = conversions / shared-link clicks; acquisition rate = converted customers / referrers. Signup and conversion share a denominator, so do not multiply them. Because these are industry medians, some conversion values exceed signup values. The vendor discloses neither sample nor period (grade C).
- Apparel15.8%
- Banking13.4%
- General retail10.8%
- Lending9.4%
- Health / beauty9.1%
- Consumer goods7.2%
- Insurance6.6%
- Financial services6.5%
- Telecom5.2%
- Education4.1%
- Hotels / travel2.2%
- B2B1.3%
| Industry | Share rate | Signup rate | Conversion rate | Acquisition rate |
|---|---|---|---|---|
| B2B | 4.8% | 7% | 1.3% | 17% |
| Banking | 15.6% | 17.9% | 13.4% | 30% |
| Financial services | 17% | 4.9% | 6.5% | 7% |
| Telecom | 15.9% | 17.5% | 5.2% | 8% |
| Apparel | 5.2% | 15.2% | 15.8% | 15% |
| General retail | 7.7% | 16.2% | 10.8% | 21% |
| Consumer goods | 4.9% | 20.2% | 7.2% | 11% |
| Health / beauty | 7% | 13.4% | 9.1% | 14% |
| Hotels / travel | 7.3% | 10.6% | 2.2% | 11% |
| Insurance | 26% | 8.9% | 6.6% | 22% |
| Lending | 30% | 22% | 9.4% | 16% |
| Education | 14.8% | 6.8% | 4.1% | 13% |
Downstream figures are also available: banks/credit unions report 60% signup-to-account-opening and 67% opening-to-qualification; telecom reports 38% signup-to-order and 80% order-to-installation.
Other Datasets
Mention Me 2021 reports sector averages across over 450 brands. The purchase-rate denominator is not defined in the text (grade C).
| Sector | Share rate | Purchase rate |
|---|---|---|
| Fashion | 6.3% | 14.4% |
| Beauty | 7.5% | 30% |
| Utilities | 8.1% | 18.5% |
| Travel | 9% | 25.7% |
| Finance | 9.9% | 33% |
| Home | 6.8% | 20.6% |
| Gifts | 6.2% | 30.1% |
Post-purchase pages generated 62% of new referred customers in fashion and 68% in home, suggesting that prompt placement strongly affects sharing volume.
ReferralCandy's sharing-action rates cover 500 Shopify stores from July 2025–June 2026, using prompt exposures as the denominator. Other rows below come from separate studies by the same vendor, mostly June 2025–May 2026, with different samples (grade C). A sharing action is a button click; actual delivery is unknown.
| Metric | Value |
|---|---|
| Top-25% sharing-action threshold | 4.64% |
| Top-10% sharing-action threshold | 13.38% |
| Time to first referral sale, measurable programs | Median 14 days; 68% within a month; 89% within 90 days. 23 days below 100 monthly orders |
| One-time referrers | 83% of referrers; 44.9% of referred customers |
| Referrers with 11+ referrals | 1.1% of referrers; 30.2% of referred customers |
| Chat/email share | 69.2% of share-button clicks |
| Stores with more chat/email clicks than public social clicks | 93.7% |
| Stores reaching a meaningful referral revenue share | Cash/commission rewards about 2.2 times coupon-only rewards, observational |
Align denominators before comparing
These three datasets do not measure the same sharing rate. They also include only existing programs, introducing survivorship bias. Choose one source and construct the full funnel using its denominator definitions.
Simulation Inputs
| Stage | Input | Evidence and conditions |
|---|---|---|
| Prompt exposure | Mean 30%, | Judgment; no public data found |
| Sharing | Mean 3%, | Below ReferralCandy's 4.64% top-quartile threshold; denominator is prompt exposures |
| Invitations per sharer | Uniform 1–4 | Judgment |
| Invitation conversion | Mean 3%, | Judgment between Leskovec's 0.5% and Aral/Walker's 10% |
| Incremental share | Uniform 0.5–1 | Judgment; unknown before on/off experiment |
Sharing and invitation-conversion inputs combine differently defined sources and are judgment assumptions. Do not use Dropbox's 35%, PayPal/Airbnb/Uber anecdotes, or Andrew Chen's 20–30% as inputs.
Start with the total active customer base, then apply each stage rate to estimate referrals. The formulas below show both the first generation and the cumulative effect across generations.
is the average number of new customers brought in by one customer.
When is below 1, the sum across generations is finite. At or above 1, the sum diverges; question the inputs first.
Referral-program intervals
Many inputs lack evidence, so keep ranges wide. Sharing and invitation conversion combine different denominators as judgment assumptions. The output includes incremental share. First inspect which input drives uncertainty.
Each input alone moves to its 5th and 95th percentile with others at the mean. Measure the top one first.
- Invitation conversion0.9customers ~ 28.0customers
- Share rate2.9customers ~ 20.8customers
- Invitations per sharer4.7customers ~ 15.6customers
- Prompt exposure6.8customers ~ 14.0customers
- Incremental share7.1customers ~ 13.2customers
8,000 runs. Stages are independent. If they move together the real interval is wider; if they move in opposite directions it is narrower. The 90% interval holds only rate uncertainty around the expected value; the actual count adds chance variation in counts. Vertical lines mark the 5th percentile, median (darker), and 95th. The last bin collects values above the 99th percentile.
Check Top-Down Estimates Against Observed Outcomes
Multiplying vendor metrics can easily produce an overly optimistic forecast. Compare with large real-world datasets.
| Cross-check anchor | Value | Context | Evidence |
|---|---|---|---|
| Cumulative successful referrals per customer at observation | 0.18 organic; 0.30 referred | Cashback app, 41.2 million people, average tenure about four years | Gershon and Jiang, B |
| Purchases per referral | About 0.51%, calculated | Online retail | Leskovec et al., B |
| Share acquired by referral | About 30% | Same cashback app | Gershon and Jiang, B |
| Share acquired by referral | 5.1% | My Yoga Teacher, 54,349 people | Gershon and Jiang additional dataset, B |
If your exceeds cumulative successful referrals of 0.18–0.30 over average tenure of about four years, question the inputs.
Public Cases
Without controls, these are directional only.
| Case | Figure | Limitation |
|---|---|---|
| Dropbox, 2010 | Referrals contributed 35% of daily signups; two-sided incentives increased signups 60% | Founder presentation |
| Toss Securities, 2021 | More than 1.7 million first-event participants; new-account figures in Korean Cases | Total in-app event participation; friend-invitation contribution not isolated |
| Toss entertainment event, 2025 | More than seven in ten result viewers shared; 65% of share-referred visitors participated | No monetary reward; potentially existing-app participation, not signup conversion |
| PayPal, Airbnb, Uber | Widely repeated, but primary sources unverified | Excluded from this edition's numerical evidence |
Andrew Chen describes successful referrals as 20–30% of acquisition mix, but this is an opinion without supporting evidence (D).
Missing Evidence
Searches for Toss, Karrot, Baemin, KakaoBank, and others found no public Korean invitation-participation or invite-to-signup rates. Sharing and participation figures appear in Korean Cases. For B2B referral success, Implisit's 2014 referral lead-to-win rate of 3.6% is almost the only value found with a clear denominator (CRM leads, grade C); no recent equivalent was found. No public cannibalization share—the fraction rewarded despite arriving anyway—was found. Measure it by turning the program off for a subset of customers.
Creative, Retargeting, and Saturation
Review experimental effect sizes for targeting fit, messaging, retargeting timing, frequency, and budget saturation, and apply them to simulations.
Referral Experiments and Limits to Spread
Examine research on referred-customer value, incentive experiments, and viral diffusion, and why assuming K above 1 is risky.