All answers

The Prevailing
Answers for
Business model:
Business model, pick up to two

How do we measure whether our referral program brings customers we would not have won anyway?

The PrevailingPublished Nov 2, 2026

Our answer

Keep a random 10 to 20% of customers from every referral prompt for 30 days, sized first with a sample size check. Compare new customers and revenue between the groups, not the referrals your tool attributes.

Confidence
Medium
Strongest proof
1 controlled test
1 practitioner, 4 opinions or claims
Sources
6 sources
6 statements

New here? Take the 1-minute tour

Why this is the answer

3 reasons and our judgment

A random holdout shows what referrals add. Eightx suppresses the post-purchase referral prompt for a random 10 to 20% of customers for 30 days and compares attributed revenue. Its planning benchmark has most programs failing that test in the first 90 days.

Holdout customers must see no referral prompts anywhere. ATTN Agency (paid media agency) keeps its holdouts out of every channel, email included. A holdout that gets the same incentive through another channel breaks the measurement (Voucherify, a promotion software maker).

Small brands need a bigger holdout than they expect. In one sizing example, detecting a 20% relative lift on a 2% conversion rate takes about 25,000 people per group (ATTN Agency). A 10% holdout on a small list often detects nothing worth acting on (Kixie).

Our judgment

  • Score the referral program on the new customers it adds. We compare new customers and store revenue between the groups, not only attributed revenue, combining Eightx's holdout with Voucherify's scoreboard.
  • Size the referral holdout group on new-customer counts before launch. At ATTN Agency's example of about 25,000 people per group, for a 2% baseline and a 20% relative lift, a 10% holdout needs a base of about 250,000 customers. Smaller brands need a larger share or a longer test.
  • Don't drop a program on one early failed holdout. Rerun it later, going by Eightx's planning benchmark that most programs pass after 12 to 18 months.
  • Judge referred customers on repeat orders and further referrals as well as cost. We apply the Rotterdam researchers' advice to track referral quality, not only volume, and WiserReview's referral acquisition cost.

Our judgment, built from Eightx, Voucherify, ATTN Agency, Kixie, Rotterdam School of Management and WiserReview: our reading of these sources, not a finding from any one of them.

What the sources say

6 statements from 6 sources
  • Ting Li and co-authors2026-10Supports

    A Rotterdam School of Management article on a field experiment paying loyalty points to 160,000+ telecom app users: an uncertain sender reward raised referrals 20.9%, an uncertain recipient reward cut them 12.3%.

    “The most effective referral design combines an uncertain reward for the sender with a guaranteed reward for the recipient”

    Read it on rsm.nl

  • KixieSep 27, 2026Caveat

    Kixie, which sells sales texting software, says to randomize SMS holdouts by customer, size them from baseline conversion and the smallest lift worth acting on, and keep calls and email equal across groups.

    “Ten percent is not a rule, it is a habit.”

    Read it on kixie.com

  • WiserReviewJun 24, 2026Supports

    WiserReview, which sells review software, advises comparing a referral reward with your customer acquisition cost. It says a $10 to $20 reward can pay off if ads cost $30 per customer. It gives no data.

    “If you normally spend $30 to acquire a customer through ads, a $10-$20 referral reward may still be profitable.”

    Read it on wiserreview.com

  • VoucherifyJun 2, 2026Supports

    Voucherify, which sells referral and promotion software, says holdout testing is the gold standard for incrementality, as long as the control group never gets the same incentive through another channel. No data.

    “Once your holdout starts receiving the same incentive through another channel, the measurement falls apart.”

    Read it on voucherify.io

  • EightxJun 1, 2026Supports

    Eightx, a fractional CFO firm, says mature DTC refer-a-friend programs drive 5-10% of revenue at the median and up to 30% at the top quartile. It says to build loyalty first and referral second.

    “Mature DTC refer-a-friend programs drive 5-10% of revenue at the median and up to 30% at the top quartile”

    Read it on eightx.co

  • ATTN AgencyMar 12, 2026Supports

    ATTN Agency's general guide says holdout tests usually hold out 10 to 20% of the audience from all channels; at a 2% baseline, detecting a 20% lift needs about 25,000 people per group.

    “Holdout testing randomly excludes a portion of your audience from marketing campaigns, then compares business outcomes between the exposed (treatment) and unexposed (holdout) groups.”

    Read it on attnagency.com

In the full answer

For readers who sign in
  • What to do this week3 steps
  • The numbers behind it2 figures from cited sources
  • Where each source standsA chart placing each source on the question
  • The strongest disagreementThe best case against the answer, from a cited source
  • How sure we areWhat its confidence rests on, and what would change the answer

Read next

More in Retention and loyalty