How to increase referral and repeat purchase rate

To increase referral and repeat purchase rate, measure it (second purchases plus referred deals, divided by customers won, at month 12), then give it an owner and a clock. The biggest effect measured directly on this stage is referring itself: in one telco field experiment, recommenders’ defection rate fell from 19% to 7% within a year.

• The metric: (customers with a second purchase + referred deals closed) ÷ customers won in the cohort, read at month 12.
• Why it compounds: in a study of about 10,000 bank customers, referred customers were worth at least 16% more than similar customers acquired in other ways.
• Biggest lever measured on this stage: joining a referral programme. Recommenders’ defection fell from 19% to 7% and their monthly revenue rose 11.4% (Journal of Marketing, 2013).
• Biggest lever measured next to it: response speed. Firms that tried to contact web leads within an hour were nearly 7x as likely to qualify them (Harvard Business Review; 1.25 million leads, 42 US firms).
• What to change first: add a referral-source field to the CRM, then put the ask on a clock: within 14 days of the customer’s first measurable result.

“My customers buy once and never come back”: how referral and repeat rate is measured

Referral and repeat purchase rate is stage 17 of the 17 sales pipeline stages and what each one costs. It happens after the sale, so it never appears in a funnel diagram and usually has no owner. Measure it as two numbers read from the same cohort, because the two halves are fixed by different levers:

• Repeat rate = customers from the cohort who made a second purchase within 12 months ÷ customers won in the cohort.
• Referral rate = closed deals whose source field names a customer from the cohort ÷ customers won in the cohort.

Use a cohort (every customer won in one quarter), not a calendar month. A monthly ratio mixes last week’s buyers with customers from two years ago, so it moves when nothing has changed. Before you trust a move, check the noise. For a cohort of 200 customers with a 20% repeat rate, the standard error is √(0.2 × 0.8 ÷ 200) = 2.8 percentage points. At 50 customers it is 5.7 points, which means a move from 20% to 25% cannot be told apart from chance.

A referral and repeat rate you cannot split into its two halves cannot be fixed, because repeat and referral respond to different levers.

Why referral and repeat is the stage that compounds after the sale

Every other stage runs on people you paid to reach; this one runs on people who already paid you. Two peer-reviewed studies carry the evidence. In Schmitt, Skiera and Van den Bulte (Journal of Marketing, 2011), the authors tracked about 10,000 customers of a German bank for almost three years. Referred customers were worth 40 euros (at least 16%) more over six years than comparable customers. After 33 months, 82.0% of referred customers were still active, against 79.2% of the rest. Set against the 25-euro reward, the authors estimate a return of about 60%.

The second study shows the effect running in the other direction. Garnefeld and colleagues (Journal of Marketing, 2013) found that making a referral changes the referrer. In a field experiment with a cellular telecoms provider, recommenders’ defection fell from 19% to 7% within a year. Their average monthly revenue grew 11.4% compared with a matched control group. Referral drives repeat purchase, and that loop is why the stage compounds.

A customer who refers you becomes a better repeat customer, so the referral ask is also a retention lever.

The levers that raise referral and repeat rate, ranked by effect size

No public study compares these levers head to head. The effects below come from different studies, settings and outcome measures, so they are not directly comparable; the ranking is our judgement, marked down where the evidence comes from a different stage.

Lever 2 has the largest raw multiplier, but it ranks second because HBR measured it on all online enquiries, not on referred ones. Keep our own figure separate from that study: in our own client work on paid-ad leads, we typically see speed to lead lift conversion about 3x on its own. That is an operator figure, not research, and we have not measured it on referred enquiries separately. How to cut response time is covered in speed-to-lead conversion levers.

The first-result clock: when to ask for a referral and a second purchase

Referral and repeat rates are usually low because nobody is scheduled to ask. The first-result clock is our decision rule for fixing that. It sets three timers, each started by an event, not a date:

1. 14 days from the first measurable result: ask for the referral. Garnefeld et al. found the loyalty effect of recommending is “particularly pronounced for newer customer–firm relationships”, so an ask held back until the annual review misses the window where it helps most.
2. 1 hour from a referred enquiry arriving: make first contact, including evenings and weekends. The referrer’s credibility rides on your speed.
3. 1x the customer’s normal buying cycle: open the second-purchase conversation. By 2x the cycle the customer is lapsing, and the problem has become signal-based expansion and win-back rather than repeat purchase.

The 14-day and 1x-cycle timers are our starting settings to test on your cohort, not measured thresholds. The 1-hour timer comes from the HBR lead-response study.

Under the first-result clock, the customer’s first result starts the referral ask. A calendar date does not.

Worked example: what a 10-point move in repeat rate is worth

Illustrative inputs; substitute your own.

• Cohort: 200 customers won in a year, average first sale $4,000.
• Repeat rate 20%, average second purchase $2,500: 200 × 0.20 × $2,500 = $100,000.
• Referral rate 0.10 closed referred deals per customer, each worth $4,000: 200 × 0.10 × $4,000 = $80,000.
• Stage-17 revenue today: $180,000.

Now raise repeat rate to 30% and referral rate to 0.15. Repeat revenue becomes 200 × 0.30 × $2,500 = $150,000, and referral revenue becomes 200 × 0.15 × $4,000 = $120,000. The total is $270,000, up $90,000 or 50%. At a $4,000 first sale, that equals 22.5 extra new customers. Multiply 22.5 by your own fully loaded acquisition cost to see what the same revenue would cost through the front of the funnel.

Comparing two 200-customer cohorts, the 10-point move is about 2.3 standard errors of the difference (10 ÷ 4.3), so it clears the usual bar of 2; on cohorts of 50 it would be about 1.2.

On a 200-customer cohort, a 10-point rise in repeat rate plus half a referral per ten customers adds 50% to stage-17 revenue with no extra acquisition spend.

What should I change first? The threshold table

Fix the first row that describes you before moving down.

If referrals have not just been low but have been falling for two quarters, that is decay rather than a missing process. The active-referrer test for referrals that have dried up is the right diagnostic for that.

For referral and repeat rate, measuring comes first: a missing referral-source field outranks every other fix.

What running a referral and repeat programme yourself costs in hours

All of this can be run in-house. The honest cost for the 200-customer example:

• Measurement: one day to add the source field and backfill it, then about 2 hours a month for the cohort report. That is roughly 32 hours in the first year.
• The referral ask: a personal message at the 14-day mark takes about 10 minutes per customer. Across 200 customers, that is about 33 hours a year.
• The repeat trigger: a CRM task per customer, dated from their own buying cycle. Setup takes about a day, and after that it is part of normal account work.
• The one-hour response: this is the part that breaks. A 40-hour roster covers 24% of the 168 hours in a week, and referred enquiries arrive when the referrer happens to mention you, not during your office hours.

The first three total about 70 hours in the first year of ordinary account-manager work. The fourth is a staffing question, not a skills one. Businesses that decide the fourth is not worth rostering tend to hand it to an AI appointment-setting service that answers and books around the clock.

The referral and repeat stage costs roughly 70 hours in the first year to run by hand for 200 customers. The hard part is being available at all hours, not the workload.

Referral and repeat purchase questions

What is a good repeat purchase rate?

There is no credible cross-industry benchmark for repeat purchase rate. Published figures are mostly vendor numbers from a single platform, and they use different windows and denominators. Measure your own rate on a 12-month cohort and compare it with the same cohort a year earlier. A move is only real if it is larger than about two standard errors of the difference: roughly 16 points for cohorts of 50 customers at around 20%, or 8 points for 200.

Should I pay customers for referrals?

Usually, yes, but think about who receives the reward. In four experiments, Ryu and Feick (Journal of Marketing, 2007) found that rewards increased referral likelihood, most of all for referrals to weak ties and for weaker brands. For strong ties and stronger brands, giving at least some of the reward to the referred friend worked better. In the bank study, a 25-euro reward returned about 60% over six years.

Do referred customers stay longer than other customers?

In the best-documented case, yes, modestly. Schmitt, Skiera and Van den Bulte (2011) found that 82.0% of referred bank customers were still active after 33 months, against 79.2% of comparable non-referred customers. The retention gap persisted over time, while the margin gap eroded.

How quickly should I follow up a referral?

Within the hour if you can. In a study of 1.25 million online leads at 42 US companies reported in Harvard Business Review, firms that tried to contact a lead within an hour were nearly seven times as likely to qualify it as firms that tried even an hour later. That study covered all online leads, not referrals specifically. In a separate audit of 2,241 US companies, the average response, among those that responded within 30 days, was 42 hours.

Does asking for referrals annoy existing customers?

The evidence points the other way. In a telecoms field experiment reported in the Journal of Marketing (Garnefeld et al., 2013), customers who made a referral became more loyal: their defection rate fell from 19% to 7% within a year. An ask sent to a customer with an open complaint is a different matter, so check for open issues first.