Field Notes · Note 001 · Customer value

The second purchase,
the window that decides customer value.

By Baptiste Pigaux28 September 20268 min read

A new customer is expensive to acquire. What decides whether they become profitable is almost never their first basket: it is whether they come back a second time, and when. This note shows the window in which that return is played out, and how to trigger it.

The question

Why does the second purchase matter more than the first, when is it played out, and what can marketing do during that period?

What we observed

Rounded orders of magnitude, observed in two assignments, anonymised. Every customer base is different; benchmark data will complement them.

The data

The observations come from two assignments with companies that sell directly to repeat customers. We publish neither their names, their sectors nor their volumes: only rounded orders of magnitude.

A common point that matters for what follows: in both cases the products are used up and bought again, with a usage cycle of a few weeks to a few months.

The method

We group customers by month of first purchase (a cohort), then follow, month after month, the share of each cohort that has made a second purchase. Three measures come out of it:

• the cumulative second-purchase rate at 3, 6, 12 and 18 months;
• the month when second purchases peak;
• the probability that a customer who has bought only once comes back, depending on the time elapsed.

We then compare cohorts with each other: if they look alike, the curve becomes a forecasting tool.

What the curves show

1. Value comes from frequency more than from basket size

In both bases, what separates large customers from small ones is not primarily basket size but the number of orders. A large customer’s average basket was only about 1.2 times that of an average customer. Bringing a customer back therefore matters more than making them spend more at once.

2. The second purchase comes early

Second purchases concentrate between the second and sixth month, with a peak around M+3 or M+4 when the product has a usage period. Nearly 8 repeat purchases in 10 happen within six months of an order.

3. The window closes quickly

After six months the curve flattens. A customer who still has not repurchased by month seven still has about a 45% chance of coming back at some point; by month twenty-four, only about 15%. Every month lost after the window costs customers.

4. Cohorts look alike, so it is predictable

Successive cohort curves were very close. Once yours is measured, the first months of a new cohort are enough to anticipate its value at one year, and therefore how much you can pay to acquire a customer.

The limits

What to do about it

First, measure your own window: three columns are enough (customer ID, date, amount). Then fill it. A typical welcome journey, to adapt to your usage cycle:

1. Day 0Order confirmation, with the loyalty programme benefits.
2. Day 1Thank-you and brand story: give a reason to come back, not a discount yet.
3. Day 10How to get the most from the product: a product used well gets bought again.
4. Day 30A personalised recommendation, with or without an offer (to be tested).
5. End of useA reminder just before the product runs out: the most profitable moment of the whole journey.

Always keep a control group that receives nothing: it tells you what the journey really brings in. And track three indicators per cohort: the second-purchase rate at 90 and 180 days, the median time between the first two orders, and the share of one-time buyers at 12 months.

Frequently asked questions

Baptiste Pigaux founded HOCUS after more than 20 years turning data and machine learning into business decisions, for more than 30 organisations. LinkedIn profile