Turning first orders into second orders
Most people who buy a bag of beans online buy once and never come back. Here is how to tell who is likely to return, and what to send them, without spamming everyone.
The problem
The café is busy and the roastery keeps growing, but most people who order beans online order once. Every customer gets the same newsletter, whether they bought a bag of espresso last week or a gift box last Christmas.
What was really causing it
The store knows what each customer bought and when, but nothing acts on it. Someone who bought a bag of espresso three weeks ago is about to run out, and nobody tells them. Café regulars and online buyers are two lists that never meet.
The store knows what each customer bought and when, but nothing acts on it.
What the research says
- 71% of consumers expect companies to deliver personalized interactions (McKinsey & Company, The value of getting personalization right, or wrong, is multiplying)
- 76% get frustrated when that does not happen (McKinsey & Company, The value of getting personalization right, or wrong, is multiplying)
For a roaster, personalization does not need a clever algorithm. It means the obvious things done at the right time: a reminder when the bag is nearly empty, a brewing tip for the beans someone actually bought, and no espresso newsletter for the person who only buys decaf for the filter machine.
What we built
Groups based on what people bought and when, and a few journeys that fit each: a reminder timed to when a bag runs out, brewing tips after a first purchase, a win-back after a long gap, and an easy way to turn a regular order into a subscription. All measured against a holdout group.
- Grouped Customers grouped by what and when they bought.
- Timed Messages sent when they are useful, not on a fixed weekly blast.
- Measured Every journey compared with a holdout group that gets nothing.
- Pruned Journeys that do not earn their keep are switched off.
What stays as it is
Your online store, the till in the café and your email platform.
Where AI fits, and where it does not
AI can draft message variations for a person to approve, and suggest likely next purchases. Timing and audience rules stay plain and visible.
What it is worth
What comes next
A simple loyalty view that links café regulars to their online orders, and roast suggestions on the site based on the same data.
Questions people ask
- Will this mean more emails?
Usually fewer. The weekly message to everyone is replaced by a handful sent when they are useful, and a journey that does not beat the holdout group is switched off.
- What is a holdout group?
A small random share of customers who get none of the new messages. Comparing them with everyone else shows what the messages actually changed, instead of crediting them with orders that would have happened anyway.
- Do we need new software?
Most store and email platforms can already do this. The work is in the groups, the timing and the measurement.
- The value of getting personalization right, or wrong, is multiplying, McKinsey & Company (2021)
Canonical URL: https://ugibo.com/case-studies/first-orders-into-second-orders