Last Updated on April 18, 2026
AI recipe engine
8 orders to learn you
11% reactivation rate
from lapsed customers
The move
Gousto is a recipe box company. They deliver ingredients and recipes to your door. In a market where HelloFresh and Mindful Chef are fighting for the same customers, churn is the biggest problem. People sign up, try it for a few weeks, then cancel. The margins are thin. Keeping customers matters more than finding new ones.
Gousto’s answer was a recommendation engine called “Rouxcommender.” It analyses your ordering patterns, what you’ve cooked before, what you’ve skipped, what ingredients you favour, and predicts what you’ll want next. The latest version uses transformer-based architecture, the same technology behind ChatGPT, to understand preferences.
The breakthrough was speed. They halved the number of orders needed to understand a customer’s preferences, from 16 recipes down to 8. That means the system gets useful faster, and customers start seeing personalised menus before they get bored and cancel.
On the email side, Gousto partnered with Paperplanes on a reactivation campaign targeting churned customers. Using dynamic creative that pulled in weekly meal updates and QR codes for easy re-subscription, they achieved an 11% reactivation rate. Customers who received the campaign also showed significantly higher retention rates over time.
Why it worked
Most subscription businesses treat personalisation as a marketing feature. A “recommended for you” section on the homepage. A personalised subject line. Gousto treats it as the core product experience. The menus you see, the order they appear in, the recipes highlighted in your emails, all shaped by what the algorithm has learned about you.
Getting to useful personalisation in 8 orders instead of 16 is a commercial decision, not just a technical one. Most subscription churn happens in the first month. If your system doesn’t learn fast enough, customers leave before it gets good. By halving the learning curve, Gousto gives the algorithm time to prove its value before the customer gives up.
The reactivation work shows the same thinking applied to lost customers. Instead of generic “we miss you” emails, Gousto sent personalised direct mail showing specific meals the customer would actually want. That relevance is why 11% came back, a rate that significantly outperforms industry averages for win-back campaigns.
The principle
Personalisation that reduces churn is worth more than personalisation that increases clicks. The goal isn’t to make your emails feel clever. It’s to make your product so tailored to each customer that leaving feels like a loss.
Speed to relevance matters. The faster your system learns what a customer wants, the less time they spend in the “generic experience” phase where most cancellations happen.
Steal this
You probably don’t need a transformer-based AI. But you do need the thinking behind how Gousto uses data to keep customers.
Track what customers choose and what they skip. You don’t need machine learning to notice patterns. If a customer always buys your vegetarian options and ignores the meat ones, stop showing them meat. Manual segmentation based on behaviour beats no segmentation at all.
Speed up your learning curve. What’s the minimum number of interactions you need to understand a customer’s preferences? Find it and optimise for it. Ask the right questions in onboarding. Use early behaviour to segment quickly. Don’t wait for months of data before personalising.
Make reactivation personal. When a customer leaves, don’t send a generic discount code. Send them something based on what they actually bought or used. ‘We’ve got three new vegetarian boxes this week’ is more compelling than ‘15% off your next order’ if the customer was a vegetarian subscriber.
Measure retention, not just engagement. Opens and clicks are vanity metrics for subscription businesses. Track how many customers are still active after 30, 60, and 90 days. If personalisation isn’t improving those numbers, it’s not working, regardless of how good your open rates look.
See the brand in action
Gousto showing how they make recipe boxes feel personal on TikTok.
Sources & further reading
- How Databricks Helps Gousto Create Personalised Recipe Recommendations at Scale · Databricks
- How Gousto Achieved 11% Customer Reactivation · Paperplanes
- Gousto on Developing an ‘Empathetic’ Data Strategy · Marketing Week
The Whito verdict
Gousto shows that the most valuable marketing isn’t always about acquisition. Their recommendation engine and personalised reactivation campaigns are retention tools disguised as product features. They reduce churn by making the experience better for each individual customer, not by offering discounts or running flashy campaigns.
If your business has a churn problem, look at how quickly your product becomes personal. If new customers are getting a generic experience for too long, that’s probably where you’re losing them.
Scale StageAre you keeping the customers you’re winning?
Get a free Growth Report and find out whether your retention is as strong as your acquisition.
Get Your Free Growth Report