What changed
On October 5, The Very Group announced a partnership with Made With Intent to personalise shopping experiences using behavioural signals.
Very reported a one-fifth increase in revenue per customer for one strategy during its first month, alongside an overall conversion improvement. That’s a promising early signal, but it isn’t a 20% lift across the business. The announcement doesn’t provide the sample size, comparison method or impact on margin.
How it works
Made With Intent describes a system where the ecommerce team chooses the experiences and commercial strategy, while its technology adjusts the audiences and timing.
In everyday terms, the team sets the options. The technology helps decide when to use them. That’s the supplier’s description; it doesn’t establish how much work Very’s team saves.
Why this matters
A shopper pausing on a product page might need clearer delivery information, reassurance about sizing or a better price. Giving every hesitant visitor a discount can be an expensive way to answer the wrong question.
A well-timed prompt could remove an obstacle. Repeated pop-ups or confusing offers could create another one.
The opportunity is to make those decisions more relevant without building endless manual rules. Someone still needs to own the offers, set commercial limits and check what the system is doing.
For brands funding promotions on a retailer’s website, there’s another question: did the offer generate additional profitable demand, or subsidise a purchase that would have happened anyway?
What we would test
Start with one category and one clear problem, such as shoppers leaving a product page without adding anything to their basket.
Agree a small set of approved responses. These could include useful product information, delivery details or a capped incentive. Keep a randomly assigned comparison group on the existing experience so there’s a fair basis for judging the result.
Measure:
- Profit per eligible visitor, accounting for discounts, returns and relevant costs
- Conversion and revenue per visitor
- Discount spend per order
- Returns, complaints and page performance
Review new and returning customers separately. Agree the measurement window before launching, and allow enough time for returns to show up. Include software and operating costs in the business case.
Keep clear limits on how often offers appear, who qualifies and what customer data is used.
Our take
This is a useful application of AI to a decision retailers already make every day: what would help this customer take the next step?
The next evidence to look for is sustained profit improvement across a meaningful share of traffic, alongside an experience customers find helpful.
Before buying another personalisation tool, identify the moment in your customer journey that needs a better decision. Then test whether the technology actually improves it.
