Case Study · Sneakers & Streetwear · AI & CRM
True 1-to-1 personalisation at scale — powered by AI.
The Sole Supplier couldn’t offer discount codes or control pricing. So we built an AI agent that did the next best thing — finding the right product, on sale, for each individual user, and writing a personalised push notification around it. The results changed how we think about reactivation entirely.
133%
Uplift in purchases within 7 days vs. original message
75%
Uplift in release reminders set vs. original message
1-to-1
Truly personalised push notifications — product, copy, and image — built individually for every user at scale
Client
The Sole Supplier
Sector
Sneakers & Streetwear
Engagement
Head of CRM
Focus
AI-powered reactivation
Tools
Braze · AI Agent
The Challenge
Winning back churned users when your most obvious tools aren’t available.
Reactivating churned users is a challenge for any brand. Discounts and promotional codes are a well-proven lever — but as an affiliate business, The Sole Supplier has no control over pricing or promotions. That doesn’t mean winning users back is impossible, but it does mean the approach needs to work harder on relevance rather than relying on an incentive.
The reactivation journey already included an automated push notification, with liquid templating pulling in products from the catalogue. But the messaging and featured products were broadly targeted rather than truly personalised — the same copy structure going to everyone, with limited consideration of individual user behaviour or interests. There was a clear opportunity to do more.
The constraint
Without the ability to offer discount codes or promotional pricing, the most effective lever available was relevance — finding exactly the right product, for exactly the right person, at exactly the right moment. The question was whether that could be delivered at scale.
The Insight
Relevance is a more powerful incentive than a discount — if you can deliver it at scale.
The insight was simple: a churned user doesn’t need a generic message nudging them back. They need to see something that feels like it was chosen specifically for them — a product they’d actually want, at a price that makes it feel like an opportunity worth taking.
The problem was scale. Writing truly personalised messages for thousands of churned users wasn’t achievable manually. But with an AI agent handling the product selection, copy, and imagery for each individual, it suddenly was.
The question became: could AI-powered personalisation outperform our existing broadcast approach — in a head-to-head A/B test, in the same journey, with the same audience?
What We Built
An AI agent that builds a personalised push notification — product, copy, and image — for every individual user.
Rather than a single message sent to everyone, we built an AI agent that runs at the point of send for each individual user. It selects a relevant product currently on sale, writes bespoke push copy around it, and pulls the right image from the product catalogue — all automatically, all at scale. The result lands in the user’s notification tray as something that feels hand-picked, not broadcast.
AI Agent · How It Works
1
User enters the reactivation journey
A churned user meets the criteria to receive a reactivation push. The AI agent is triggered at the point of send.
2
Agent selects the right product
The agent reviews the user’s behaviour and interests alongside current on-sale inventory, identifying the most relevant product for that individual.
3
Agent writes the push copy
Bespoke notification copy is generated around the selected product — personalised in tone, product reference, and framing. No templates, no liquid text.
4
Agent finds and attaches the product image
The right product image is pulled from the catalogue and added to the notification automatically — no manual asset management required.
5
Personalised push is sent via Braze
The fully assembled notification — unique to that user — is delivered. To the recipient, it feels like someone chose this specifically for them.
Before vs. After
Original message
Broadcast push with liquid templating
The same copy structure sent to all churned users, with product names pulled in via liquid text from the catalogue.
— Same message framework for every user
— Product pulled automatically but copy not written around it
— No consideration of individual user preference or behaviour
AI agent version
Fully personalised push built per user
Product, copy, and image selected and written individually for each user by the AI agent at the point of send.
— Product chosen based on individual user behaviour and interests
— Copy written specifically around the selected product
— Image automatically sourced and attached from the product catalogue
The Results
133% more purchases. 75% more reminders set. Same journey, same audience.
The AI agent version was tested directly against the original message within the same reactivation journey — same audience, same timing, same channel. The results were decisive. Users who received the AI-personalised push were 133% more likely to make a purchase within 7 days, and 75% more likely to set a release reminder — one of the strongest indicators of long-term re-engagement on the platform.
133%
More purchases within 7 days
75%
Uplift in release reminders set
Users who received the AI-personalised push were significantly more likely to set a release reminder — a strong signal of genuine re-engagement with the platform.