150% lift in repeat orders through AI-driven loyalty personalisation

PizzaRev's loyalty programme had been running for three years and accumulated 180,000 registered members — a substantial customer asset. The problem was almost none of them were active. Only 12% of members placed more than one order per month, and the average loyalty member visited just 1.7 times per quarter. The programme's economics were negative: the cost of points issuance and redemption was not being offset by incremental visit frequency.

Marketing communication was a weekly mass email blast — the same creative, same offer, same copy sent to all 180,000 members regardless of their order history, location, or favourite products. The team knew this was inefficient but lacked both the technical infrastructure and the analytical capability to do anything different. Email open rates had declined from 14% two years prior to 9% currently, with unsubscribe rates trending upward.

The loyalty platform, POS system, and email provider were three completely separate systems with no integration between them. There was no automated trigger when a customer became inactive — the first marketing communication a lapsed customer received was often the same generic blast as an active weekly visitor. The organisation had the data to do personalisation; it simply sat in three separate systems with no connective tissue between them.

We integrated the POS system, loyalty platform, and email provider (Klaviyo) via a centralised customer data platform, creating a unified profile for every loyalty member that combined their complete order history, redemption behaviour, location data, and communication engagement metrics. The integration surfaced patterns that had been invisible: 40% of lapsed members had stopped visiting immediately after a single negative experience with a specific menu item.

A clustering model segmented the 180,000-member base into 11 distinct behavioural archetypes — from 'Tuesday Lunch Regulars' who visited during the working week and ordered the same items consistently, to 'Special Occasion Visitors' who only appeared on birthdays and anniversaries based on redemption patterns. Each archetype had a distinct propensity profile for different offer types: discounts worked for price-sensitive segments, new product previews worked for variety-seekers, and recognition messaging ('You're our 500th visitor this month') worked for community-motivated customers.

An offer optimisation engine was built to match each customer to the highest-propensity offer type at send-time, taking into account their archetype, their recent engagement level, and the day and time of the communication. Send-time optimisation was applied at the individual level using Klaviyo's predictive analytics, rather than a single scheduled send time for all recipients.

A real-time personalisation layer was deployed feeding Klaviyo for email, their mobile app for push notifications, and an SMS gateway for high-propensity reactivation campaigns. Every outbound communication — whether triggered by a customer action, a scheduled campaign, or a lapse threshold being crossed — was personalised at the offer, copy, and send-time level. Segmentation was refreshed nightly based on the previous day's transaction data.

An eight-week A/B test comparing personalised versus generic communications across the full member base provided the performance validation. Personalised emails achieved a 34% open rate versus 9% for the control group. The 'next best offer' feature, launched in the customer app after six weeks, showed personalised offers to customers as they approached the restaurant — driving a measurable uplift in last-mile conversion for customers who had opened the app but not yet ordered.

A lapsed customer reactivation programme — targeting members who had not visited in 60+ days with a sequenced email-SMS-push notification series — reactivated 18,000 previously dormant members within the first three months. The programme's ROI was calculated at 3.1× based on the incremental visits and average order value from reactivated members versus the cost of the communications and offer redemptions.