AI ApplicationCommercial ExperimentStrategic AdvisoryIn the lab

Recommendations Engine

A personalisation engine that turns every patron's unique trail of behaviour into recommendations that feel genuinely individual.

01 - The Problem

Reasonable recommendations aren't personal ones.

Most venues recommend performances using a narrow set of inputs - staff knowledge and simple "people who bought this also bought that" logic.

  • - Thousands of patrons receive recommendations that are merely reasonable rather than genuinely personal
  • - Marketing teams spend valuable time deciding what to promote instead of focusing on strategy and creativity

02 - The Insight

Every patron leaves behind a unique trail of interests - not just the shows they've attended, but the genres they gravitate towards, how adventurous they are, how recently they've engaged, and how similar their behaviour is to thousands of other patrons. The challenge isn't finding data. It's turning all of those signals into recommendations that feel personal.

Instead of asking

Which show should we promote this week?

We reframed it as

Which performances is this individual most likely to enjoy right now?

03 - The Build

How it works.

The Recommendations Engine combines multiple sources of intelligence to build a personalised recommendation for every patron.

  1. 01

    Behavioural intelligence

    Ticket purchase history, booking frequency, attendance patterns, and patron affinity scores generated by PIE.

  2. 02

    Event intelligence

    AI-generated event summaries, genre and thematic similarities, and similarity between productions.

  3. 03

    Recommendation logic

    Multiple recommendation algorithms, intelligent ranking and weighting, and continuous refinement as new data arrives.

  4. 04

    A ranked, personal portfolio

    The result is a ranked portfolio of performances that is unique to every individual.

04 - The Output

What users actually get.

For every patron, the system produces a personalised list of recommended events. These recommendations aren't simply "similar shows" - they're chosen because the engine believes they represent the strongest opportunity to deepen that patron's relationship with the venue. They can power personalised emails, customer portals, digital consultants, future website experiences, and audience insight tools.

  • - Personalised emails
  • - Customer portals
  • - Digital consultants
  • - Future website experiences
  • - Audience insight tools

05 - The Impact

What changed.

  • 01

    More relevant customer communications

    grounded in individual behaviour, not broad segments

  • 02

    Increased patron engagement

    through recommendations that feel genuinely personal

  • 03

    Better discovery of lesser-known performances

    beyond the obvious hits

  • 04

    Reduced manual campaign planning

    freeing marketing teams for strategy and creativity

  • 05

    Improved commercial performance

    driven by relevance, not volume

The shift

"Generic campaign marketing"

"Genuine personalisation - helping patrons find experiences they'll value"

06 - The Future

Where this goes next.

The Recommendations Engine is designed to become the recommendation layer across an organisation. Powered by PIE, it can support a continuous personalisation model rather than a campaign-based one.

It can be extended to

  • - Automated "next best performance" recommendations
  • - Cross-venue discovery
  • - Personalised season planning
  • - Subscription suggestions
  • - Dynamic website content
  • - AI-powered audience assistants

Long term, tools like this could

  • - Recommendations become continuous, not campaign-based
  • - The question shifts from "what should we market today?" to "what should this patron discover next?"
  • - Personalisation becomes an always-on layer, not a periodic activity

Building something similar?

If this resonates with a problem you're working on - let's talk.