How we cut a healthcare partner's LLM spend to $0 per chart
RETAIL & COMMERCE

Personalised experiences that drive growth.

We built a real-time personalisation engine that learns what each shopper wants and adapts recommendations across web, app and store, lifting conversion and basket size.

  • Recommendation Systems
  • Machine Learning
  • Real-Time Data
Client
Leading retail partner
Industry
Retail & Commerce
Timeline
5 months
Capabilities
AI/ML/Personalisation
Smarter engagement
Loyal Customers
THE CHALLENGE

The problem wasn't the traffic. It was treating every shopper the same.

"Customers don't want more choice. They want the right choice."

The retailer had rich purchase and browsing data across channels, but shoppers saw the same generic pages and offers. Relevance was limited to a few manually maintained segments.

Scale

Thousands of products and millions of visits.

Context

Intent shifts by visit, device and moment.

Speed

Recommendations must appear in milliseconds.

OUR APPROACH

From behaviour to recommendation.

We designed a real-time approach that turns customer signals into relevant experiences on every channel.

Customer Signals

Browse, search, purchase

Unified Profile

Resolve identities across channels

AI Model

Learn preferences and context

Ranking

Score products in real time

Merchandiser Review

Set guardrails and goals

Experience

Personalised pages and offers

THE SOLUTION

Intelligence designed around the shopper.

We built a personalisation engine that plugs into the storefront and marketing tools, giving every customer a relevant experience while merchandisers stay in control.

Real-time Recommendations

Products ranked for each visit.

Customer Segments

Dynamic audiences that update themselves.

Channel Integration

Web, app, email and in-store.

Merchandiser Control

Business rules and guardrails stay in place.

Personalisation console showing a shopper profile, live signals, ranked product recommendations and a conversion uplift chart
INSIDE THE SYSTEM

Built for relevance.
Built for scale.

Our architecture serves personalised results in real time: scalable through peak trading, privacy-aware and easy to extend with new channels.

Technical details
Layered system architecture
  • Data IngestionEvents, orders, catalogue
  • Identity & ProfilesUnified customer view
  • Modeling LayerRecommendation and propensity models
  • Real-time ServingLow-latency ranking APIs
  • IntegrationStorefront, CRM, email
  • Merchandising WorkflowRules, testing and feedback
SecureConsent and privacy built in
ScalableHolds through peak trading
FlexibleAny channel or catalogue
ReliableSafe fallbacks always on
THE RESULTS

Meaningful results for growth.

By combining machine learning with real merchandising workflows, we helped our partner achieve measurable improvements.

2.5x

Conversion uplift

On personalised placements.

35%

Higher ADV

Larger average order value.

Real-Time

Recommendations

Updated on every interaction.

BEFORE AFTER

From one-size-fits-all to one-to-one.

The same catalogue. The same customers. A fundamentally better way to shop.

  • Generic pagesEveryone sees the same thing
  • Manual segmentsUpdated a few times a year
  • Channel silosWeb, app and store apart
  • Missed intentRelevant products buried
  • Personalised pagesTailored to each visit
  • Live audiencesSegments update in real time
  • Connected journeyOne profile across channels
  • Relevant discoveryThe right product surfaces first
WHAT WE LEARNED

Personalisation works when it feels helpful.

The biggest gains came from testing small, explainable changes with merchandisers rather than chasing complexity. Clear guardrails kept recommendations on-brand and trusted.

Understand

What shoppers actually value.

Test

Prove lift before scaling.

Design for trust

Be relevant, not intrusive.

READY TO BUILD WHAT'S NEXT?

Not sure where to start? Start with the problem.

Tell us what you're trying to achieve. We'll help you explore the right approach from idea to impact.

View All
Contact us

Have a problem worth getting right?

Send us the paragraph you'd normally write to a colleague. We'll get back to you within 24 hours.

hello@tathastha.com