E-commerce & Digital

Personalised product recommendations

Surface relevant next-best products based on browse history, purchase history, and customer segment. Displayed on product pages, basket, and post-purchase confirmation.

AutomationPattern-matchingRevenueCustomer experience

5–12%

uplift in units per order

Opportunity assessment

Business Impact
3

Moderate impact. Meaningful cost or revenue improvement, likely noticeable at function level but modest relative to total business scale.

Feasibility
3

Moderate effort. Requires configuration, prompt engineering, and testing. A capable team can get there but expect several months.

Data Readiness
2

Demanding data requirements. Relies on clean, integrated data across multiple systems. Significant data preparation work typically needed.

Risk Exposure
4

Low risk. Limited external exposure. A human reviews output before it reaches anyone outside the team.

Change Complexity
3

Moderate people impact. Part of someone's working day changes. Requires training and some adjustment time, but roles remain broadly the same.

Tooling required

Specialist AI toolFine-tuned model

Things to consider

  • Amazon's recommendation engine drives an estimated 35% of total revenue. The system uses collaborative filtering and content-based filtering across homepage, product pages, checkout, and email (McKinsey).

  • Requires a sufficient volume of interaction data (browse, purchase, add-to-basket) to generate meaningful recommendations. Businesses with fewer than 10,000 active customers per month may not have enough signal.

  • Specialist recommendation engines (Algolia, Klevu, Nosto, Dynamic Yield) exist for this use case and are usually faster to value than custom builds. Evaluate before deciding to build.

  • Recommendation placement matters as much as recommendation quality. Test placements (product page vs. basket vs. post-purchase) before optimising algorithm performance.

  • A common failure mode is 'you might also like' recommendations that are completely obvious — matching socks with trainers when the customer is already buying socks. Focus on surfacing non-obvious but highly relevant items that the customer might not have found independently.

  • Experiment starter: If you have any recommendation capability already live (even manual 'frequently bought together' rules), A/B test it against a collaborative filtering model built on your purchase history data. Measure click-through rate on recommendations and units per transaction. A measurable uplift in either metric justifies investing in a more sophisticated engine.

AI Transformation Playbook

Ready to assess your own opportunities?

The playbook gives you the full 150-opportunity directory, scoring tools, and 100+ templates for every stage of an AI transformation programme.