Possibility 01
Suggest relevant catalogue items from a shopper’s stated requirements.
Marketing, content & creative production
Personalisation and recommendation experiences
Help visitors find relevant information or products using clear rules or models suited to your data and audience.
Explore your recommendation experienceHelp people find what fits their needs.
Define relevance
Agree the user need, available options and information that may be used.
Build the matching approach
Choose rules or a model and provide sensible fallback results.
Evaluate the suggestions
Test relevance, fairness, privacy and how users can change their preferences.
Explore each step. The approach is shaped around your tools, people and review needs.
See how it could workThe work behind the idea
Help people find what fits their needs.
A broad catalogue or resource library can ask visitors to do too much searching. A recommendation experience can narrow the choices using stated preferences or appropriately obtained data, with clear limits on the conclusions it draws.
Possibility 02
Tailor a resource selection to a visitor’s chosen business challenge.
Illustrative use cases, ready to adapt to your business. These are not claims of completed client work.
From question to a workable approach
How this could
come together.
- 01
Define relevance: agree the user need, available options and information that may be used.
- 02
Build the matching approach: choose rules or a model and provide sensible fallback results.
- 03
Evaluate the suggestions: test relevance, fairness, privacy and how users can change their preferences.
Define it together
A useful scope.
A clear starting point.
Available to discuss and scope. The right approach depends on your systems, information, people and the decisions you want to support.
What we can scope
- Define the recommendation objective, available data and privacy boundaries.
- Build and evaluate an explainable experience with appropriate fallbacks.
What helps us get started
- The products, resources or options to recommend
- Permitted preference or behaviour data
- Rules, exclusions and examples of useful matches
Worth thinking through
Recommendations should not rely on unjustified sensitive inferences or imply suitability beyond the information available. Users need clear choices and a way to explore alternatives.
Let’s talk about your next step
Make recommendations more relevant to your visitors.
Tell us what people are choosing and which signals you can appropriately use. We can explore a recommendation approach with clear controls and evaluation.
Explore your recommendation experience