Recommendation EnginesThe Right Thing, to the Right User.

A recommendation engine is a model that ranks the items the user is most likely to engage with — products, articles, videos, jobs, the things the user actually sees. The work is the data pipeline, the model, the ranking, the A/B test framework, the eval suite and the feedback loop. The model is one component, the system is the product.

We build recommendation engines with the data pipeline, the model, the ranking, the A/B test framework, the eval suite and the feedback loop as part of the architecture from sprint one. The output is a recommendation engine that lifts conversion, not a model that lives in a notebook.

This is the wrong engagement if the catalogue is genuinely small (a "related items" widget will do) or if the use case is a one-off report.

• Right Thing, Right User — A recommendation engine that surfaces the right thing, not a model that lives in a notebook.
• A/B Tested — An A/B test framework, with the metrics and the variants the team can act on.
• Feedback Loop — The feedback loop, the model retraining and the monitoring the recommendations need.
• Real-Time Capable — Real-time ranking for the surfaces (homepage, product page) that need it.