GEIZHALS TRUSTS IN DANUBE.AI
Geizhals: Personal recommender without any personal data
A personalized product comparison API - integrated by Geizhals directly into their own product pages.
- Superior
- personalization from the first request
- Millions
- of offers ranked in production
- 0
- training records required
- 3–7%
- Interaction rate
- Industry average: <1%
- 2–5%
- Conversion rate
- Industry average: <0.5%
- 72%
- Net conversion
- Conversion per unique interaction
We deliver personalized product comparison as an API. Geizhals, Austria's biggest platform for finding a product's best price, calls it from their own product pages and renders the results in their own design. The API ranks millions of offers against what a shopper actually cares about - price, delivery, ratings, specifications - returns the comparison in milliseconds, and needs no personal data and no historical clickstream to do it.
Delivered as an API
The deliverable is a comparison API, not a hosted frontend. Geizhals sends a product context, gets back a ranked, explainable comparison, and renders it inside their existing product page - full control over markup, styling and placement stays on their side.
Integrated on the product page
Geizhals integrated the API into their product pages: browsing a product surfaces alternatives with similar or better price/value, side by side with the item the shopper is looking at.
No user data required
The API personalises to the visitor's preferences without receiving, storing or profiling any personal data - the request itself carries enough context to infer why someone is looking at a product.
Compare and boost features
Shoppers can compare features across alternatives and boost the attributes that matter most to them; each adjustment is one API call and the comparison re-ranks instantly.
Live at scale in Austria
Geizhals is the biggest platform for finding a product's best price in Austria; the API serves its catalogue in production, for example in the mobile phone category.
“The cool thing about this solution is that danube.ai finds out why a user browses a product. It works really well - without any user data.”
Capabilities
Multi-criteria ranking
Weigh dozens of product attributes at once without training a model on historical clickstreams.
One call, ranked comparison
Send the product context, get the ranked alternatives back in milliseconds - no retraining, no batch job, no frontend lock-in.
Explainable responses
Every position in the response carries the criteria that produced it, so your own UI can show shoppers why.
All solutions
Try Genetic AI on your own data
Convert your data into evolutionary format with DarwinNGC - or get in touch directly.