Beaconsfield Labs
Adaptive experimentation, rigorous by design. Built on peer-reviewed research from Cambridge.
The Team
We're repeat founders backed by experience scaling products to tens of thousands of users. Our previous startup raised £4M and served 30,000 users — now we're channelling that into something new.
Co-Founder
Student at the University of Cambridge. Previously co-founded and scaled a venture-backed startup to 30,000 users, raising £4M in funding. Focused on product, growth, and go-to-market.
Co-Founder
PhD researcher at the University of Cambridge specialising in biostatistics — specifically adaptive and covariate-adjusted randomised control trials. His research is the technical foundation of everything we build.
£4M
Previously Raised
30K
Users Served
2
Cambridge Founders
The Research
Our work is rooted in Lucas's doctoral research at Cambridge — bringing clinical-trial rigour to every domain where decisions are made under uncertainty.
Dynamically shift traffic towards better-performing variants during the experiment — not after. Fewer users see losing treatments.
Account for known confounders — device, geography, time of day — that classical A/B tests ignore, giving you cleaner and faster results.
Every method we use comes with formal error control — Type I error, power, and confidence intervals that hold up under adaptive designs.
Current Focus
Traditional A/B tests lock you into a 50/50 split — half your traffic goes to the losing variant for the entire experiment. That's lost revenue you can't get back. We fix that.
Fixed splits send equal traffic to every variant regardless of performance. If variant B is clearly worse, you're burning budget until the test ends.
Our algorithms shift traffic toward winning variants in real time. You still get statistical rigour — but with dramatically less wasted spend.
By minimising exposure to underperforming variants, companies recover revenue that traditional testing methods throw away.
Equal traffic to both variants throughout the test
Traffic shifts to the winning variant in real time
We're building in public. Join our mailing list for research updates, product news, and early access.
No spam. Unsubscribe anytime.