Squoosh Bench: Evaluating UX-1 on 318 Historical Website A/B Tests
Abstract
Experimentation teams must decide which proposed website changes to test with live traffic. We evaluated whether forecasts from simulated shoppers could inform that choice. We screened 412 historical website A/B tests and evaluated four systems on the 318 that ran longer than 21 days. Each system predicted an original win, a variant win, or an inconclusive result. Page roles were masked and recorded outcomes were withheld during inference. Squoosh UX-1, a model fine-tuned on human browsing data, simulated 1,000 shoppers per arm and correctly classified 267 of 318 outcomes (84.0%, 95% Wilson interval 79.5-87.6%, treating experiments as independent). Qwen3 Next 80B, the strongest general-purpose configuration on outcome accuracy, correctly classified 196 of 318 (61.6%). UX-1 favored the variant in 126 experiments. In the live tests, 118 of those variants won (93.7%), four lost, and four had inconclusive results. The remaining 192 experiments included 19 additional variant wins.
Community
AI has made building a website change take minutes, but finding out whether customers like it still takes weeks of live testing. Squoosh simulates how people use a site, so teams can see what their customers will likely do before anything goes live.
This paper is the proof. We took 318 real website A/B tests that had already finished, hid the results, and asked our model, UX-1, to pick the winner. It got 267 right (84%). The best general-purpose AI model we tried got 62%.
Teams use this to check designs, find where customers get stuck, test pricing and offers, and pick which ideas are worth a live test.
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- What Changes When Fact-Verification Scores Improve? Evidence and Answer Accounting Across Trained Verifiers and LLMs (2026)
- The Company a Question Keeps: Auditing Open Jev-Style Models for Retrieval (2026)
- Benchmarking candidate coverage and rejection policy transfer in typed decision models (2026)
- Same Scores, Different Decisions: Evaluating JEV and Language Models for Legal Document Understanding (2026)
- Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts (2026)
- Truncated Reasoning Under a Budget: Evaluating System-1 Models as Decision-Makers (2026)
- Measuring User Success in Randomized Search Experiments: Interpreting Good Abandonment in Generative AI Search (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper