ABSee / Compare / DoWhatWorks alternatives
CompareDoWhatWorks alternatives. ABSee: 10,000+ experiments from 1,150+ brands.
ABSee is experimentation intelligence for ecommerce: a record of 10,000+ experiments run by 1,150+ brands on their live sites, with the page before, the version tested, and whether it was kept. It works inside Claude, ChatGPT or Gemini, so you ask in plain words and get real experiments back.
Updated October 2026 · Source: ABSee record · Methodology →
| Option | What it is | What you see | How you use it |
|---|---|---|---|
| ABSee | 10,000+ experiments run on live sites by 1,150+ ecommerce brands | Before and after screenshots, dates, and whether the brand kept the change | Ask in plain language inside Claude, ChatGPT or Gemini |
| DoWhatWorks | A service that tracks live experiments on brands' websites | Detected tests and a prediction of which version is likely to win | A web app and a newsletter |
| ABWatcher | A service that watches brands' websites for live A/B tests | Both versions of a test, with an AI-written summary | A web app |
| GoodUI | A library of UI patterns built from A/B test results | Named patterns and the tests behind them | A website and a newsletter |
| Baymard Institute | Ecommerce UX research from usability testing | UX guidelines and site reviews against them | A website and a review tool |
Why teams pick ABSee
- Real experiments, not hypotheticals. Every entry is an experiment a named ecommerce brand ran on its own live site.
- The page before and the version tested, side by side. You see exactly what changed, with the dates it ran.
- What the brand did next. Each experiment records whether the brand kept the change or rolled it back. How this is decided.
- Inside the assistant you already use. Ask in Claude, ChatGPT or Gemini and get the evidence for and against your own idea, before you spend your own traffic on it.
- Updated daily, ecommerce only. Home, category, product, cart, checkout and search pages, from 1,150+ brands.
How ABSee works
ABSee connects to Claude, ChatGPT or Gemini over MCP, the open standard assistants use to call outside tools. You ask in plain words, for example whether anyone rolled back a free-shipping bar, and the answer comes back as real experiments: the brand, the page, the before and after, and what each brand did next. If nobody has tested an idea, you are told that too.
You can also browse the record in the ABSee app, by page and by brand, and keep the experiments you want in your library.
ABSee and DoWhatWorks
DoWhatWorks tracks live experiments on brands' websites and predicts which version of a test is likely to win. ABSee takes a different route: it does not forecast a winner, it shows what the brand actually did next. That is the brand's own decision, made with its own data, and it is recorded for every one of the 10,000+ experiments.
ABSee is also built for ecommerce teams only, and it lives inside your AI assistant rather than in another dashboard to check. If your question is "has anyone tried this, and did they keep it?", that is the question ABSee answers.
When something else fits better
If you need UX research guidelines from usability testing, Baymard Institute publishes them. If you want a library of named UI patterns, GoodUI collects them.
Questions
What is DoWhatWorks?
DoWhatWorks is a service that tracks live experiments on brands' websites and predicts which version of a test is likely to win.
What is the best DoWhatWorks alternative?
For ecommerce teams, ABSee: 10,000+ experiments from 1,150+ brands, each with the page before, the version tested and whether the brand kept it, and you can ask it inside Claude, ChatGPT or Gemini.
How do I try ABSee?
Connect your AI at app.absee.it/start and ask ABSee a question inside Claude, ChatGPT or Gemini, or book a demo on a problem from your own site.
How is ABSee different from DoWhatWorks?
ABSee covers ecommerce only. Each experiment records the page before, the version tested and whether the brand kept it, and it works inside your AI assistant. ABSee does not predict winners; it shows what brands actually did next.
Can I see which version won?
Not the brand's own numbers: no outside tool sees those. ABSee records what the brand did next, whether the tested version became the page, which is the decision the brand made with its own data.
Has anyone rolled back a free-shipping progress bar?
Works in Claude, ChatGPT or Gemini once ABSee is connected.
Connect your AI →