Before
Tested
MonclerProduct pageProduct page size-prompt button
What changed Relabels the purchase button to prompt size selection first.
Seen 3 Aug to 18 Sep 2026
Outcome being confirmed
ABSee / Database
DatabaseA database of real ecommerce A/B tests. What brands tested, and what they kept.
Yes — ABSee keeps a record of 10,000+ real experiments run by 1,150+ ecommerce brands: the page before, the version tested, when it ran and whether it was kept.
Updated October 2026 · Source: ABSee record · Methodology →
BeforeTestedMonclerProduct pageHow it works
Watch it work. About 25 seconds.
1 Connect
MCP serverhttps://mcp.absee.it/mcpCopied- Claude
- ChatGPT
- Gemini
Paste it into your assistant’s connector settings, then sign in to ABSee.
Add one link to the assistant you already use.
2 Ask
Has anyone tested warning shoppers that items in their bag aren't reserved?ABSee · search_experiments — mini-cart 'not reserved' warningSearching…Ask anythingAsk in plain words.
3 See the experiment
Has anyone tested warning shoppers that items in their bag aren't reserved?Yes, one close match: Timberland, in its mini-cart.
Before
TestedTimberlandMini-cartMini-cart 'not reserved' warningWhat changed Adds a line warning that items in the bag are not reserved.
Likely hypothesis If the mini-cart warns that items aren't reserved, then more shoppers will go on to checkout, because they will want to secure what they picked.
Strategic takeaway Explicitly stating that carted items are not reserved leverages the fear of missing out, prompting faster checkout decisions.
Seen 19 May to 4 Oct 2026
ShippedReply…A real experiment: what changed, why, and what the brand did next.
4 Weigh your idea
If we remove secondary content from the product page,then more shoppers will add to bag,because there is less to distract them.
ABSee · evidence_for_hypothesis — secondary content on the product page✓Comparable experimentsIntelligence planEvidence for
ReebokShipped'Best For' section removed
Evidence against
ZimmermannRolled backProduct page sizing tool removal
Real experiments. Real outcomes. Inside Claude, ChatGPT or Gemini.
State it as if / then / because — and see the evidence for and against.
An illustration of the format, built from public experiments. Your assistant returns the comparable experiments ABSee finds for your own hypothesis.
- Experiments
- 10,000+
- As of October 2026
- Ecommerce brands
- 1,150+
- As of October 2026
Updated daily. New experiments are added as they are seen, and running experiments get their outcome once the brand has decided.
What is in each recordBeforeTested
TimberlandMini-cartMini-cart 'not reserved' warning
One experiment, everything you get. The same for every record.
ABSee is experimentation intelligence for ecommerce: a record of real experiments run by 1,150+ brands, with the page before, the version tested, and whether it was kept. An experiment here means two versions of the same public page served to ordinary visitors at the same time. A page that simply changes for everyone is not counted.
BeforeTestedTimberlandMini-cart- What changed
- A shopper expanding the mini-cart sees a newly added warning message stating that items in their bag are not reserved. This explicit nudge encourages them to check out immediately to ensure they do not miss out on the items.
- Likely hypothesis
- If Timberland warns in the mini-cart that items in the bag are not reserved, then more shoppers will go on to checkout, because the fear of losing what they picked makes securing it feel urgent.
- Strategic takeaway Intelligence plan
- Explicitly stating that carted items are not reserved leverages the fear of missing out, prompting faster checkout decisions.
- Principle Intelligence plan
- Loss Aversion
Seen 19 May to 4 Oct 2026
ShippedIntelligence plan
- Brand and page. The brand, and the page type the experiment ran on.
- The page before, as a visitor saw it.
- The version tested, as a visitor saw it, with the change framed in orange.
- What changed between the two versions, in plain words.
- Likely hypothesis. ABSee’s reading of the idea behind the test, written as if, then, because. It is a reading, not the brand’s own words.
- Strategic takeaway and principle. The lesson to carry to your own pages, and the UX principle it rests on. Part of the Intelligence plan.
- Dates seen. When the experiment was first seen and when it ended. Each record carries its own dates, so you can tell a test from last week from one that ended last year.
- What the brand did next. If the tested version became the page, it was kept (Shipped). If the original stayed, it was rolled back. That is the brand's decision, read from the public page, not a measured lift: ABSee never sees a brand's own numbers. Part of the Intelligence plan. How this works, and its limits, is on the methodology page.
Pages covered
The shopper's whole path. Mobile and desktop recorded separately.
The record covers the home page, category pages, product pages, the cart, checkout and search, plus changes that run across the whole site, such as headers, navigation and banners. Mobile and desktop versions are recorded separately, because brands often test them separately.
Home
Home page experiments, including headers, navigation and banners.
Category
Category and listing page experiments.
Product
Buy box, imagery, price display, size and fit, urgency, social proof.
Cart
Cross-sell, free-shipping bars and reassurance.
Checkout
Reassurance, express pay and coupon fields.
Search
Search results page experiments.
Logged-in areas are never visited, and prices personalised to an account are not recorded. Everything in the record is something any visitor could have been served. Browse A/B test examples with the before and after.
From the record
Before
Tested
Before
Tested
AéropostaleMini-cart
Before
Tested
BalenciagaProduct page
Three more real experiments. Before, tested, and what happened next.
The chip shows what the brand did next: Shipped means the tested version became the page; Rolled back means the original stayed.
BeforeTestedKaren MillenProduct page
Green Add to Bag button
What changed Turns the Add to Bag button from black to green.
Seen 27 Aug to 27 Aug 2026
Rolled back
BeforeTestedAéropostaleMini-cartMini-cart free shipping meter addition
What changed Adds a progress bar showing how much more to spend for free shipping.
Seen 16 May to 13 Jul 2026
Outcome being confirmed
BeforeTestedBalenciagaProduct pageProduct page colour-swatch consolidation
What changed Gathers every colour and material into one row under the product.
Seen 3 Aug to 17 Sep 2026
Outcome being confirmed
Evidence for and against
BeforeTested
BeforeTested
ZimmermannProduct page
Bring a hypothesis. See who tested it, and what they kept.
On the Intelligence plan your assistant files comparable experiments as evidence for, against, or closest match. An illustration of the format, built from public experiments. Your assistant returns the comparable experiments ABSee finds for your own hypothesis.
HypothesisIntelligence plan
If we remove secondary content from the product page,then more shoppers will add to bag,because there is less to distract them.
For
BeforeTestedReebokProduct page
'Best For' section removed
What changed Removes the 'Best For' section from the product details list.
Seen 7 Aug to 4 Oct 2026
Shipped
Against
BeforeTestedZimmermannProduct pageProduct page sizing tool removal
What changed Hides the automated size recommendation tool.
Seen 21 Jul to 5 Aug 2026
Rolled back
What you can ask it
Ask before you spend a test slot. In Claude, ChatGPT or Gemini.
The record is built to answer the questions a product, growth or CRO team asks before it spends a test slot. Answers come back as real experiments, each with its before and after, the brand, the page and the dates. An answer can be empty: if nobody in the record has tested an idea, that is what you are told.
- Find experiments
Who is testing a sticky add-to-bag right now?
Comes back with experiments with that change, each with a before-and-after thumbnail, the brand, the page and the dates.
- What brands did nextIntelligence plan
Has anyone rolled back a free-shipping progress bar?
Comes back with experiments with that idea and what each brand did next: kept or rolled back.
- Find experiments
What's been tested against fit uncertainty on mobile product pages?
Comes back with mobile product-page experiments aimed at size and fit.
- Weigh a hypothesisIntelligence plan
Our hypothesis: if we add a low-stock message on product pages, then more shoppers will buy now instead of later, because scarcity prompts a decision. What is the evidence for and against?
Comes back with comparable experiments filed as evidence for, against, or closest match.
- See what's new
What changed in fashion and footwear this week?
Comes back with new experiments in that category from the past week.
- Check competitors
Which of these competitors does ABSee cover, and what have they tested on checkout?
Comes back with the names ABSee has experiments for, then their checkout experiments.
Use it inside Claude, ChatGPT or Gemini: connect ABSee and ask in your own words, which takes about a minute (how to connect). Or browse experiments by page and by brand in the ABSee app, and keep the ones you want in your library.
How it differs
A record, not a rulebook. Different sources answer different questions.
Pattern libraries such as GoodUI collect test results and turn them into reusable patterns. UX benchmarks such as Baymard publish guidelines from large-scale usability research and expert review. Both are useful, and both answer a different question.
| ABSee | Pattern libraries | UX benchmarks | Case-study posts | |
|---|---|---|---|---|
| Built from | Experiments brands served to visitors on their public pages | Test results, turned into reusable patterns | Large-scale usability research and expert review | One team's write-up of one test |
| What you get | The page before, the version tested, the dates and what the brand did next | Patterns to reuse | Guidelines | The story of one test |
| It answers | Who tested this on a page like mine, and did they keep it? | Which pattern might fit here? | What does good look like, by research? | How did that team run that test? |
ABSee does not rate designs or publish guidelines. Teams often use all three together: a guideline or pattern suggests an idea, and the record shows who tested it on a similar page and whether they kept it.
Open and on a plan
Try it free. Go further on a plan.
Anyone can connect their assistant at app.absee.it/start to try it. Pricing is set per team after a short demo.
- What have others tested?
Library
Search real experiments, see before and after screenshots, browse brands and categories, and see patterns across the market.
- What's the evidence for our hypothesis?
Intelligence
Adds outcomes (kept or rolled back), evidence for and against an idea, the UX principles behind each pattern, competitor tracking and prioritisation.
Questions
Questions
Is there a public database of A/B test results?
ABSee keeps a record of 10,000+ real experiments run by 1,150+ ecommerce brands, recorded from their public pages. Each record has the page before, the version tested and when it ran. Brands do not publish their own numbers, so no outside record can show a measured lift; ABSee shows what each brand did next.
Can I see competitors' experiments?
If ABSee covers them, yes. Ask your assistant which of your named competitors are in the record, then ask what they tested. ABSee does not publish its full brand list, so the check is done one name at a time.
Does it show which version won?
It shows which version the brand kept: whether the tested version became the page or the original stayed. That is the brand's decision, not a measured lift. Outcomes are tracked and are part of the Intelligence plan.
Which brands are covered?
1,150+ ecommerce brands, mostly mid-to-large and weighted to fashion and to Europe. The list itself is not published; you can check any brand by name once your assistant is connected.
Ask your assistant
Who is testing a sticky add-to-bag right now?
Works in Claude, ChatGPT or Gemini once ABSee is connected.
Connect your AI →