Ecommerce A/B Testing Ideas: 25 Things You Can Test
25 ecommerce A/B test ideas across discovery, category pages, product pages, cart, checkout and offers, each tied to the evidence that justifies it and a metric.
Quick answer
Good ecommerce A/B tests come from evidence, not lists. The 25 ideas below cover discovery, category pages, product pages, cart, checkout and offers. Each is paired with the evidence that would justify running it and the metric to judge it by. Pick ideas that match problems your data already shows, on pages with enough traffic, write a clear hypothesis, set a primary metric and guardrails such as margin and returns before launch, run to the planned sample size in full weeks, and record every result.
How to Use This List
Don't run these in order. Start from your funnel and research: find the stage that loses the most shoppers and the evidence for why, then look for ideas in the matching group. The method behind testing, including sample size and analysis, is covered in ecommerce A/B testing; how to run testing as an ongoing program is in ecommerce experimentation framework. Shopify stores can also use 50 Shopify CRO testing ideas.
Worth noting
These are hypotheses, not proven wins. We don't quote uplift figures because results depend on your store, traffic and implementation.
Discovery: Search, Navigation and Filters (Ideas 1–5)
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 1 | Visible search field in the mobile header instead of an icon | Searchers convert well but mobile search use is low | Revenue per session |
| 2 | Autocomplete with product images and prices | Many searches are refined or abandoned | Search-to-product view rate |
| 3 | Navigation labels rewritten in shoppers' words | Tree tests or recordings show people opening the wrong menus | Category click-through, product views |
| 4 | Filter chips above the grid on mobile instead of a hidden drawer | Filter use on mobile is far below desktop | Product view rate from listings |
| 5 | No-results page with corrections, popular categories and alternatives | High exit rate after zero-result searches | Search exit rate |
Category Pages (Ideas 6–8)
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 6 | Product cards showing a deciding attribute (sizes in stock, key spec, colour count) | Shoppers bounce between listing and product pages | Add-to-cart rate from listing sessions |
| 7 | Default sort changed from newest to best-selling or relevance | Low click-through from the first rows | List click-through |
| 8 | Two products per row on mobile instead of one | Shoppers scroll deep without clicking | List click-through, product views |
Product Pages (Ideas 9–14)
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 9 | Estimated delivery date next to the add-to-cart button | Visits to the shipping page followed by exits | Add-to-cart rate |
| 10 | Rating and review count near the product name and price | Shoppers scroll to reviews before deciding | Add-to-cart rate |
| 11 | Gallery leading with the product in use or at scale | Low image engagement; questions about size or look | Add-to-cart rate |
| 12 | Fit or sizing guidance shown inline next to the size selector | Size guide opens followed by exits; fit-related returns | Add-to-cart rate (guardrail: returns) |
| 13 | Sticky add-to-cart bar on mobile | Long mobile pages with low mobile add-to-cart | Mobile add-to-cart rate |
| 14 | Short benefit summary above the fold | Recordings show scrolling without interaction | Add-to-cart rate |
Not sure which tests your evidence supports?
ZSpace builds prioritized test roadmaps from your funnel data, recordings and user research.
Cart (Ideas 15–18)
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 15 | Progress toward a free-delivery threshold | Many orders sit just below the threshold | Average order value (guardrail: conversion, margin) |
| 16 | Delivery cost estimate shown in the cart | Cart visits followed by exits or policy-page views | Cart-to-checkout rate |
| 17 | Express wallets shown in the cart | Mobile shoppers start checkout at a low rate | Checkout starts, completed orders |
| 18 | Cross-sells moved below the checkout button or removed | Recordings show cart distraction; low cart-to-checkout | Cart-to-checkout rate (guardrail: AOV) |
Checkout (Ideas 19–22)
What you can test here depends on your platform. Hosted checkouts restrict changes; on Shopify, customizing the information, shipping and payment steps with checkout UI extensions requires Plus. See Shopify checkout optimization.
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 19 | Guest checkout made the most prominent option | Drop-off at the account step | Checkout completion |
| 20 | Address autocomplete | Frequent address errors or long form times | Checkout completion |
| 21 | Payment methods reordered or local methods added per market | Payment-step exits concentrated in certain markets | Payment step completion |
| 22 | Returns and security reassurance near the pay button | Exits at payment; survey mentions of trust | Checkout completion |
Offers and Trust (Ideas 23–25)
| # | Test | Run it when evidence shows | Primary metric |
|---|---|---|---|
| 23 | Returns window and process stated plainly on product pages | Returns questions in support and surveys | Add-to-cart rate (guardrail: returns) |
| 24 | Bundle or multi-pack offered as the default choice | Customers often buy several units separately | Revenue per session (guardrail: margin) |
| 25 | Risk-reducer such as a guarantee instead of a first-order discount | Discount-acquired customers rarely return | Conversion and margin per new customer |
Write Each Test as a Hypothesis
Use one template for every test: because we observed [evidence], we believe [change] for [audience] will cause [effect on behavior], which we'll measure with [primary metric], while watching [guardrails]. It forces a link between evidence and change and makes results easier to learn from.
Choose Guardrails Before You Launch
- Gross margin or discount rate for any pricing or offer test
- Returns rate for sizing, imagery and expectation-setting tests
- Average order value for cart and cross-sell tests
- Page speed for tests that add scripts or media
- Error and payment failure rates for checkout tests
When Not to A/B Test
Some changes should just be made: broken functionality, accessibility failures, misleading information and obvious usability bugs. On low-traffic stores, tests may take too long to reach a reliable answer; test larger changes on the highest-traffic pages and use user research for the rest.
Record Every Result
Keep a searchable log of hypothesis, evidence, variants, dates, sample, results with uncertainty, decision and screenshots. Losing and flat tests are as valuable as winners because they stop the team retesting the same idea.
Want tests that teach you something?
Talk to ZSpace about a CRO audit and test roadmap and variant design.
Conclusion
The best test ideas are the ones your evidence already points to. Use this list to find a matching idea, write it as a hypothesis, set guardrails, run it properly and record the result. For the full diagnostic starting point, see the ecommerce CRO audit.
Common questions
The idea tied to the biggest, best-evidenced problem on a high-traffic page. Look at where the funnel drops most and what recordings, surveys and tests say about why, then pick the test that addresses it.