37 A/B Test Ideas Worth Actually Running
A practical list of A/B test ideas across headlines, pricing, forms and trust signals, grouped by where they tend to pay off and how to prioritize them.
The best A/B test ideas come from a specific friction point in your own funnel data, not a generic list — but a good list still speeds up the move from "we should test something" to a specific, testable hypothesis. Below are ideas grouped by where they tend to pay off.
Before running any of these, make sure you understand what A/B testing actually requires to produce a trustworthy result — traffic, a clear hypothesis, and patience — because the best idea on this list still fails if it's run without those three things in place.
On this page
How do you generate ideas that are actually worth running?
Start from where the data already points, not from a brainstorm. Pull your funnel report, find the step with the steepest percentage drop, and generate ideas specific to that step rather than general ideas for the whole site. A test idea that isn't tied to an observed friction point is a guess wearing a hypothesis's clothes.
Session replay and support tickets are the two most underused idea sources. A recording of a real visitor hesitating at a specific field, or a support ticket asking a question your page should have already answered, hands you a test idea with the hypothesis already half-written. Most testing tools and CRO tools worth using bundle one or both of these alongside the actual split-testing engine, precisely because ideas and execution live better in the same workflow.
Competitor pages are a weaker but still useful source, with one caveat: a competitor showing a pattern doesn't mean the pattern works for them, only that they haven't removed it yet. Treat a competitor's page as a source of ideas to test on your own traffic, not as evidence of what already works — the only evidence that counts is a result from your own funnel.
What should you test on your headline and hero copy?
Specificity against vagueness, almost every time. Replace an adjective-heavy headline with one naming the exact outcome or number a buyer cares about, and test it against the original rather than assuming the specific version wins — it usually does, but "usually" is not "always," and the exceptions are where you learn something real about your own audience.
17.5% lift
Other headline ideas worth queuing: naming the buyer's role or industry directly instead of writing generically for everyone; leading with the outcome instead of the feature that produces it; and testing a headline that states a number — a timeframe, a price, a percentage — against one that doesn't. Numbers tend to read as more credible than adjectives, because they're falsifiable in a way "powerful" and "seamless-sounding" copy isn't.
What pricing and layout tests tend to win?
Showing a number where one was hidden, first. If your pricing page currently routes every visitor through a form, testing a visible starting price against the current form-gated version is one of the highest-leverage tests most sites haven't run, and it ties directly into broader pricing page work if the result is positive.
| Category | Typical effort | Typical payoff |
|---|---|---|
| Headline / value prop | Low — copy only | Medium to high |
| Pricing display / tiers | Low to medium | High, but slower to read |
| Form fields | Low — remove, don't add | Medium, compounding |
| Social proof placement | Low | Low to medium |
| Full page redesign | High | Unpredictable, hard to attribute |
Layout ideas beyond pricing: testing the order of sections on a long page — proof before pricing versus pricing before proof — and testing whether a sticky call-to-action bar on mobile outperforms one that scrolls away. Both are cheap to build and both address a real, observable behavior rather than a stylistic preference.
What form and checkout tests are worth trying?
Removing fields before you redesign anything. Every optional field you can defer to after signup or after the first purchase is a field that isn't costing you conversions today. Test a shorter form against your current one before you spend time on visual polish — the field count usually matters more than how the form looks.
Other checkout-specific ideas: testing a guest checkout option against an account-required flow, testing whether showing a progress indicator reduces abandonment on a multi-step form, and testing the exact wording of an error message — a message that tells someone what to fix converts differently than one that just says something's wrong.
One more worth queuing early: testing the order fields appear in, not just how many there are. A field asking for payment details before shipping information reads as more committal than the reverse order, even when both versions collect exactly the same information by the end. Sequence changes cost almost nothing to build and are easy to overlook next to bigger ideas.
What social proof and trust tests move the needle?
Placement more often than content. The same testimonial or logo bar can underperform in one position and overperform in another, because trust signals work by reducing hesitation at the exact moment it occurs — right before a commitment, not decoratively somewhere on the page. Test moving proof next to the action itself before you test which proof to show.
Specific ideas: a named, attributed testimonial against an anonymous quote; a real usage or customer count against no count at all if you can state one honestly; and a trust badge or guarantee statement placed directly next to the price rather than in a footer.
Do not manufacture a proof point you don't have. A fabricated count or a testimonial attributed to no one specific is a test that can technically "win" while quietly damaging trust in a way your conversion number won't show you until a customer calls it out later. If you don't have real proof to test yet, that's a finding too — it means collecting real proof is the higher-priority project, ahead of testing how to display proof you don't have.
How do you prioritize a long list of ideas?
Score expected impact
How much traffic sits behind the page, and how directly the idea addresses an observed drop-off.
Score confidence
Is this idea backed by data — a session recording, a support ticket pattern — or is it a hunch?
Score effort
Copy and config changes score higher than anything requiring new development.
Multiply, don't add
A high-impact, low-effort idea should always run before a similarly-scored idea that needs a full sprint to build.
How do you know an idea was actually worth testing, after the fact?
Look at more than whether it won. A test that wins by a small margin on a low-traffic page taught you less than a test that lost clearly on a high-traffic one — the second result rules something out with confidence, which is real information even though it feels like a loss. Reading real results honestly, including the losses, is what keeps a queue of ideas actually improving over time instead of just cycling through the same handful of safe bets.
Questions people also ask
Where should a new testing program start?
Wherever the funnel loses the highest percentage of visitors on your highest-traffic path — usually the page right before checkout, signup, or a form submit. That page has the most attempts and the most room to improve, which gets you a readable result fastest.
How many test ideas should be queued at once?
Enough for four to six weeks of testing, not a year's backlog. A shorter queue forces you to re-prioritize based on what you just learned instead of running down a list written before you had any real data.
Do small changes like button color actually matter?
Rarely on their own. A button color test wins occasionally, but it's a weak use of traffic compared to testing what the button says or what happens after it's clicked. Save the small stuff for once the bigger ideas are exhausted.
Should every idea on this list get tested?
No. Treat it as a menu to pull from based on your own funnel data, not a checklist to complete in order. An idea that doesn't match a real friction point on your site is a wasted test slot.
Sources
- How We Doubled Conversions With A/B Testing — Semrush (checked 22 September 2026)
- A/B Testing 101 — Nielsen Norman Group (checked 22 September 2026)
- What is A/B Testing? A Practical Guide With Examples — Wingify (VWO) (checked 22 September 2026)
- Evan's Awesome A/B Tools — Evan Miller (checked 22 September 2026)
Get a test queue built from your own funnel, not a generic list
Book a free 30-minute consultation with Ibrahem and Jeremy. We look at your site while we talk, and you leave with a prioritised plan.
Who wrote this
Jeremy ZavaCo-founder
Leads growth and client partnerships.
Works on: CRO, SEO
Reviewed before publishing by Ibrahem Helal, Co-founder.
- A/B Testing: The Complete GuideWhat A/B testing is, when your traffic can support it, how to run a split test that holds up, and how to read the result without fooling yourself.
- Best A/B Testing Tools, ComparedA working list of A/B testing tools by team size and budget, with what each one actually changes about how you test and what it costs to run.
- Best CRO Tools for a Real Testing StackThe CRO tools that actually matter, grouped by what each one is for — testing, session replay, feedback and analytics — with honest cost ranges.
- How to Read and Report CRO ResultsCRO results mean nothing without the hypothesis, the sample size and the confidence level attached — here is how to read them honestly and report them well.