A/B test ideas for landing pages (that are worth testing)
George Peric · Conversion designer · 6 min readShort answer
The landing page A/B tests most worth running change what visitors decide on: the headline’s match with the ad, the first action, proof near the button, the offer, form length, the call to action and the mobile layout. Prioritize ideas by potential, traffic and effort, and skip button colors and tiny copy tweaks unless you have a lot of traffic.
Key takeaways
- Test changes a visitor would notice and that answer a real doubt, not cosmetic details.
- Start with the first screen of your highest-spend page: headline, first action and proof.
- Rank ideas with a simple score (potential, importance, ease) and test the top one first.
- Judge every test on revenue per visitor, with quality as a guardrail.
What makes a landing page A/B test worth running?
A good test changes something the visitor actually notices and uses to decide, and it comes from evidence, not a hunch. Evidence can be analytics (a big drop on the first screen), recordings (people hunting for the price), surveys (“I wasn’t sure it would work for me”) or a clear mismatch with the ad. Each idea below comes with the reason it might work. If that reason doesn’t apply to your page, skip it.
What are good A/B test ideas for landing pages?
Fifteen ideas, grouped by area of the page.
Headline and message match
- Repeat the ad’s promise in the headline. Why it might work: visitors spend their first seconds checking whether they are in the right place. Reusing the ad’s words answers that at once. See message match.
- Show the same visual as the ad. Why it might work: a familiar image confirms the click before anyone reads a word.
- Build one page per ad angle instead of one page for all ads. Why it might work: a page written for “price” buyers and a page for “speed” buyers can each be more specific than one page trying to serve both.
First action
- Replace the first button with the first question. Why it might work: on lead and quiz funnels, a one-tap answer feels easier than a button that leads to a form, and it starts the commitment.
- Make the first action continue the ad’s question. Why it might work: if the ad asked “Cooking for one, two or a family?”, letting people answer that exact question feels like the next step.
Proof
- Move the rating and review count next to the call to action. Why it might work: doubt peaks at the moment of clicking, so proof helps most where the decision happens.
- Use specific reviews that answer the main objection. Why it might work: a review that says “I didn’t feel groggy in the morning” does more than five generic stars if grogginess is the main fear.
Offer
- State the price and what is included on the landing page. Why it might work: hidden prices make people click away to find them, and some never come back.
- Test a different offer framing: a bundle, a guarantee or a trial. Why it might work: the offer often matters more than the design. A clear guarantee lowers the perceived risk of trying.
Form
- Remove fields you don’t need. Why it might work: every field is work. Nielsen Norman Group recommends cutting unnecessary questions and keeping optional fields to a minimum.
- Split a long form into short steps. Why it might work: easy questions first build momentum before the sensitive ones. See multi-step vs single-step forms.
Call to action
- Rewrite the button to name the outcome. Why it might work: “Get my estimate” tells people what they get, “Submit” tells them what they give up.
- Add reassurance next to the button. Why it might work: a short line such as “No obligation” or “Free returns” answers the last doubt at the moment of clicking.
Mobile
- Add a sticky call to action on mobile. Why it might work: on a long phone page, the button scrolls away right when someone is convinced.
- Shrink the hero image so the first action fits on the first phone screen. Why it might work: a big image can push the button below the fold, and people may not scroll to find it.
How do these ideas compare on effort and potential?
Use this as a starting point, then adjust it to your page and your data. “Potential” is a rough guess of how much the change could move results on a typical paid landing page, not a promise.
| Idea | Area | Effort | Potential |
|---|---|---|---|
| Repeat the ad’s promise in the headline | Message match | Low | High |
| Show the same visual as the ad | Message match | Low | Medium |
| One page per ad angle | Message match | High | High |
| First question instead of first button | First action | Medium | High |
| First action continues the ad’s question | First action | Medium | High |
| Rating and review count next to the button | Proof | Low | Medium |
| Reviews that answer the main objection | Proof | Low | Medium |
| Price and what is included on the page | Offer | Low | Medium |
| Bundle, guarantee or trial framing | Offer | Medium | High |
| Remove unneeded form fields | Form | Low | Medium |
| Split a long form into steps | Form | High | High |
| Button names the outcome | Call to action | Low | Low |
| Reassurance next to the button | Call to action | Low | Low |
| Sticky call to action on mobile | Mobile | Low | Medium |
| Smaller hero so the action fits on screen | Mobile | Low | Medium |
How do you prioritize A/B test ideas?
Score each idea and test the highest score first, starting on the page with the most traffic and spend. I use PIE: Potential (how much it could improve), Importance (how much traffic and money goes through that page) and Ease (how hard it is to build and test). Score each from 1 to 10 and average them. ICE works the same way with Confidence instead of Importance, which rewards ideas backed by evidence.
A few rules of thumb on top of the score:
- Fix bugs before you test. A broken form or double-firing tracking is a fix, not a test.
- Go top-down. The first screen is seen by everyone, so tests there reach the most visitors.
- Bigger changes on low traffic. Small effects need a lot of traffic to detect. With little traffic, test bold changes.
- One question per test. If you change the headline and the form at once, a win won’t tell you which one worked.
Example
Say a skincare brand’s audit shows that most mobile visitors leave on the first screen, and the headline says “Welcome! Browse the collection” while the ad promised “Clear skin in 30 days”. The headline idea scores high on potential (it fixes the biggest leak), importance (it is the top-spend page) and ease (it is a copy change). The multi-step form idea scores lower on ease and matters less, because few people reach the form yet. So the headline goes first.
What should you not A/B test?
Don’t spend traffic on changes too small to notice or too small to detect.
- Button colors. As long as the button stands out and is readable, the color is rarely what stops people. Contrast and placement matter more than the hue.
- Tiny copy tweaks on low traffic. Swapping one word in a subhead may have a real but small effect, and on a page with modest traffic the test can run for months without a clear answer.
- Things that are just broken. Slow pages, failing forms and wrong tracking should be fixed, not tested.
- Several unrelated changes at once, unless you are deliberately testing a full redesign against the old page.
- Ideas with no reason behind them. “Let’s see what happens” tests rarely teach anything you can use on the next page.
How should you measure a landing page test?
Use revenue per visitor as the primary metric, quality as a guardrail, and run at least two full weeks. Clicks and form starts can rise while revenue falls, for example when an easier first step brings in less serious leads. Revenue per visitor captures both volume and value. To turn a win into yearly money, see how to calculate the revenue impact of an A/B test.
For a worked example of several of these ideas applied to one page, see my landing page redesign case study. If you want a testing program run on your pages, see A/B testing or book a call.
FAQ
What is the best first A/B test for a landing page?
Usually the headline or the first action on the first screen of your highest-spend page, because every visitor sees it. If the headline doesn’t repeat the ad’s promise, start there.
How many A/B tests should I run at once?
On one page, one at a time, so the results don’t mix. You can run tests on separate pages or separate traffic at the same time.
Are button color tests worth it?
Rarely. If the button is easy to see and read, its color is unlikely to be what holds visitors back. Testing the button’s wording, placement or the proof next to it is usually a better use of traffic.
How much traffic do I need to A/B test a landing page?
Enough conversions in each version to detect the difference you care about. With low traffic, test bigger changes and expect longer tests. A sample size calculator will tell you how long a given test needs.
Sources
Conversion designer, CXL certified in Conversion Optimization and Ecommerce Marketing. I design ads, landing pages and funnels and A/B test them.