# How to test landing pages for ads

> Split ad traffic evenly between two page versions, change one thing, and wait for enough visitors. Three ways to split, a sample size example and costs in rupees.
- **Author**: Marxx
- **Published**: 2026-11-23
- **Category**: Testing
- **URL**: https://marxx.ai/library-guides/how-to-test-landing-pages-for-ads

---

To test landing pages for ads, send equal, random shares of the same ad traffic to two versions of the page, change only one thing, and compare orders or leads per visitor. Wait until each version has enough visitors before you pick a winner.

## Three ways to split traffic

| Way | How it works | Catch |
|---|---|---|
| Split-test tool | A tool sends each visitor to page A or B at random | Needs a tool. Google Optimize ended on 30 September 2023 |
| Two ads, two pages | Each ad points to a different page address | Meta may deliver the ads to different people, so you mix ad delivery with page effect |
| One page, then the other | Run page A one week, page B the next | Weakest. Days, offers and moods differ week to week |

Google says Optimize is no longer available, and names AB Tasty, Optimizely and VWO as A/B testing providers it is collaborating with on Google Analytics integrations ([Google Analytics Help](https://support.google.com/analytics/answer/12979939), checked 2026-10-20).

Meta's A/B test variables are creative, audience, placements and custom ([Meta](https://www.facebook.com/business/help/1597318281091985), checked 2026-10-20). None is a page test by name. Meta also warns that informal testing, such as switching ad sets on and off, can overlap audiences.

## Step by step

1. **Pick one change.** A new headline, a shorter form or a different offer. Not all three.
2. **Check [message match](/glossary/message-match).** The page must repeat the promise in the ad.
3. **Split traffic evenly** using one of the ways above.
4. **Tag each page's traffic** so you can compare it in GA4. See [how to set up UTM parameters for Meta ads](/guides/how-to-set-up-utm-parameters-for-meta-ads).
5. **Decide the finish line first.** For example: "10,000 visitors on each page."
6. **Compare conversion rate,** not clicks. See [conversion rate](/glossary/conversion-rate).
7. **Keep the winner and test the next change.**

## A worked example

Say page A turns 2.0% of ad visitors into orders. A new page B looks better. After 1,000 visitors each, A has 20 orders and B has 26 (2.6%). The numbers are made up.

That gap is small enough to be luck. A standard sample size formula, at 95% confidence and 80% power, says you need about 9,800 visitors on each page to trust a 2.0% against 2.6% difference.

At INR 12 a click, 19,600 visitors cost INR 2,35,200. That is why small stores should:

- Test big changes, such as the offer or the whole layout, which show up fast.
- Test a nearer step, such as add to cart, which happens more often than a purchase.
- Skip tiny tweaks, like a button colour.

## Mistakes to avoid

- Stopping after one good day
- Changing two things at once
- Sending page A your ad traffic and page B your email traffic
- Ignoring [landing page view rate](/glossary/landing-page-view-rate). If many clicks never load the page, fix speed first

More ideas for what to change are in [how to improve conversion rate from ads](/guides/how-to-improve-conversion-rate-from-ads). A starting layout is in the [landing page template for ads](/templates/landing-page-template-for-ads).

## FAQ

### How long should a landing page test run?

Until each version reaches the visitor count you set before you began. Run it for at least a full week so weekdays and weekends both count.

### Can I test a landing page with Meta's A/B test?

Meta's A/B test variables are creative, audience, placements and custom, and none is a page test by name. For pages, use a split tool, or accept that two ads with two page addresses mix the ad effect with the page effect.

### What should I test first?

The offer and the headline. They change what people see first and what they are asked to do.

---
- [More Testing articles](https://marxx.ai/library-guides/category/testing)
- [All articles](https://marxx.ai/library-guides)