How do I test two versions of a landing page to see which gets more leads?

This page may contain affiliate links from which Blur Studio earns commissions.

Testing two versions of a page against each other, rather than guessing which headline or layout works better, is one of the more reliable ways to actually improve a landing page's conversion rate instead of redesigning based on opinion. If your site is built and hosted on Netlify, a platform that builds your site from its connected code and serves it to visitors, the specific split-testing tool built into that platform for this job is being phased out, so it's worth knowing that before you plan around it. My wider view of the platform is in a Netlify review for small business and marketing sites.

Netlify's split testing is closed to new projects

Netlify's own documentation states that, as of August 26, 2026, split testing can no longer be added to new projects. It had stayed in beta the whole time it was available, and sites that already had it configured before that date keep working, but it isn't something you can newly set up now. On top of that, it only ever supported one test at a time per site, and it's explicitly documented as breaking when combined with Functions, Edge Functions or Agent Runners, which limits how much of a modern Netlify site it could actually sit alongside even while it was available.

What to use instead

A dedicated A/B testing tool like VWO. These tools are built specifically for this job: splitting traffic between variants, tracking which one converts better, and reporting the result with statistical confidence rather than a raw guess. They work independently of your hosting platform, so you're not tied to whatever a host happens to support.

A Google Optimize replacement. Google retired Optimize itself a while back, and several tools have positioned themselves as direct replacements, generally offering similar visual-editor-based testing without needing to touch code. Worth comparing a couple directly against your actual use case before committing, since "Google Optimize replacement" covers tools with meaningfully different depth and pricing.

The free, manual version: run A for a month, then B for a month. This isn't a true simultaneous split test, since time-based factors (seasonality, a marketing push, a news event) can confound the comparison between the two periods. But for a small site without the traffic to justify a dedicated testing tool, it's a genuinely workable, zero-cost way to get directional evidence, as long as you're honest with yourself that it's a weaker signal than a real simultaneous split.

The traffic problem nobody mentions

A/B testing needs enough visitors and enough conversions on each variant for the result to be statistically meaningful, not just a difference that looks real but is actually noise. If your landing page gets a few dozen visits a month, any A/B test, on any tool, will likely produce a result that looks decisive but isn't reliable, since the sample size is simply too small to distinguish a real effect from random variation. Before investing in a testing tool, do a rough gut check: do you have at least a few hundred visits a month to the page you want to test, with a reasonable number of actual conversions on each variant? If not, your energy is better spent on getting more traffic first, or making an obviously better change based on good judgment rather than a test that can't actually tell you anything reliable. A cheaper first step is testing a new offer with a one-page website.

A worked example: a service business testing a headline manually

A home services company wanted to know whether leading with price ("Free estimates, same-day service") or leading with trust ("Licensed, insured, 200+ five-star reviews") converted better on their main landing page. Their traffic, roughly 600 visits a month to that page, wasn't quite enough to justify a dedicated paid testing tool's cost for a single test, but was enough to get a reasonably useful read from the simple sequential method: run the price-led headline for one full month, then the trust-led headline for the next full month, tracking the same conversion action (a quote request form submission) both times.

The price-led version converted noticeably better, but the business was careful to check for an obvious confound before trusting the result fully: the trust-led month happened to overlap with a seasonal dip in search volume for their service, which on its own could have depressed that month's numbers regardless of headline. Rather than treating the first result as final, they ran the trust-led headline again the following month, outside the seasonal dip, and got a result closer to parity between the two, concluding correctly that the first comparison had been partly confounded by timing rather than purely reflecting headline performance. That second check is exactly the discipline the sequential method requires and a true simultaneous split test wouldn't have needed.

A practical note on picking what to test first

Not every element on a page is worth testing, and spreading limited traffic thin across several small variations at once just slows down getting a useful answer on any of them. Pick the single element most likely to actually move the needle, usually the headline, the primary call-to-action, or whatever's directly above the form, rather than testing a minor color change or a small wording tweak first. A bigger, more consequential variation also tends to produce a clearer, faster-to-detect difference than a subtle one, which matters more when your traffic is already limited. Run that one test to a clear conclusion before moving to the next element, rather than changing several things loosely at once and ending up unable to say which change actually drove the result. Discipline here matters more than the tool you pick, since even the most sophisticated testing platform can't rescue a test design that changed too many things at once to draw a clear conclusion from. If you still need the page itself, getting a one-page landing page live today covers options and costs.

What I'd actually tell you

If you're on Netlify and were hoping to use its own split testing, stop planning around it; it's not available for new setups and the honest move is to pick a dedicated tool instead. If your traffic is genuinely too low for a meaningful test, I'd skip testing altogether for now and instead make the change you believe is better, based on good copywriting and design judgment, revisiting formal A/B testing once your traffic justifies it. Start by working out whether your website brings you any leads at all.

How I'd approach this

When a client wants to test landing page variants and their traffic genuinely supports it, I'll wire in a dedicated testing tool rather than rely on anything bundled into hosting, specifically because that keeps the testing capability independent of hosting decisions down the line. When traffic doesn't support a real test, I'll say so plainly rather than set up tooling that will just produce a misleadingly confident but statistically meaningless result, and I'll build the site itself to make future testing easy to add once the traffic is there.