Funnel vs A/B Testing: Which One Sells More From Your Traffic?

A/B testing and a funnel in front of your page both promise more sales from the same traffic, but they answer different questions. An A/B test finds the one version of your page that works best for the average visitor. A funnel gives different visitors different answers. Which one you need depends on your traffic and on how many kinds of buyer you have. This is part of our guide to increasing your conversion rate without more traffic.

What each one actually does

An A/B test splits your visitors at random between two versions of a page, say two headlines, and counts which version sells more. When the test ends, everyone gets the winner. The page is still one page for everyone; it's just a better one.

A funnel in front of the page asks each visitor a few questions before the offer. Their answers decide what they see next: a result written for their situation and a button to the offer that fits them. A beginner and an expert, or a small buyer and a big one, end up in different places.

Picture a hypothetical yoga studio that sells drop-in classes, a beginner course and a teacher training. An A/B test can tell the studio which headline sells more classes overall. A funnel can send the nervous beginner to the beginner course and the experienced student to the teacher training, each with a reason that speaks to them.

An A/B test improves the page for the average visitor; a funnel stops treating visitors as average.

When A/B testing wins

A/B testing is the right tool when three things are true.

• You have the traffic. A test needs enough visitors to tell a real difference from luck. To detect a page moving from 2% to 2.5% at the usual 95% confidence and 80% power, you need about 13,800 visitors on each version (we show the math in conversion rate optimization with low traffic). Sites with tens of thousands of visitors a month can run tests like this every few weeks.
• You have one specific question. "Does a headline about the result beat a headline about the method?" is a good test. "Is our page good?" is not.
• Your visitors want broadly the same thing. If most people arriving are the same kind of buyer looking at the same offer, there's one best page to find, and testing finds it.

When those hold, testing is hard to beat. It replaces opinions with evidence, and it catches changes that seemed smart but hurt sales.

Test when you have the traffic, a sharp question, and visitors who mostly want the same thing.

When a funnel wins

A funnel earns its place when the problem isn't the page's wording but the fact that one page has to serve very different people.

• You sell several offers. Three plans, five courses, a product range. The visitor first has to work out which one is theirs, and a page can only lay the options side by side. Questions can do the choosing with them.
• You have several kinds of buyer. Beginners and experts, individuals and companies, people with a $50 budget and people with $5,000. Whatever the page says to one group is wrong for another. An A/B test can only pick which group to favour.
• Many visitors aren't ready yet. They're interested but unsure it's for them. A result that answers their specific doubt does what a general page can't.
• The page itself is already clear. A funnel can't fix an unclear offer or a lack of proof. If the page fails those, fix it first; see why your sales page isn't converting.

There's published evidence that adapting the page to the visitor can help. eBay ran a controlled A/B test of exactly this: pages adapted to each shopper against the same page for everyone, and the adapted version raised purchases by 7.34% (Lo et al., RecSys 2021). That's eBay's traffic and eBay's catalog, so read it as evidence the idea can work, not as a number you should expect.

Use a funnel when the page's real problem is that it has to say one thing to people who need different things.

The honest catch with a funnel: proving it worked

A funnel doesn't need a statistically significant test to start working, because it's a change to how the page talks to people, not an experiment. But you still want to know whether it helped, and that needs data too.

On a small site, the practical way is a careful before-and-after comparison: four full weeks before the funnel against four full weeks after, same traffic sources, nothing else changed. On a larger site, you can run a real A/B test of the funnel against the plain page, which is the cleanest proof.

It also helps to measure the step before the sale. Snacked, which builds funnels from your sales page's URL, counts clicks to each of your offers out of everyone who opened the funnel, on every plan. With a short snippet on your thank-you page it counts the sales too. It does not split your traffic for you or run A/B tests, so the comparison is yours to set up.

A funnel changes the experience right away, but proving it worked still takes a fair comparison.

They aren't rivals: using both

The best setups use both, each for what it's good at.

1. Test the funnel itself. If you have the traffic, A/B test the funnel against the direct page. It's the single most useful test you can run on a page with several offers.
2. Test inside the funnel. Once the funnel works, the same rules apply to its parts: the first question, the result text, the button label. Big changes first.
3. Test the page behind it. The offer page still has to convince the visitor once they arrive. Headline and proof tests on that page still pay off.
4. Keep the funnel short. Every extra question is a chance to lose someone. If you test anything, test fewer questions against more.

For a small site without the traffic to test, the order is simpler: fix the page, add the funnel if you sell to different kinds of buyer, and compare full weeks before and after.

Use a funnel to send each visitor to the right offer, and A/B tests to improve each step once you have the traffic.

Quick comparison

Recap

• An A/B test finds the best single page for everyone; a funnel gives different visitors different paths to the offer that fits them.
• Test when you have high traffic, one clear question and visitors who want the same thing. Use a funnel when you sell several offers or to several kinds of buyer.
• They combine well: test the funnel against the page when traffic allows, and compare full weeks before and after when it doesn't.

For the full picture of what moves your conversion rate, read our guide to increasing your conversion rate without more traffic. To see whether your page needs fixing before either, run it through the free sales page grader.