To test a landing page headline, don’t write two versions. Write at least five, ideally ten, that make genuinely different promises. Split your traffic between them, cut the clear losers after a few days, keep the best few running until enough people have bought, and pick the winner by revenue per visitor. And if you can, test the headline together with the image and the button in one multivariate test, so you find the best combination instead of the best headline on its own.
The headline is usually the first thing I test on a new funnel page. It’s one line of text, it takes minutes to change, and every visitor reads it. Below is why it deserves to go first, how I write variants that are worth testing, with examples for a course, a webinar and a $27 ebook, and how to read the result without fooling yourself.
Key takeaways
Test the headline first on most pages. It sits on the first screen, where people decide whether to stay, and it costs almost nothing to change.
Start with 5 to 10 versions, not 2. Two versions only tell you which of two guesses was better. Cut the losers early and keep the best running.
Test angles, not words. “Get more clients” vs “Get more customers” teaches you nothing. Outcome vs objection does.
Test the headline, image and button together. A multivariate test with a multi-armed bandit finds the best combination in one go.
Judge on revenue per visitor. If the revenue comes later, use leads per visitor or calls per visitor until it does.

Why test the headline first
Three reasons, and they all come back to the first screen.
Everyone sees it. Your guarantee, testimonials and FAQ only get read by people the headline convinced to stay. Fix the top and every section below it gets more readers.
It decides whether people stay. Visitors judge in a few seconds whether they’re in the right place. The headline is the main thing they use to decide.
It’s cheap to change. A new headline takes five minutes. A new offer, video or page design takes days.
This matters even more on a phone. In the paid-ad funnels I run, 90% or more of the traffic is mobile, and on a phone the first screen is often just the headline, the image, one sentence and a button. Those few things carry the whole page.
One exception: if the page is broken (a slow load, a form that fails, sales that aren’t tracked), fix that first. My landing page CRO checklist is a good pass to run before you test anything.
How many headline versions to test
Most guides tell you to test two. I think that’s a waste of a test. Why just two? If both are mediocre, you’ve spent a month finding the better of two mediocre headlines.
Here’s how I do it instead:
Write 10 versions, each from a different angle (the six angles are below).
Split the traffic equally to start, so every version gets a fair chance.
Cut the clear losers early. After a few days, the bottom versions are usually obvious. Turn them off and give their traffic to the rest.
Keep the best five, six or seven running longer, until you have enough sales to trust the result.

A multi-armed bandit split does step 3 for you. It sends more visitors to the versions earning more and fewer to the weak ones while the test runs, so you don’t pay to keep showing a losing headline. In Splitlab you can test up to 10 or 20 versions this way.
How many you can run depends on your traffic. With a lot of traffic, start with ten. With less, start with five, and expect the test to run longer.
How to write headline variants that are genuinely different
A headline test is only as good as the difference between the versions. If they all make the same promise, the result will be noise, and even if one “wins”, you won’t know why.
So I don’t start with words. I start with an angle: the reason someone would want this. Here are the six angles I use most, each with why it can work:
Outcome. Say the result they get. Why: people buy the result, not the product. This is the safest starting point and often the control.
Pain. Name the problem they have right now. Why: someone who feels the problem stops scrolling when you describe it better than they could.
Specific number or proof. Put a concrete number or result in the headline. Why: specifics sound checkable, so they feel more believable than adjectives.
Objection. Answer the main reason people don’t buy. Why: if most buyers hesitate over the same thing (time, money, “will it work for me”), answering it first removes the brake.
Audience call-out. Name who it’s for. Why: “for fitness coaches” makes the right people feel the page was written for them.
Curiosity vs clarity. A curiosity headline hides the answer to make people read on. A clear one says exactly what they get. Why it’s worth testing: curiosity can pull more clicks, but it can also pull the wrong people.
Write one or two headlines per angle and you already have ten versions.

Here’s how that looks on three info products. All the headlines below are examples I wrote for this post, not results from real tests.
A course: outcome vs objection
Say you sell a $497 course that teaches people to start a bookkeeping business from home.
Outcome: “Start a home bookkeeping business and land your first 3 clients in 60 days”
Objection: “No accounting degree? You can still start a bookkeeping business from home”
What this tests: whether your buyers are pulled by the result, or held back by doubt about whether they’re qualified. If the objection wins, that’s a big lesson. The main job of your whole funnel is giving permission, not selling the outcome.
A webinar registration: audience call-out vs pain
Say you run a free webinar that sells a coaching program for coaches.
Audience call-out: “Free training for coaches stuck under $5k a month”
Pain: “Posting every day and still no clients? Free training shows why”
What this tests: whether naming the audience beats naming the frustration. The call-out is narrower. It may get fewer registrations but more people who are a fit for a $997 program.
A $27 ebook: specific number vs curiosity
Say you sell a $27 ebook of Instagram hooks for coaches.
Specific number: “50 copy-and-paste Instagram hooks for coaches”
Curiosity: “The one-line opener nobody scrolls past”
What this tests: whether buyers want to know exactly what’s inside before they pay, or respond to intrigue. On a $27 product people decide fast, often on the first screen.
If one angle wins, you learn something about your buyers that you can use everywhere else. That’s the bar for a headline variant.
Test the headline, image and button together
Here’s the problem with testing one thing at a time. Say you test the headline this month and the image next month. The winning headline was chosen with the old image. The winning image might not work with the winning headline at all. And you’ve spent two months.
A multivariate test fixes that. You set up several headlines, several images and several buttons, and the tool shows visitors every combination:
3 headlines × 2 images × 2 buttons = 12 combinations, all running at the same time.
With a multi-armed bandit split, the weak combinations get less traffic early, so the test doesn’t need as many visitors as a plain multivariate test.
At the end, you get the best combination, not just the best headline.

I also test the headline and the subheadline as one block. They’re read together, so they should be written and tested together. A great headline with a subheadline that no longer fits it will lose for the wrong reason.
I explain the trade-offs in A/B testing vs multivariate testing.
Keep the message match with the ad or post
Before you write any variant, check what the visitor saw just before they landed. Say someone taps a Meta ad that reads “Stop posting every day for zero DMs”. The page should open with that same promise. If your variant suddenly talks about “building a personal brand”, people feel they’re in the wrong place and leave, and the variant loses for a reason that has nothing to do with its angle.

There are two clean ways to handle this:
Test within the ad’s promise. Keep the core promise from the ad in every version, and vary the angle around it.
Test the whole message, ad and page together. If you want to test a completely new angle, run a new ad with that angle and send it to the matching headline.
How long to run the headline test
It depends on your traffic.
High traffic: sometimes a week is enough to see a clear winner.
Normal traffic: around two weeks is a good length.
Low traffic: plan on a month or more.
Whatever the length, a few rules hold:
Run it in full weeks. Why: people who arrive on a Saturday often behave differently from people on a Tuesday.
Keep each visitor on one version. Why: if visitors flip between versions on a second visit, the groups aren’t comparable. A testing tool handles this. Sending one ad to page A and another ad to page B is not a headline test, because the ads bring different people.
Cut losers, but don’t crown a winner early. Why: dropping a version that’s clearly far behind is safe. Calling the final winner after three days isn’t, because the lead swings a lot early on.
Write down what you expect before you start. Something like “Because buyers doubt they’re qualified, the objection headline will raise revenue per visitor.” Why: a written reason tells you what you learned even when the test loses.
The full step-by-step is in my guide on how to A/B test landing pages.
Judge the winner on revenue per visitor
This is where most headline tests go wrong. The tool says B got more opt-ins, so B wins. But opt-ins aren’t money.
A curiosity headline is the classic case. It can pull more people into your funnel, because they want to know the answer. But some of them signed up out of curiosity, not because they wanted what you sell.
Here’s an example with made-up numbers to show the effect. A webinar registration page for a $997 coaching program, 1,000 visitors on each version:
A (clarity): “Free class: fill your coaching calendar without cold DMs.” 200 people register (20%). 6 buy. That’s $5,982, or $5.98 per visitor.
B (curiosity): “The weird DM trick nobody talks about.” 280 people register (28%). 4 buy. That’s $3,988, or $3.99 per visitor.

B wins on registrations by a lot. A makes about 50% more money per visitor. (Six sales against four is far too few to call a real winner. These numbers only show the direction.)
Revenue per visitor is the main number for almost everyone. But sometimes the revenue is delayed, like a sales call that closes two weeks later. Then judge on the step right before the money, per visitor:
Leads per visitor on an opt-in page.
Calls booked per visitor in a book-a-call funnel.
Revenue per visitor as soon as the sales come in.
I go deeper on it in revenue per visitor vs conversion rate.
What to do with the result
A headline test doesn’t just give you a better headline. It tells you which reason to buy works best on your audience. Use it everywhere:
Ads. If the objection angle won, write your next Meta and Instagram ads around that objection.
Emails. Use the winning angle in subject lines and the first line of your launch emails.
The rest of the funnel. Open your webinar, your sales page and your sales video with the same angle.
The next test. Keep the winner as the new control and test a fresh batch of angles against it.
If the test ends with no clear difference, that’s a result too. It tells you the headline isn’t what’s holding the page back, so test something bigger, like the offer.
Quick checks before you launch the test
Run every variant through these before any traffic hits it.

The 5-second test. Show the first screen to someone who doesn’t know the offer for five seconds, then hide it. Ask what it’s selling and who it’s for. Why: if they can’t say, cold traffic won’t get it either.
Read it on a phone. Open each variant on your own phone, not in the editor. Why: a headline that fits on one line on a laptop can turn into five lines and push the button off the first screen.
Check the match. Put the ad or email next to each variant. Why: a variant that breaks the ad’s promise will lose because of the mismatch, not because of its angle.
Confirm revenue is tracked by variant. Make one test purchase or registration on each version. Why: if sales aren’t credited to the variant the visitor saw, you can only judge on clicks.
How Splitlab helps
Splitlab runs A/B, multivariate and multi-armed bandit tests on funnel pages, including pages built in ClickFunnels and GoHighLevel. You can test up to 10 or 20 headline versions, or headlines, images and buttons together, and let the bandit move traffic to the winners. Every order, upsell and booking is credited to the version the visitor saw, so revenue per visitor decides. See the features or book a demo.
FAQ
How many headline variants should I test at once?
At least five, ideally ten, each from a different angle. Start with an equal split, cut the clear losers after a few days, and keep the best few running. A multi-armed bandit split does the cutting for you. With low traffic, start with five.
How long should a landing page headline test run?
It depends on your traffic. With a lot of traffic a week can be enough, around two weeks is typical, and with low traffic plan on a month or more. Run full weeks, and don’t crown a winner while the lead is still swinging.
Should I test the subheadline with the headline?
Yes. They’re read together, so write and test them as one block. If you want to test the image and button too, use a multivariate test so you find the best combination.
Should I test the headline alone or with other elements?
Testing the headline, image and button together in a multivariate test saves time and finds the combination that works best. If you test them one at a time, the winning headline may not fit the winning image.
What metric should decide a headline test?
Revenue per visitor for any page that leads to a sale. If the revenue comes later, use leads per visitor or calls booked per visitor until it does. Opt-ins and clicks alone can crown a headline that lowers sales.
My final thoughts
Most headline tests I see fail before they start. They test two versions that say the same thing in different words, so whichever one “wins” teaches you nothing.
If you take one thing from this post, make it this: test many angles, not two wordings. Start with ten, cut the losers fast, and test the headline together with the image and the button. When one combination wins, you’ve learned what your buyers care about, and that lesson carries into your ads, your emails and your webinar.
And judge it on revenue per visitor. If you want to run your first headline test with every sale tracked to the version the buyer saw, book a demo and we’ll set it up with you.
The best headline isn’t the cleverest one. It’s the one that brings in people who buy.
Built for revenue
Test for revenue, not just clicks.
Run A/B, multivariate, and multi-armed bandit experiments — with real revenue tied to every variation.
We’ll increase your conversion rate by 30% or you don’t pay. Terms apply.
✓
Revenue per variation
✓
A/B & MVT
✓
Multi-armed bandit
Test for revenue, not just clicks.
A/B, multivariate and multi-armed bandit experiments, with real revenue tied to every variation.


