15 November 2026

How to A/B Test a Booking Page That Fills Your Diary

An A/B test on a booking page shows half your visitors the current page and half a changed version. You then compare how many complete a booking. It works when you have enough traffic, change one thing at a time and judge the result by paid, kept appointments, not clicks. If your page gets only a few hundred visits a month, fix the obvious leaks first and test later. At Plexo, we start from the booking path, and a 90-minute business audit is where we find which step to test.

Plexo

plexo.online

Find the Booking Path’s Biggest Leaks

Plexo’s 90-minute business audit identifies operational constraints and tailored solutions to help align your marketing with revenue.

Visit

TL;DR

A/B testing compares version A (your current page) with version B (one change) on live visitors, split randomly and equally.
Test one variable at a time. Start with the first screen, the trust signals near the button, the number of form fields and the confirmation page.
Pick one main metric, such as completed bookings per qualified visit. Add guardrails: no-shows, cancellations and revenue per visitor.
Low-traffic pages cannot give a reliable answer. Thousands of visitors are often needed.
Roll out a clear winner. Keep the original if the test is inconclusive. Write down every result.

What an A/B test on a booking page is
Is your page ready to test?
What to test first
Set up the test properly
Choose the right metrics
How long should it run?
Read the result and roll it out
Common mistakes
Where the booking page sits in the wider system
Perspective
How Plexo can help
FAQ
Sources

What an A/B test on a booking page is

An A/B test is a simple experiment. Version A is the page you have now. Version B is the same page with one change. Visitors are split into two groups at random and in equal numbers. You then compare what each group did.

The UK government’s digital guidance describes it as like a randomised controlled trial for design choices. It also says you need many users to reach statistical significance. That means the difference you see is unlikely to be luck.

The Nielsen Norman Group, a respected user research firm, defines it as testing two or more variations against predetermined business metrics with a live audience. It also says to change only one design element per variant.

That is the whole idea. The hard parts are choosing what to change, what to count and when to stop.

Is your page ready to test?

Not every booking page should be tested yet. Two things come first.

You need enough traffic. The same guidance warns that A/B testing suits busy pages. Low-traffic pages often need thousands of users before a result means anything. Here is a simple illustration. Say 1,000 people visit your page each month and 5 in 100 book. That is 50 bookings. A change that lifts bookings by one tenth adds five bookings. Five is within normal month-to-month swings. You would not know whether the change worked.

You need a page that works. If the calendar shows slots that are not real, or the confirmation email never arrives, a test only measures the damage. Book a test appointment yourself on your phone. Check that the confirmation email and the calendar entry agree. Fix what is broken before you split traffic.

If traffic is thin, there are better moves. Look at where people leave. Check calls and form fills. Our guide on finding booking leaks with call tracking shows how a 90-minute audit does this. Our article on reducing CAC by fixing booking first explains why this often beats buying more ads.

What to test first

Start where visitors decide in seconds. Four areas give the most to learn from.

1. The first screen

A visitor should know three things at once: am I in the right place, what can I book, and what do I do next. Check that the first screen shows the service, how long it takes, the starting price and where it happens. If any are missing, people leave before they open the calendar.

Test one change here. For example, change the headline so it names the service and the outcome. Leave everything else alone.

2. Trust signals near the button

Reviews, credentials and plain cancellation terms all help. Their position matters too. Test the same proof placed above the button against placed below the form. Do not change the proof and the position in one test.

An event ticketing guide on writing copy that turns browsers into buyers is a helpful source of ideas for the wording of headlines and calls to action on a page like this.

3. The form

Every extra field costs you some people. Ask only what you need to confirm the appointment. Test a shorter form against the current one. Or test the order of the fields. The order that tends to work is service first, then practitioner, then time, then personal details, then confirmation.

4. The confirmation page

Many teams forget this one. The confirmation page decides whether people show up. Test clear next steps, a calendar add button and the cancellation rule in plain words. A visitor should not meet your cancellation policy for the first time in a reminder text.

Read the page as a sequence

A booking page is a sequence, not a menu. Cost and duration come before the calendar. The cancellation rule comes before the book button. When you pick what to test, walk through the page in this order and find the first step where you would hesitate as a customer. That step is your first test.

Set up the test properly

A sound setup has five parts.

One hypothesis. Write it in one sentence. “If the first screen shows the price and duration, more visitors will start the calendar.”
One change. Change one element. If you change the headline, the photo and the button together, you will not know which one did the work.
One main metric. Choose it before the test starts. Do not pick the winner after seeing the data.
A fixed run time. Decide how long to run it and how many visitors you need. Do not stop the moment one version looks ahead.
Equal treatment. Follow-up calls, emails and scripts must be identical for both versions. Otherwise you are testing your team, not the page.

For guidance on sample size and duration, the UK government page says to base them on monthly visitors, the size of change you expect and other factors. A sample size calculator will do the sum for you.

Choose the right metrics

Clicks are easy to count. They are also easy to misread. A version with more clicks on the book button but fewer finished bookings is not better.

Use a main metric that sits close to revenue. Good choices are:

completed bookings divided by qualified visits
bookings made within seven days of the visit
paid deposits per visitor

Then add guardrails. These are metrics that must not get worse while the main one improves.

Metric	Why it matters
Completed bookings per qualified visit	The main result
No-show rate	A page that books people who never come is not a win
Cancellation rate	Shows whether terms were clear
Revenue per visitor	Conversion and revenue can point in different directions
Rebooking rate	Shows whether the first visit was worth it
Form errors and time to complete	Shows friction

Record the stages of the path as events: page view, calendar opened, slot chosen, details entered, booking submitted, booking confirmed. Then you can see where people leave. Our guide to GA4 funnel exploration walks through setting this up.

Break results down by segment

PostHog, a product analytics company, warns that an overall improvement can hide a drop for one group, such as mobile users or visitors from one channel. Always split results by device and traffic source. A version that wins on desktop and loses on mobile needs a second look, because most booking visits in wellness start on a phone.

How long should it run?

Run it until you reach the sample size you set at the start. Then keep it going to the end of a full week, so weekday and weekend behaviour are both counted. Seasonality matters too. A test during a school holiday or a promotion will not match a normal month.

Traffic from different sources can behave differently. If a big ad campaign starts halfway through, the mix of visitors changes and the result is muddied. Keep paid campaigns steady during the test.

Do not peek and stop early. A small early lead often fades.

Read the result and roll it out

The Nielsen Norman Group says a statistically significant variant should usually become the new design, and an inconclusive test should keep the original. That is a good rule.

Three things to do after a result:

Check the guardrails. If bookings rose and no-shows rose with them, the net gain may be small.
Check the segments. Confirm the win holds on mobile and on your main traffic source.
Write it down. Record the hypothesis, the change, the dates, the numbers and what you decided. A log of small tests builds up faster than one big redesign.

A/B testing tells you what changed in behaviour. It does not tell you why. The Nielsen Norman Group suggests pairing it with qualitative research. That can be as simple as watching five people try to book, or reading the questions that come into your inbox.

Common mistakes

Changing several things at once. The result is unreadable.
Testing a page with too little traffic. You get noise and act on it.
Picking the metric after the fact. Any test can look like a win if you search long enough.
Counting clicks, not bookings. Look at completed, paid, kept appointments.
Ignoring what happens after the booking. A page that attracts the wrong clients costs you later.
Not testing the whole path. The page may be fine while the confirmation email, the reminder or the phone call loses the client.

Where the booking page sits in the wider system

For a wellness brand, the booking page is one step in a longer path. People arrive from search, social, a Google Business Profile or an email. They book. They are reminded. They attend. They rebook. A test on the page only moves one step.

These pages go deeper on the other steps:

If the numbers going into your tests are wrong, the results will be wrong too. That is why tracking comes before testing.

Perspective

We see a booking page test as a business question first and a design question second. The question is not whether the button should be green. It is where the booking path loses people, and what that loss costs.

A test only helps when the numbers behind it can be trusted. That means bookings tied to the source that brought them, and revenue tied to bookings. Without that, a team can run test after test and still not know whether revenue moved.

This is why our work links content, operations and revenue in one system. The booking page is one part. Follow-up, reminders, the CRM and reporting are the others. We give clients a live operating view of those systems, so a result on the page can be checked against what happened to revenue.

Our advice is simple. Fix what is plainly broken. Measure the path. Then test one thing at a time, and write it down.

Jordan

How Plexo can help

Plexo runs fixed-scope business audits for wellness brands. The Plexo Business Audit is a 90-minute session that finds the operational constraints holding growth back. It costs 499 AUD, one-off, and ends with a tailored 90-day plan.

For one wellness brand, we took monthly revenue from $65k to $110k by working on operations and marketing integration together. After the audit, we can manage the plan with you. That includes CRM work, content, retention and live dashboards for revenue tracking.

You can see the full range on our services and pricing page. Or book an audit or growth call when you are ready.

FAQ

How much traffic do I need to A/B test a booking page?

There is no single number. It depends on your current booking rate and the size of change you want to detect. Guidance from the Nielsen Norman Group says low-traffic pages often need thousands of users to reach significance. A sample size calculator gives you a figure for your own rate.

What should I test first on a booking page?

Start with the first screen. Make sure it shows the service, duration, starting price and location. After that, test the position of trust signals, the number of form fields and the confirmation page. Change one thing per test.

What is the best metric for a booking page test?

Use completed bookings per qualified visit as the main metric. Then watch no-shows, cancellations and revenue per visitor as guardrails. A version that adds bookings but also adds no-shows may not be a real gain.

Can I test more than one change at once?

Yes, but it needs a different method called multivariate testing, and it needs much more traffic. For most wellness brands, one change per test gives clearer answers.

What if the test shows no clear winner?

Keep your original page. An inconclusive result is still useful. You have learned that this change did not matter much, and you can move on to the next idea.