A/B Testing You changed the site. Nobody knows if it helped.
We A/B test it and keep the winner.We split your visitors between two versions, count the calls, chats and forms each brings, and keep the winner.
- One change per test
- Run in the tools you already pay for
- If testing won’t work, we say so
Before any test
Not every idea needs a test.
A broken form needs fixing, not testing. So we sort your list first.
- Form fails on phones
- Page loads slowly
- Button goes nowhere
Fix now
Broken things. They need no test, and testing them wastes your visitors.
- The headlineAB
- The offerAB
- The main photoAB
Test
Matters of taste. Things sensible people disagree on. Your own visitors settle them.
Too quiet to test? If a fair test would take months, we say so and make the clear fixes without one.
Too early to call
An early lead is often luck.
Stop a test the day B pulls ahead and you may keep a lucky winner.
A read date, set before launch. We read the result on that date, not before.
At least one full week. Weekdays and weekends both count. Optimizely’s minimum is “one business cycle (seven days)”.
Why “winners” fade. Peeking daily makes the tool’s score “meaningless” (Evan Miller, 2010).
Sources: Optimizely docs and Evan Miller, checked 17 Nov 2026.
Put the change your team argues about in the form.
Get StartedOne change per test
Change five things. Learn nothing.
After a redesign, enquiries drop and nobody can say which change did it.
5 changesWhich one?
Old page against new? That tells you which page wins, not why.
1 changeThis one.
One change per test. Then the result tells you exactly what moved your enquiries.
How a test runs
Four steps. One number decides.
That number is real enquiries: forms, calls, WhatsApp chats and bookings. Clicks don’t count.
-
1Pick one change.
Headline, offer, photo or form, and you approve it. Big changes need fewer visitors to prove.
-
2Split visitors in half.
At random, in your own testing tool if you have one. A person checks both versions on phones.
-
3Wait for enough.
We work out the visitors needed before it starts.
-
4Keep the winner.
You decide. We build it in and take the test down.
The test log
Every test, logged and yours.
Kept, dropped or no difference, every test goes in a log you keep.
Test logExample: a car workshop
- 3 MarHeadline: “Fixed-price servicing”KEPT
- 31 MarOffer: free pick-upDROPPED
- 12 MayPhoto: workshop or mechanicNO DIFFERENCE
- 9 JunForm: phone number firstKEPT
- 14 JulButton: WhatsApp firstDROPPED
- Next test
A draw counts too. It saves you building a change that made no difference.
It stays in your accounts. The systems, the code and the accounts are yours.
A new manager starts informed. What was tried, what won, and no headline gets tested twice.
Your team
The people behind your tests.
One team runs your website, ads and tests, so nobody changes the ads halfway through a test.

Quality specialist.
Checks both versions on phones before visitors see them.

A dedicated account manager.
One person who knows your business and owns the work.
- 1We review your details.
- 2We talk it through with you.
- 3We propose a solution.
- Priority access to Luna. Partners go to the front of the queue with our AI assistant.
- We accept only a limited number of partnerships each quarter.
Questions
Before you split your visitors.
Traffic, time and tools, answered straight.
What is A/B testing?
A/B testing, also called split testing, is a fair race between two versions of one thing on a website, such as a headline, an offer or a form. Visitors are split at random. SuperToasty’s A/B testing service runs that race for businesses with steady traffic and keeps the version that brings more enquiries.
Read date, set before launch.
How long should an A/B test run?
An A/B test should run until each version has had the number of visitors worked out before it started, and for at least one full week, so weekdays and weekends both count. Optimizely’s guidance is “a minimum of one business cycle (seven days)” (checked 17 Nov 2026). Enough visitors means the gap between the versions is too big to be luck. We set the read date before launch.
Do we have enough traffic to A/B test?
A site has enough traffic to A/B test when a fair test of one change would finish in weeks, not months. That depends on how often visitors enquire now and how big the change is: a new offer needs far fewer visitors to prove than swapping the order of two photos (Optimizely, checked 17 Nov 2026). We work it out from your own numbers first. If the answer is months, we say so.
Why did our test winner stop working once we switched it on?
A test winner usually fades for one of four reasons: the test was stopped early while luck favoured one side, it counted clicks instead of enquiries, the ads changed halfway, or the new version won on desktop and lost on phones. We fix the read date and the deciding number before launch, and one team plans your ads and the test together.
What does your A/B testing service include?
An A/B testing project with SuperToasty includes enquiry counting checked before any test, a sort of your ideas into things to fix and things to test, a written plan for each test, both versions checked on phones, the result read on the planned date, the winner built into your site, and a test log you keep.
Do we have to pay for an A/B testing tool?
An A/B testing tool can cost nothing extra. HighLevel says split testing in funnels is “available across all plans” (help article, updated 9 Sep 2026), and GrowthBook’s Starter plan costs nothing for up to three users, with unlimited traffic (pricing page, checked 17 Nov 2026). Google Optimize closed on 30 September 2023. We use what you already pay for first, and the account stays yours.
Can we run A/B tests in GoHighLevel?
GoHighLevel can split test any funnel step: you create a variation of the step and set what share of visitors sees each version. HighLevel’s help centre says split testing “will only be possible once you have added a domain to your funnel” (updated 9 Sep 2026). If you already pay for HighLevel, we can run your tests there rather than add another tool to your bill.
Will A/B testing hurt our Google ranking?
A/B testing doesn’t hurt Google rankings when it follows Google’s own rules: show Google the same versions visitors see, mark the original page with a canonical tag (a line telling Google which page is the real one), use a temporary 302 redirect, not a permanent 301, and remove the test once it ends (Google Search Central, updated 10 Dec 2025).
How is A/B testing different from conversion rate optimisation?
A/B testing is one method inside conversion rate optimisation, the whole job of turning more of your visitors into enquiries. A/B testing proves which of two versions works better. Finding where visitors leave is Conversion Audits. Writing a better offer and better words to test is Offer & Copy Optimisation. Our Conversion Rate Optimisation page helps you pick which your site needs first.
Keep going
Not quite what you need?
See the rest of this group, or the service closest to this one.
