You read that the best growth teams test everything, so you set an A/B test live on your homepage and waited. Eleven weeks later, the dashboard still reads collecting data — not enough visitors for a result. Here is what no testing-tool blog will tell you: at your traffic, A/B testing isn't rigour. It's a queue you'll never clear.

Direct answer: For most owner-operated European businesses, A/B testing should not be your first conversion move. Below roughly 10,000 visitors a month, a split test rarely reaches statistical significance before the problem outlasts the test. Diagnose the obvious leak, fix it, ship it — then measure the only number that pays you: enquiries that turn into clients.

That's not an excuse to stop measuring. It's the opposite. It's about measuring the right thing, on a site that can't feed a test.

The advice is everywhere because the people giving it sell the tool

Notice who keeps telling you to test everything. Most of the time, it's a company that sells A/B testing software.

That doesn't make them wrong about big sites. It makes them quiet about small ones. A testing platform has no reason to publish the one sentence that matters to you: if your site doesn't get much traffic, this product won't help you yet. So the advice arrives stripped of its precondition, and you inherit a tool built for traffic you don't have.

The result is a strange kind of guilt. You change a headline, enquiries tick up, and a part of you whispers that you didn't prove it. Real marketers run tests. You just... changed things.

Drop the guilt. The question was never whether testing is rigorous. It's whether you have the visitors to make it rigorous. For most European founders, the honest answer is no — and it isn't close.

The numbers nobody runs before they hit start

A valid A/B test needs more conversions than most small sites produce in a year.

Here are the figures the calculators bury. A page converting at 2–5% typically needs 1,000–2,000 conversions per variant to detect a 10–20% lift at 95% confidence. To catch a 10% improvement on a 2% baseline, you need roughly 60,000 visitors per variation — about 120,000 visitors for one two-way test. Below 10,000 visitors a month, A/B testing "should not be your primary optimisation tool" at all; a test only turns viable on a page pulling at least a few thousand visits and 100 conversions per version.

Now put your own numbers in. Say your site gets 3,000 visits a month and converts at 2% — both ordinary figures for an owner-operated business. That's 60 conversions a month, total. To reach even the low end — 1,000 conversions per variant, 2,000 across the test — the test would have to run, uninterrupted, for about 33 months. Nearly three years to learn whether one button beat another. (That's a worked example, not a law of physics — but the order of magnitude is the point.)

Three years isn't a test. It's a hostage situation.

At your traffic, the test outlives the problem

A test you can't finish is worse than no test, because it freezes the page while you wait.

Picture the dashboard from the top of this piece. Eleven weeks in, still collecting data. Meanwhile the page it's testing hides your phone number three scrolls down, the form asks for nine fields, and your ad budget keeps spending against all of it. You're not de-risking a change. You're paying rent on a broken page to protect a result that will never arrive.

It gets thinner. The average small or local business site gets 1,000–10,000 visitors a month, and most sites convert between 1% and 4%. Even the companies that do have the traffic mostly don't win: across large analyses of tens of thousands of experiments, only somewhere between 1 in 5 and 1 in 8 tests produces a clear, significant winner. So the upside you're waiting three years for is, more often than not, "no detectable difference."

There's a European wrinkle on top. Most testing tools sort visitors with cookies and scripts that fall under EU consent. A share of your already-small audience declines them, so those people never enter the test at all. Your sample isn't just small. It's smaller than the dashboard admits.

What to do instead: the senior move

Stop optimising the test. Fix the business.

The work that moves a small site isn't a 1.5% nudge you can't even measure. It's the obvious leak you can find in an afternoon. Here's the order a senior partner runs it in:

• Diagnose before you change anything. Watch ten real sessions. Read what people type into the form before they abandon it. Open the page on a mid-range phone on mobile data — the way your buyers actually see it, not the way it looks on your fast laptop. The leak is usually loud: a hidden price, a vague headline, a form nobody wants to fill, a page that crawls. You don't need a test to spot a phone number three scrolls down.

• Make changes big enough to feel. When you can't run a clean test, small tweaks are guesswork and invisible. So make the change that doesn't need a calculator to notice — rewrite the offer, halve the form, put the price on the page, let people message you instead of filling in fields.

• Read the qualitative signal. Session recordings, a five-person user test, the questions that land in your inbox. None of it is statistically pure, and none of it needs to be. It tells you why people leave — the one thing a split test never explains.

• Then measure the number that pays you. Not impressions. Not a dashboard full of green arrows. Enquiries that become clients — tracked from click to client, not click to dashboard. Ship the change, watch real enquiries over the next few weeks, and judge it like an owner, not a lab technician.

This is the loop we run on a conversion-first website: find the leak, fix it, ship it, read the real number, go again next week. No three-year tests. The work compounds because it keeps moving — and you can watch that number in the workspace instead of guessing at it.

When you actually should A/B test

A/B testing earns its place the moment your traffic can feed it — so here's the honest threshold.

If a single page reliably pulls a few thousand visits and around 100 conversions per version inside two to four weeks, test away. Keep it to two variants, not five — every extra version splits your traffic and pushes significance further out of reach. Test changes big enough to matter, not font choices. And as you grow into real testing volume, that's exactly the kind of work worth handing to a senior partner — one fixed monthly fee, every price on the Plans page, no year-long lock-in.

Until then, testing isn't the discipline. Shipping is. On a small site, A/B testing is procrastination with a dashboard — and you have better things to do with the next three years.

FAQ

How much traffic do you need to A/B test

Enough to produce roughly 1,000–2,000 conversions per variant within a few weeks — in practice, several thousand visits and at least 100 conversions per version each month. Below about 10,000 monthly visitors, A/B testing shouldn't be your primary tool, because you won't reach a valid result before the problem you're testing has moved on.

What should I do instead of A/B testing

Diagnose the obvious conversion leak by watching real sessions and reading what visitors type, then make changes big enough to feel — rewrite the offer, cut the form, show the price. Ship the change and judge it by real enquiries that turn into clients over the following weeks, not by a significance badge on a dashboard.

Is A/B testing worth it for a small business

Usually not as a first move. At a few thousand visits a month, a test can take years to reach significance, and even well-resourced teams see a clear winner in only about 1 in 5 to 1 in 8 tests. Fix the visible problems first, and save testing for when your traffic can genuinely feed it.

Plan. Build. Iterate.

That's the loop — and it beats a test you'll never finish. Bring us the page that's quietly losing you enquiries, and we'll find the leak instead of babysitting a dashboard.

Book a 30-min call — bring your worst-performing page, leave with a fix-it list.