A/B testing for small businesses (and when you don't have enough traffic)

A/B testing sounds like the scientific way to improve a website: show half your visitors version A, half version B, and keep the winner. For big sites it is. For most small business sites, it's often impossible to run a meaningful test, and pretending otherwise leads to decisions based on noise. This guide explains how testing works, how to tell whether you have enough traffic, and what to do instead if you don't.

Help me choose

Do you have enough traffic to A/B test?

Does the page you want to improve get more than a few conversions a week?

How an A/B test works

Pick one page and one goal, such as contact form submissions from a service page.

Create a variation that changes something meaningful: the headline, the offer, the form length, the page structure.

Split traffic randomly so each visitor sees one version, and keeps seeing the same one.

Run the test until you have enough data, decided in advance, not until one version looks ahead.

Compare the conversion rates and check whether the difference is bigger than chance would explain.

Why traffic matters so much

Small numbers are noisy. If version A gets 3 enquiries and version B gets 5, that looks like a big win for B, but a difference that small could easily be luck. To be reasonably confident a difference is real, you need many conversions per version, and the smaller the improvement you're trying to detect, the more you need.

You can work this out for your own site with a free sample-size calculator (Evan Miller's is a well-known one). You enter your current conversion rate and the smallest improvement you'd care about, and it tells you how many visitors each version needs. For many small business pages, the answer works out to months of traffic, sometimes longer. Over that time, seasons change, campaigns start and stop, and the test stops being a clean comparison.

A rough rule of thumb: if a page gets only a few conversions a week, a classic A/B test on that page is unlikely to give a trustworthy answer in a reasonable time. Run the calculator with your own numbers to see.

Mistakes that produce false winners

Stopping early because one version is ahead. Early leads often vanish. Decide the sample size up front.
Testing tiny changes like a slightly different button shade. Small changes produce small effects, which need huge samples to detect.
Testing many things at once and declaring a winner on whichever metric moved.
Ignoring the calendar. Run tests in whole weeks, so weekday and weekend behavior are both included.
Counting the wrong goal. More clicks on a button is not the same as more qualified leads.

Tools

Google Optimize, which many small businesses used, was shut down by Google in 2023. Options now include dedicated testing tools, testing features built into some website builders and landing-page tools, and the experiment features in ad and email platforms. Features and pricing change often, so check the current plans, and check whether a tool's script slows your site down, since a testing script can delay how fast the page appears.

What to do when you don't have enough traffic

Low traffic doesn't mean you're guessing. It means using different methods.

1. Fix the obvious problems first

Many conversion improvements don't need testing because the problem is clear: a broken form, a button that doesn't show on mobile, a headline that doesn't say what you do, a page that takes ages to load. Work through the 30-minute conversion self-audit before testing anything.

2. Watch real visitors

Session recording and heatmap tools (Microsoft Clarity is free) show where people click, how far they scroll and where they hesitate. Ten recordings of people abandoning your form are worth more than a month-long inconclusive test. Make sure your privacy notice and cookie consent cover the tool.

3. Ask people

Customer interviews: a few short calls with recent customers about what nearly stopped them.
On-page polls: one question on a pricing or contact page.
Five-second tests: show a page briefly to someone outside your business and ask what it's about.
Sales and support teams: they hear the objections every day.

4. Make bold changes, then compare before and after

Since you can't detect small effects, make changes big enough that, if they work, the effect would be clear: a new headline and hero, a much shorter form, prices added to a page. Then compare a few weeks before with a few weeks after.

To make before-and-after comparisons fairer:

Change one main thing at a time, and note the date.
Compare equal lengths of time, ideally the same period in the previous year too if your business is seasonal.
Look at conversion rate (conversions divided by sessions, as Google Analytics measures it), not just total leads, in case traffic changed.
Check whether traffic sources changed. A new ad campaign can move results on its own.

This isn't as rigorous as a controlled test, and you should treat the results as a strong hint rather than proof.

5. Test where you do have volume

Paid ads: ad platforms can test ad copy and send traffic to two landing pages.
Email: subject line tests on a list of a few thousand can be meaningful.
High-traffic pages: your homepage or a popular blog post may have enough visitors even if service pages don't.

Frequently asked questions

What should I test first if I do have the traffic?

Start with changes closest to the decision and most likely to matter: the offer, the headline, the form, the pricing presentation. Leave cosmetic changes for later.

Can AI tools predict which version will win?

Some tools claim to predict attention or preference. They can be a useful sanity check, but they aren't a substitute for what your real visitors do. Treat their output as an opinion.