Quick answer: A/B testing means showing two versions of a page to different visitors at the same time and measuring which one gets more conversions. On a small website, start with your highest traffic page and test big, meaningful changes such as the headline, the offer or the form, not small details like button colours. Decide on one goal before you start, run the test for at least two full weeks, and do not stop it the moment one version looks ahead.
A/B testing has a reputation as something only big online stores do. Small sites can benefit too, as long as they adjust their approach to their traffic levels.
How A/B testing works
You create a variation of an existing page, called version B, while the original stays as version A. A testing tool splits your visitors so each person consistently sees one version. The tool records how many visitors in each group complete your goal. When enough data has built up, you can tell whether the difference is real or just chance.
The important part is that both versions run at the same time. Comparing this month with last month is not a test, because too many other things change between months.
Is your site ready for A/B testing?
Testing needs enough visitors and conversions to reach a reliable result. As a rough guide:
- If a page gets only a handful of conversions a month, a test may take many months to give a clear answer
- If a page gets steady daily traffic and regular conversions, testing is realistic
- The smaller the difference between versions, the more traffic you need to detect it
If your traffic is low, you can still improve conversions. Fix obvious problems first, using analytics, visitor recordings and a review of your key pages. Our guide to landing page elements that convert and our CRO audit checklist are good places to start.
Tools for small sites
- Google Analytics Content Experiments is free and built into Analytics. You create a separate URL for each variation and Analytics splits the traffic. It is a good starting point if you already use Analytics goals.
- Optimizely lets you create variations in a visual editor without building new pages, and is popular with marketing teams.
- Visual Website Optimizer (VWO) offers similar visual editing along with heatmaps.
- Unbounce and similar landing page builders include A/B testing for pages built with them.
- WordPress plugins such as Nelio A/B Testing let you test pages, headlines and themes from inside the dashboard.
Whatever you choose, check that it does not noticeably slow your page or cause a visible flash of the original before the variation loads.
What to test first
On a small site, every test is precious, so test changes with a real chance of a big effect.
High impact ideas
- The headline. Does it describe the customer’s problem and your solution clearly?
- The offer. A free consultation, a quote within 24 hours, a guide, a trial or a discount.
- The call to action. Its wording, placement and whether there is one clear action or several competing ones.
- The form. Fewer fields, or splitting a long form into steps.
- Proof. Adding reviews, examples of work, guarantees or accreditations near the decision point.
- Page length. A short page versus a longer one that answers more questions.
Low impact ideas to skip for now
- Button colours
- Small wording tweaks
- Image swaps that do not change the message
These can matter on very high traffic sites, but on a small site they rarely produce a detectable difference.
How to run a good test
1. Start with a hypothesis
Write down what you are changing, why, and what you expect. For example: “Visitors are unsure about price, so adding a ‘typical project cost’ section will increase quote requests.” A hypothesis makes the result useful even if the test loses, because you learn something about your customers.
2. Pick one primary goal
Choose the conversion that matters, such as quote requests or purchases, and set it up in your tool before launch. Clicks on a button are not the same as completed enquiries.
3. Split traffic evenly
Fifty percent to each version is standard. Make sure returning visitors keep seeing the same version.
4. Run it long enough
Run tests for full weeks, so weekday and weekend behaviour are both included. Two weeks is a sensible minimum, often longer on small sites. Decide the duration or sample size before starting.
5. Do not peek and stop early
Results swing wildly in the early days. Stopping a test as soon as one version pulls ahead is the most common way to get a false winner. Let it run its planned course.
6. Check the result properly
Most tools report a confidence or significance level. Treat anything below a high level of confidence as inconclusive. Also check results by device, since a change can help desktop visitors and hurt mobile visitors.
7. Record what you learned
Keep a simple log of each test: the hypothesis, the dates, the result and what you will do next. Over time this becomes a valuable record of what your customers respond to.
Common mistakes
- Testing several changes on different pages at once, so you cannot tell what caused the result
- Running a test during an unusual period, such as a big promotion or a holiday
- Forgetting to check the variation works properly on phones
- Calling a winner on too little data
- Never implementing the winner once the test ends
SEO and testing
Search engines are fine with A/B testing done properly. Google’s guidance is to use rel=”canonical” on variation URLs pointing to the original, use temporary 302 redirects rather than permanent ones when redirecting to variations, avoid showing search engine crawlers different content from users, and end tests once you have a result.
A sample first test plan
Here is what a sensible first test could look like for a service business:
- Page: the main service page that gets the most organic traffic
- Hypothesis: visitors leave because they cannot see what happens after they enquire
- Change: add a short three step “what happens next” section beside the form
- Goal: completed quote requests, tracked by a thank you page
- Duration: four full weeks, or until the planned sample size is reached
Simple, focused plans like this are easier to run, easier to read and easier to act on.
FAQ
How many visitors do I need for A/B testing?
It depends on your current conversion rate and how big a change you expect. Online sample size calculators can give an estimate before you start.
Can I test more than two versions?
Yes, but each extra version needs more traffic. On small sites, stick with two.
What if the test shows no difference?
That is still useful. It tells you the change did not matter to visitors, so you can move on to a bolder idea.
Want help testing the right things?
We help businesses find the changes most likely to win more customers, then test them properly. Learn how we turn visitors into customers.
