Most advice on landing page A/B testing is written for sites with 50,000 visitors a month. A plumber in Etobicoke or a roofer in Oakville might get 400. Run a textbook split test on that traffic and you will wait a year for an answer, or worse, declare a winner that was just noise.
Landing page A/B testing for local businesses works, but only if you change how you do it. This guide covers what to test, how much traffic you actually need, and how to get useful answers from small numbers.
Why standard A/B testing advice fails local service sites
A split test needs enough conversions in each version to tell a real difference from random variation. Local service pages have two problems: low traffic and low conversion volume. A page with 500 visits and a 6% call or form rate produces 30 leads a month. Split that in half and each version gets 15. A swing of five leads either way proves nothing.
The fix is not to skip testing. It is to test bigger changes and set realistic expectations.
How much traffic you need
These are rough figures at 95% confidence and 80% power, for detecting a 20% relative improvement. Use any free sample size calculator to check your own numbers.
| Current conversion rate | Target rate (20% lift) | Visitors needed per version |
|---|---|---|
| 3% | 3.6% | about 13,900 |
| 5% | 6% | about 8,200 |
| 10% | 12% | about 3,800 |
If your landing page gets 600 visits a month, even the best case above takes around 13 months. That is not a test, that is a hobby.
What to test when traffic is low
Small tweaks like button colour need huge samples because the effect is tiny. With low traffic, test changes big enough to move conversion by 30% or more. In order of likely impact:
- The offer. Free quote versus free on-site estimate versus same-day callback. This usually moves results more than design.
- The headline. Service plus city plus a specific promise beats a generic tagline.
- Form length. Cutting a form from seven fields to three is a big swing. We cover the mechanics in our guide to contact form optimization and in multi-step versus single-page forms.
- Trust placement. Reviews, licences and guarantees above the fold versus buried lower down.
- Call versus form emphasis. For urgent trades like plumbing and HVAC, a tap-to-call button at the top often beats a form.
Run the test on paid traffic first
Google Ads traffic is the fastest way to get volume. You control the budget, the audience is consistent, and every click already costs money, so a better page pays back right away. Send a fixed share of clicks to each page version, keep ad copy identical, and run both at the same time. Do not compare this month to last month, because seasonality and ad changes will skew it.
Organic traffic works too, but a split test on a page that ranks carries a small risk if the variant changes the content a lot. Keep the main headings and topic the same.
How long to run a test
- Run at least two full weeks, so every day of the week appears twice.
- Do not stop the moment one version pulls ahead. Early leads reverse often.
- Set the sample size or end date before you start, and stick to it.
- Count real outcomes: calls, form submissions and booked jobs, not just clicks on a button.
Tracking has to be right before you test
A test is only as good as the conversion data behind it. If your call tracking misses mobile taps, or form submissions fire twice, you will pick the wrong winner. Confirm each version records calls and forms separately before you split any traffic. Our write-up on website conversion rate for local businesses covers what a healthy baseline looks like.
What to do when you cannot reach significance
Many local sites will never hit the numbers above. That is fine. Use a sequential approach instead:
- Make one major change at a time.
- Run it for four to six weeks against a stable ad budget.
- Compare lead volume and cost per lead to the previous period, and note anything else that changed.
- Keep the change if results hold or improve, and roll back if they drop.
- Treat the result as a strong signal, not proof.
It is less rigorous than a true split test, but it beats guessing, and it is honest about the uncertainty.
Frequently asked questions
How much traffic do I need for landing page A/B testing?
For a 5% conversion rate and a 20% lift, you need roughly 8,000 visitors per version. Most local service sites have far less, so test bigger changes or use a sequential before-and-after approach.
What should a local business test first on a landing page?
Start with the offer and the headline. These change visitor intent more than layout or colour, so they produce larger, easier-to-measure differences.
How long should an A/B test run?
At least two full weeks, and until you reach your planned sample size. Stopping early because one version looks ahead is the most common mistake.
Can I A/B test without paid ads?
Yes, but organic traffic is usually too low and too slow. Paid traffic from Google Ads gives you volume and control, which makes results arrive faster.
Not sure which page changes would bring in more booked jobs? Get a free SEO audit from Cadiente Digital and we will show you where your landing pages are losing leads.