A/B testing means showing two versions of a page and seeing which one gets more of the action you want. It works well for sites with a lot of visitors. A small local business might get a few dozen visits and a handful of calls a month. That is not enough for most tests to give a clear answer.
This post shows how to test anyway, what to expect and how to avoid fooling yourself. It is part of our landing page guide for small business.
The small-traffic problem
Numbers bounce around by chance. Here is a made-up example to show the idea.
Suppose version A gets 100 visitors and 4 calls. Version B gets 100 visitors and 6 calls. B looks 50 percent better. But a two-call difference could easily come from luck, a slow week or one good ad click. Run the same test again and the result could flip.
To tell a real difference from noise you need many more visitors and conversions than most local pages get in a month. We are not going to give you a magic number here, because it depends on your current rate and how big a change you are looking for. Testing tools can calculate it for you. What matters is the principle: small counts mean weak evidence.
What to do instead
You still have options.
1. Make big changes, not small ones
A button color change is a small change. It will not show up in a handful of leads. A new headline, a different offer, a shorter form or real photos in place of stock are bigger changes. Big changes make a bigger difference, so you can see them with less data.
2. Compare before and after, with care
Run version A for a set period, then version B for the same length of time. This is not as clean as a true A/B test. Seasons, weather, ad changes and holidays can all muddy the result. Write down anything else that changed so you can judge.
3. Fix the obvious first
Some problems do not need a test. A broken form, a slow page, a phone number that is not tappable and a headline that does not match the ad are all problems to fix, not to test. Use our 12-point speed and mobile check and the scorecard first.
4. Talk to your leads
Ask each new customer what made them call and what almost stopped them. Ten conversations can teach more than a month of data. Write the answers down.
5. Watch real behavior
Free tools can show recordings or heatmaps of how people use the page. With low traffic, watching 20 sessions can show you where people get stuck.
6. Test where you have more volume
If your ads send traffic to the page, you can test the ad text and keywords, which often get more impressions than the page gets visits. Start there if the page volume is too low.
What to test first
Rank ideas by how big the change is and how easy it is to do.
| Idea | Size of change | Effort |
|---|---|---|
| New headline that matches the ad | Large | Low |
| Add real reviews near the top | Large | Low |
| Shorten the form | Medium to large | Low |
| Move the call button to the first screen on mobile | Medium | Low |
| Replace stock photos with your own | Medium | Medium |
| Add a clear “how it works” section | Medium | Medium |
| Change button text from “Submit” to a clear action | Small to medium | Low |
| Change button color | Small | Low |
Start at the top. Skip the bottom unless you have plenty of traffic.
For headline ideas, see landing page headline formulas. For forms, see how many fields to ask for.
How to run a simple test
- Pick one question. “Will a shorter form get more requests?”
- Pick one measure. Calls plus form fills, tracked the same way for both versions.
- Change one main thing. If you change the headline, the photos and the form together, you will not know which mattered.
- Split traffic evenly, or run the versions one after the other for equal lengths of time.
- Decide in advance how long. Pick an end date or a visitor count before you start so you do not stop the moment one version looks ahead.
- Write down the result and what else changed.
- Keep what is better, or repeat if unclear.
Track calls and forms before you test, or you have nothing to compare. See what to track in Google Ads.
What Google says about tests and search
If your landing page also shows up in search results, there are a few rules. Google’s guidance on A/B testing and Google Search says:
- Do not show Googlebot a different page than humans. That is cloaking and it breaks Google’s spam policies, whether or not you are running a test.
- If you test with different URLs, use rel=”canonical” on the alternate URLs rather than noindex.
- If you redirect people to a test variation, use a 302 (temporary) redirect, not a 301.
- Run the experiment only as long as necessary, then remove the variations and test code.
If your landing page is only used for ads and not meant to be found in search, these matter less, but it is a good habit to follow them.
Reading results honestly
- Do not stop early just because one version is ahead.
- Do not cherry-pick the week that looks best.
- Count real outcomes, such as calls that turned into jobs, not only clicks on the button.
- Look at quality as well as quantity. A shorter form might bring more leads that are less ready to buy.
- Accept “no clear winner.” That is a result. It means the change was not big enough to matter, or you need more data.
When not to test
- When the page is broken or slow
- When tracking is not set up
- When you get almost no traffic at all (fix traffic first)
- When you are about to change the offer, price or service area (test after)
If your page gets very little traffic, work on the basics and on getting the right visitors. Our PPC guide for local businesses and how much to spend on Google Ads cover traffic. The Ad Budget Reality Check can help you see what your budget can reasonably buy.
A simple testing log
Keep a log in a spreadsheet:
| Date | What changed | Why | Measure | Result | Notes |
|---|---|---|---|---|---|
After a year, this log is worth more than any case study from someone else’s business.
Worked example: a low-traffic test plan (hypothetical)
A hypothetical electrician gets about forty visits a month to a landing page and a handful of calls and forms. What can they learn?
Month 1: Count. They set up tracking for calls and forms. They record how many came from the page, and how many became jobs. They test the page on a phone and fix a broken phone link. No testing yet. They just see where they stand.
Month 2: One big change. They replace the generic headline with one that matches the ad, and add two real reviews near the top. They write down the date and what changed.
Month 3: Compare, with care. They compare month 2 with month 1. They also note that month 2 had better weather and a seasonal rise in searches. The comparison hints that the change helped, but it is not proof.
Month 4: Another big change. They cut the form from seven fields to four. They compare lead counts and lead quality.
By month four the electrician has two changes, a log and a better page. They have not run a lab-grade test, but they have fixed things that good practice says matter.
Metrics worth watching
| Metric | What it tells you | Watch out for |
|---|---|---|
| Calls from the page | Interest from people who want to talk | Count only real calls, not hang-ups |
| Form submissions | Interest from people who prefer to write | Spam entries |
| Qualified leads | Calls and forms that fit what you sell | Needs you to judge each one |
| Booked jobs | The result that matters most | Slow to arrive; needs good records |
| Visits | How much traffic the page gets | Not a result by itself |
| Bounce or time on page | Rough hints of fit | Easy to misread |
Count the lower rows (leads and jobs) whenever you can. A change that raises form fills but lowers job quality is not a win.
Risks of testing too much
- Seeing patterns that are not there. With small numbers, luck looks like a trend.
- Changing things before the last test finished. You lose track of cause and effect.
- Letting a test run forever. Google says to run an experiment only as long as necessary.
- Optimizing the wrong thing. A page can get more clicks on a button and fewer real jobs.
- Ignoring the rest of the funnel. Slow follow-up can cancel any gain from a better page. See thank-you pages.
Test ideas by goal
To get more calls: try a larger call button, a sticky call bar on mobile or a headline that mentions calling.
To get more form leads: try a shorter form, clearer button text or an offer such as a free, written estimate.
To get better leads: try stating the service area and typical job type on the page, or adding a field that screens.
To build trust: try real photos in place of stock, or move reviews to the top.
Each idea comes with a trade-off. Write it down before you begin.
When to use a tool
If your page gets a lot of visits, a testing tool can split traffic and report results. If you use one, check that it follows Google’s guidance on cloaking, canonical tags and redirects, as covered above. If you do not have the traffic, spend your time on the changes in landing page mistakes and what makes a landing page convert instead. They are fixes backed by common sense, not tests that need volume.
Questions people ask
How long should I run a test?
Until you reach the visitor count or end date you set in advance. Google says to run an experiment only as long as necessary. For small pages that can mean months. If that is too long, make bigger changes or use before-and-after comparison with care.
Do I need special software?
Not to start. A spreadsheet and call tracking are enough for a before-and-after comparison. Dedicated tools help when you have the volume.
Can I test on my homepage?
You can, but the homepage serves many goals. A single-purpose page is easier to test. See landing page vs website.
Is it better to guess or to test?
Use good practice first. Test the questions that practice cannot answer for your audience.
Keep reading
- Landing Page Guide for Small Business: Build One That Converts
- 9 Landing Page Headline Formulas (With Local Examples)
- How Many Form Fields Should a Landing Page Have?
- 12 Landing Page Mistakes That Waste Your Ad Clicks
- Landing Page Design services
Want a landing page built for your business?
Webfactory Media is a digital marketing agency in Carlisle, PA. We design and build landing pages for local businesses, and we start with a conversation about your offer, your customers and your numbers.
