How-To · Testing
How to Pick Your First Website Test: From Audit Findings to Experiments
Turn a long audit list into one good test. A friendly guide to spotting test-worthy findings, writing a hypothesis and choosing a test that fits your traffic.
Start with a finding that touches a page people already visit and an action you care about, like a form or call button. Write a one-sentence hypothesis, pick the simplest test your traffic can support, and decide before you start how you'll judge it. On low-traffic sites, a careful before-and-after comparison is often more practical than an A/B test.
Key Takeaways
- The best first tests sit on busy pages and touch an action you care about.
- Write a one-sentence hypothesis before changing anything.
- Pick high-impact, low-effort ideas first.
- Low-traffic sites can learn a lot from before-and-after comparisons and session recordings.
Audits are great at producing lists. Fifty findings, all technically true, none obviously first. Testing is how you turn that list into learning: you make a change on purpose, watch what happens and keep what works.
Here's how to pick your first one without overthinking it.
Not Every Finding Is a Test
Some findings are just fixes. A broken link doesn't need an experiment. Neither does a missing page description or a form that stopped sending emails. Fix those and move on.
A finding becomes test-worthy when there's a real question about what to do, like "would a shorter form get more quote requests?" or "would visitors call more if the phone number were higher up?"
Find the Sweet Spot
Good first tests usually share three traits:
- Busy page. The change sits on a page plenty of people already see.
- Meaningful action. It affects a form, call, booking or sale, not just clicks on a footer link.
- Small lift. You can make the change in an afternoon, not a quarter.
Write a Hypothesis First
A hypothesis keeps you honest. It's one sentence that says what you'll change, what you expect and why.
Vague
"Let's try a new button and see what happens."
Clear
"If we shorten the quote form from nine fields to five, more mobile visitors will finish it, because recordings show people quitting at the address fields."
The "because" matters most. It's what you actually learn, whether the test wins or not.
Choose a Test That Fits Your Traffic
| Approach | How it works | Good for |
|---|---|---|
| A/B test | Half of visitors see version A, half see B, at the same time | Busy pages with lots of conversions |
| Before and after | Change the page, compare matching periods before and after | Smaller sites, simple changes |
| Qualitative check | Watch recordings or ask a few customers | Very low traffic, early ideas |
A/B tests need enough conversions to tell real differences from noise. If your page gets a handful of leads a month, a clean before-and-after comparison is usually more practical. We cover how to do that fairly in Did That Fix Actually Work?
Decide How You'll Judge It, Up Front
Before launching, write down:
- The one number that decides it, like completed quote forms per week
- How long it runs, ideally full weeks so weekdays and weekends both count
- What would make you stop early, like a broken form
Then add an annotation in GA4 on the start date. Future you will be grateful.
Frequently Asked Questions
What if my test "loses"?
You still learned something. Write down what happened and why you think so. A clear loss is more useful than a vague win.
Can I test more than one thing at once?
You can, but it's harder to know which change made the difference. For a first test, change one thing.
What tools do I need?
For a before-and-after test, GA4 and a calendar. For understanding why, Microsoft Clarity's recordings and heatmaps are free and very handy.
Sources
- Events and key events, Google Analytics Help
- About annotations, Google Analytics Help
- Microsoft Clarity, Microsoft