Most small websites already have two free tools installed. Google Analytics 4 counts what happened. Microsoft Clarity shows how people behaved while it happened. Neither of them can tell you whether a change you made actually caused a better result.
That gap is the reason I built SplitPea. It is not a replacement for either tool. It answers a different question, and the three work best when each one is left to do its own job.
Three tools, three questions
The easiest way to see where each tool fits is to look at the question it can answer on its own.
- GA4 answers "what happened?" Sessions, sources, conversions, landing pages and the trend over time. It tells you the contact page converts at 2% and that most visitors arrive on mobile.
- Clarity answers "how did people behave?" Heatmaps, scroll depth, session recordings and frustration signals such as rage clicks. It shows you that half the visitors never scroll past the hero, or that people keep clicking an image that is not a link.
- SplitPea answers "did this change cause a better result?" It splits traffic between the current page and one variant, tracks the goal you chose, and tells you whether the difference is strong enough to act on.
The first two are diagnostic. They help you find a problem and form a view about why it exists. The third is the test of that view.
GA4 no longer runs experiments
For a long time, small teams could lean on Google Optimize for free A/B testing that sat directly on top of Google Analytics. Google shut Optimize down on 30 September 2023, and GA4 did not replace it with a native testing feature. The recommended path is a third-party experimentation tool that sends its results into GA4.
Most of those tools were built for teams with a developer, a tracking plan and an enterprise budget. That left a gap for the people who used Optimize precisely because it was free and simple: small business owners, solo founders and marketers running a site on Squarespace, Webflow or WordPress.
Clarity shows the problem, not the fix
Microsoft Clarity is one of the most useful free tools a small site can install. Watching ten recordings of people abandoning a form teaches you more than a month of staring at a conversion rate.
But a recording is evidence about behaviour, not evidence about a solution. If Clarity shows visitors hesitating on a seven-field enquiry form, the obvious hypothesis is that the form is too long. Cutting it to four fields might help. It might also reduce lead quality, or make no difference because the hesitation was really about price.
Clarity can also show you how a page performed before and after you changed it. That comparison is tempting, but it is not a controlled test. Traffic mix, seasonality, a campaign ending or a competitor's sale can all move the numbers in the same fortnight. It is the same problem I ran into when working out which marketing campaigns were actually incremental: a before-and-after lift is not the same thing as a caused lift.
Where SplitPea fits
SplitPea sits after the diagnosis. You have seen something in GA4 or Clarity, you have a view about why it is happening, and you want to know whether a specific change fixes it.
The workflow is deliberately small:
- Write the hypothesis. "A shorter enquiry form will increase submissions." Writing it down before the test starts stops the result being reinterpreted afterwards.
- Make one change. The visual editor lets you edit a headline, button, image or block without a developer or a deployment.
- Pick the goal. A form submission, a click, a booking or a page visit. This is where most tests quietly go wrong, so it is worth being precise about what the metric actually counts.
- Let traffic split. Half of visitors see the original and half see the variant, and the same person keeps seeing the same version.
- Record the decision. Every test ends as ship the variant, keep the original or inconclusive, with the reason logged.
The install is a single snippet under 8KB that loads asynchronously, so it can sit alongside the GA4 and Clarity tags without a rebuild.
A worked loop
Here is how the three tools fit together on a typical service-business site.
- GA4 shows the enquiry page gets healthy traffic but a low submission rate, mostly from mobile visitors.
- Clarity recordings show mobile visitors scrolling past the form, starting it, then stalling at the fields asking for budget and company size.
- The hypothesis is that the optional-looking fields are creating friction. Removing them should lift submissions without hurting lead quality.
- SplitPea runs the four-field form against the seven-field form, with a form submission as the goal.
- The decision is logged: ship the shorter form, keep the original, or record the test as inconclusive and try something more meaningful.
- GA4 and Clarity then become the monitoring layer again. Did the shipped change hold up over the following month, and did new friction appear somewhere else?
Each tool contributes one step. None of them has to pretend to do the others' work.
What SplitPea does not do
It is not an analytics platform, so it does not try to replace GA4's acquisition and funnel reporting. It does not record sessions or draw heatmaps. It does not run multivariate tests or personalisation programmes for large product teams.
It is also honest about traffic. A small site cannot detect a small improvement quickly, and no tool changes that arithmetic. SplitPea only recommends a winner once confidence passes 90%, and it marks a test inconclusive rather than inventing a result. The practical implication is to test changes big enough to matter. A new offer, a much shorter form or a clearer headline is usually a better use of limited traffic than a button colour. The SplitPea blog has a fuller explanation of how much traffic an A/B test needs.
Why I built it this way
The idea behind SplitPea is the same one behind most of my work: a useful tool starts with the decision it needs to support. For a small business, that decision is rarely "what is our statistically optimal experimentation programme?" It is "should I keep this change or not?"
GA4 and Clarity are very good at showing you where to look. SplitPea is there for the moment after that, when you have an idea and need evidence that it works before you commit to it.
If you want to see how the product itself was built, the SplitPea case study covers the problem, the build and the stack behind it.