How Much Traffic Do You Need to Split Test on WordPress?
The short answer is: fewer visitors than you might think, but more conversions than most people have when they start.
Traffic volume is not actually the variable that matters. Conversions are. And those two things are related by your current conversion rate, which varies enormously depending on what you are measuring.
Here is how to figure out whether you have enough to run a meaningful test — and what to do if you do not.
The threshold: 50 to 100 conversions per variation
Statistical significance is what tells you whether a test result is real or just noise. Without enough conversions, a difference in conversion rate between two variations could easily be random chance rather than evidence that one version is actually better.
The working minimum is 50 conversions per variation. At that point you have enough data to start drawing reasonable conclusions. At 100 conversions per variation the confidence becomes significantly stronger.
AB Split Test enforces this automatically. As of v2.5.1, winners require at least 50 visits per variation before a result can be declared. If your test has not reached that threshold yet, the results panel shows an Underpowered badge so you know to keep waiting rather than acting on inconclusive data.
Calculating how much traffic you need
Once you know how many conversions you need, you can work backwards using your current conversion rate.
Required Traffic = Conversions Needed / Baseline Conversion Rate
Example: Your current conversion rate is 2% and you want 100 conversions per variation.
100 / 0.02 = 5,000 visitors per variation
That is 10,000 total visitors split across two variations. If your site gets 10,000 visitors a month, that test could complete in around a month. If it gets 2,000 visitors a month, you are looking at five months.
Use the calculator below to work out your own numbers:
Split Test Traffic Calculator
Current conversion rate
Conversions needed per variation
Minimum 50 for reliable results. 100+ for strong confidence.
Number of variations (including original)
Monthly visitors to this page
Visitors needed
10,000
total across all variations
Estimated duration
2.0 mo
at your current traffic
Readiness
Estimates only. Results vary by site.
How long should you run the test?
The answer depends on which test mode you are using. AB Split Test supports two approaches and they work quite differently.
Standard testing (Bayesian)
This is the default mode. Traffic is split evenly between your variations throughout the test. The plugin monitors results continuously and when there is enough statistical evidence, Autocomplete declares a winner and shows it to everyone automatically.
A minimum of one full week is recommended regardless of how fast conversions are coming in. Behavior on Mondays is different from Saturdays, and a test that only captures a few days will reflect that unevenness. Two weeks is safer. Beyond three months, seasonal drift starts to affect the data and results become harder to interpret.
Multi-Armed Bandit
This mode does not wait for the test to finish before acting on what it learns. As data comes in, AB Split Test automatically shifts more traffic toward the better-performing variation and reducing how many visitors see the losing version while the test is still running. It is better suited to high-traffic sites where the cost of sending 50% of visitors to a losing variation adds up quickly. It can also run indefinitely as a continuous optimization mode rather than ending at a fixed point.
For most WordPress sites, standard testing is the right starting point. Multi-Armed Bandit makes more sense once you are running tests regularly on pages with significant traffic.
What to do if you do not have enough traffic
If your calculator result came back red or amber, you are not stuck and you just need to measure something that happens more often than your primary conversion event.
We covered this in detail in the guide to split testing low traffic WordPress sites. The short version is below.
Move your conversion goal higher up the funnel
Instead of waiting for a purchase or form submission, measure something that happens earlier and more often.
Time Active — AB Split Test counts 30 seconds of active engagement (scrolling, clicking, interacting) as a conversion. It only counts real activity, not idle browser tabs. On a low-traffic site this is one of the most useful goals because it fires far more frequently than a lead or sale, and active engagement correlates well with downstream intent.
Scroll Depth — measure whether visitors scroll past a specific percentage of the page. If your key content or CTA sits at 60% of the page, you can test which hero variant gets more people there without waiting for a form submit.
Button or link clicks — instead of waiting for a thank-you page, measure clicks on the CTA button itself. You will accumulate conversions much faster and still get directional data on which variation is driving more intent.
Both Time Active and Scroll Depth are built into AB Split Test as native conversion goal types no extra setup needed.
Test your highest-traffic page first
Whatever page gets the most visitors is where your data accumulates fastest. For most sites that is the homepage. Even if the homepage is not where conversions happen, testing elements there with a higher-funnel goal gives you faster, more actionable results than testing a low-traffic landing page with a hard conversion goal.
The bottom line
You do not need massive traffic to run split tests. You need enough conversions to tell the difference between signal and noise. If your primary goal is slow to accumulate, move it earlier in the funnel. If your traffic is genuinely very low, run tests on engagement metrics and treat them as directional rather than definitive.
The sites that get value from split testing are not always the ones with the most traffic. They are the ones that keep running tests consistently and let the data compound over time.