How to Use Crazy Egg Experiments to Improve Conversions
How to Use a Crazy Egg Experiment to Improve Landing Page Conversions
Setting up a Crazy Egg experiment is one of the fastest ways to test landing page changes without touching your code or waiting for a developer.
Quick answer: A Crazy Egg experiment is a built-in A/B test that swaps elements on your live page for a percentage of visitors, then tracks which version drives more goal completions. To run one, install the tracking snippet, pick a single hypothesis, create a variant using the visual editor, define a conversion goal, and let it run until you hit statistical significance.
Most founders install Crazy Egg, stare at a heatmap for ten minutes, then forget about it. That's a waste. The heatmap tells you where people click. Experiments tell you what actually moves revenue. Here's how to run one that produces a real answer instead of noise.
What is a Crazy Egg experiment?
Crazy Egg calls its A/B testing feature "Experiments." It lets you create variants of an existing page using a visual editor, split traffic between them, and measure which version wins based on a goal you define (a click, a form submit, or a URL visit).
You don't need a second URL. You don't need to duplicate your page in your CMS. The tool injects the variant on the fly for a slice of visitors. That matters because it means marketers can test without waiting on engineering.
If you're deciding whether the tool is worth paying for at all before you invest time here, check our breakdown of whether Crazy Egg pricing is worth it in 2026.
How do you set up your first Crazy Egg experiment?
Six steps. Do them in order.
1. Install the tracking snippet
Grab the JavaScript snippet from your Crazy Egg dashboard and paste it into the <head> of your landing page. If you're on Webflow, Framer, or Carrd, use the custom code section. Verify the snippet fires by loading your page and checking the "Sites" tab in Crazy Egg. It should show a green connected status within a few minutes.
2. Write a hypothesis before you touch the editor
This is where most people screw up. Don't open the visual editor and start moving things around. Write down what you believe, why, and what you expect to happen. Format:
"Because [observation from data], we believe [change] will increase [metric] by [amount]. We'll know when we see [signal]."
Example: "Because our heatmap shows 68% of clicks on the hero go to the demo video instead of the CTA, we believe removing the video thumbnail and enlarging the CTA button will increase trial signups by 15%."
If you can't fill in that template, you don't have a test yet. You have a whim.
3. Create the variant
In Crazy Egg, click "Create Experiment," select the page URL, and use the visual editor to make one change. One. Not five. If you test three changes at once and the variant wins, you have no idea which change caused the lift.
Common single-variable tests worth running:
- CTA copy ("Get started" vs "Start my free trial")
- CTA color contrast against the background
- Headline specificity (generic benefit vs specific outcome with a number)
- Removing a distracting element (chat widget, secondary nav, testimonial carousel)
- Above-the-fold layout (see our post on above-the-fold problems that kill first impressions)
4. Define your conversion goal
Crazy Egg supports three goal types: click on an element, page visit (thank-you page), or form submit. Pick the one closest to revenue. A click on the CTA is a weak signal. A visit to /signup-success is stronger. A form submit tied to a qualified lead is best.
Set the goal before you launch. Changing the goal mid-test invalidates your results.
5. Set traffic allocation
For your first test, split 50/50. Once you're confident in your setup, you can weight tests differently (say, 90% control / 10% variant for risky changes on high-traffic pages), but 50/50 gets you to significance faster.
6. Launch and don't touch it
Resist the urge to peek after two days and declare a winner. Early results lie. This is the single biggest mistake founders make with any A/B test.
How long should you run a Crazy Egg experiment?
Until you hit statistical significance with enough sample size. Two rules:
-
Minimum sample size per variant. Use a calculator like Evan Miller's sample size calculator. Plug in your baseline conversion rate and the minimum lift you care about detecting. If your baseline is 3% and you want to detect a 20% relative lift, you'll need roughly 8,000 visitors per variant at 95% confidence.
-
Minimum duration: one full business cycle. Even if you hit sample size in three days, run for at least a full week. Traffic behaves differently on Monday vs Saturday. Google's own experimentation guidance recommends full-week cycles to avoid day-of-week bias.
If you have low traffic and can't hit these numbers in a reasonable timeframe, read our guide on A/B testing landing pages without wasting traffic before you spend weeks on inconclusive tests.
How do you read the results without fooling yourself?
Crazy Egg shows you conversion rate, improvement percentage, and a confidence score for each variant. Three things to watch:
Confidence level. Aim for 95% or higher before declaring a winner. Anything below 90% is noise dressed up as a signal.
Sample size per variant. If confidence hit 95% at 200 visitors, ignore it. Small samples produce wild swings. Wait for the number you calculated in step one.
Segment the result. A variant that wins overall might lose for mobile users or paid traffic. Crazy Egg lets you filter by device and source. Check both before rolling out.
If the variant loses or ties, that's still a result. Document what you learned and run the next test. The compounding value of CRO is running many tests, not winning every test.
Crazy Egg experiments vs dedicated A/B testing tools
Crazy Egg experiments are good enough for most SaaS landing pages. But they're not the deepest tool on the market. Here's the tradeoff:
| Feature | Crazy Egg Experiments | Dedicated A/B tools (VWO, Optimizely) |
|---|---|---|
| Setup time | Minutes | Hours to days |
| Visual editor | Yes | Yes, more powerful |
| Multivariate testing | Limited | Full support |
| Advanced targeting | Basic (URL, device) | Deep (behavioral, custom audiences) |
| Server-side testing | No | Yes |
| Bundled with heatmaps | Yes | No, separate purchase |
| Learning curve | Low | Steep |
For indie hackers and early-stage SaaS teams, the bundled heatmap-plus-experiment combo is the whole point. You spot a problem in the heatmap, hypothesize a fix, test it, ship the winner. All in one tool.
If you're evaluating alternatives, our comparison of Crazy Egg vs Hotjar for actionable landing page data covers where each tool wins.
What kind of lift should you expect?
Realistic expectations matter. According to a 2023 CXL analysis of hundreds of tests, the median A/B test produces no statistically notable winner, and among tests that do win, single-digit percentage lifts are common. Double-digit lifts happen, but they usually come from big structural changes (rewriting the whole hero, changing the offer), not button color swaps.
If someone tells you they get 40% lifts every test, they're either running tiny samples or testing pages so broken that anything helps.
Frequently asked questions
Can I run a Crazy Egg experiment without a developer?
Yes. Once the tracking snippet is installed (a one-time developer task), the visual editor lets marketers create variants, define goals, and launch tests without touching code. Complex changes involving new components may still need engineering help.
How much traffic do I need to run a useful Crazy Egg experiment?
Rough rule: you need enough traffic to get at least a few thousand visitors per variant within two to four weeks. If your page gets under 500 visitors per week, focus on qualitative research (heatmaps, session recordings, user interviews) before running quantitative tests.
Does a Crazy Egg experiment slow down my page?
The tracking snippet loads asynchronously and adds minimal overhead in most cases. However, running an active experiment does inject variant code that can cause a brief flicker on first load. Test your page speed before and after enabling experiments and check Crazy Egg's own documentation for anti-flicker recommendations.
Can I test more than two variants at once?
Yes, Crazy Egg supports multiple variants in a single experiment. But splitting traffic across more variants means each one hits significance more slowly. For most landing pages with modest traffic, stick to A vs B until you have enough visitors to support more.
What's the difference between a Crazy Egg experiment and a Crazy Egg snapshot?
A snapshot is a passive recording tool: heatmaps, scrollmaps, and click data on your existing page. An experiment is active: it modifies the page for some visitors and measures the impact on a goal. Use snapshots to find problems, experiments to test solutions.
Should I run a Crazy Egg experiment or use Google Optimize?
Google sunset Optimize in September 2023, so it's no longer an option. Alternatives include Crazy Egg experiments, VWO, Optimizely, and open-source tools like GrowthBook. For most SaaS landing pages under 100k monthly visitors, Crazy Egg's bundled experiment feature is enough.
What happens if my experiment finishes without a winner?
That's a valid outcome, not a failure. It means the change you tested didn't meaningfully affect conversions at the sample size you ran. Document the hypothesis and result, then design a bigger, bolder test. Small changes rarely produce measurable lifts.
Ready to turn experiment results into shipped improvements? PagePulse tracks which landing page variants ship, when, and how they perform over time, so you build a real optimization system instead of a folder of forgotten tests. Start a free PagePulse workspace and connect your first page in under two minutes.