{"id":108098,"date":"2026-02-27T10:16:22","date_gmt":"2026-02-27T18:16:22","guid":{"rendered":"https:\/\/www.crazyegg.com\/blog\/?p=108098"},"modified":"2026-02-27T10:16:23","modified_gmt":"2026-02-27T18:16:23","slug":"ai-ab-testing","status":"publish","type":"post","link":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/","title":{"rendered":"What Can AI AB Testing Tools Do? Benefits and Limits"},"content":{"rendered":"\n<p>A\/B testing tools were some of the first software to incorporate AI into their platforms in a meaningful way. And they did it long before LLMs and AI agents were <em>the <\/em>hot topic.<\/p>\n\n\n\n<p>So while many tools have slapped \u201cAI-powered\u201d onto their marketing materials (without actually improving their product), good AI A\/B testing platforms have legitimately changed the way that teams run experiments.<\/p>\n\n\n\n<p>But these new capabilities have important limits and introduce new risks. This post provides hype-free coverage of what these tools can do today.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">AI A\/B Testing: Overview<\/h2>\n\n\n\n<p>I\u2019m assuming you are already familiar with A\/B testing and how it works: form a <a href=\"https:\/\/www.crazyegg.com\/blog\/hypothesis-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">testable hypothesis<\/a>, build a controlled experiment, split your traffic, measure performance, and find a winner. If you need a refresher, start with our <a href=\"https:\/\/www.crazyegg.com\/blog\/ab-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">guide to A\/B testing basics<\/a>, because we\u2019re just going to jump right in.<\/p>\n\n\n\n<p>AI A\/B testing refers to experimentation platforms that use artificial intelligence to assist with different stages of the testing process, helping teams to:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Run more tests with less manual work<\/li>\n\n\n\n<li>Accelerate the pace of experimentation<\/li>\n\n\n\n<li>Test more complex ideas without writing code<\/li>\n\n\n\n<li>Serve more personalized experiences during testing<\/li>\n\n\n\n<li>Automatically document learnings from past tests in a searchable record<\/li>\n<\/ul>\n\n\n\n<p>The net effect is that a full-fledged A\/B testing program is now within reach for teams that don\u2019t have dedicated copywriters, designers, and developers.&nbsp;<\/p>\n\n\n\n<p>Of course it\u2019s ideal to have those resources, but AI A\/B testing tools can help close skill gaps by generating copywriting, code, and <a href=\"https:\/\/www.crazyegg.com\/blog\/conversion-audit\/\" target=\"_blank\" rel=\"noreferrer noopener\">CRO audits<\/a> instantly.<\/p>\n\n\n\n<p>These are real capabilities that exist today. Teams are already using these tools with documented success.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Why AI works for A\/B testing<\/h3>\n\n\n\n<p>I understand if you are skeptical. There is an ungodly amount of hype around AI, and plenty of brands are overselling what their products can do.<\/p>\n\n\n\n<p>But A\/B testing is genuinely well-suited to benefit from AI.&nbsp;<\/p>\n\n\n\n<p>During a test, every interaction gets recorded. Every test tells you what works and what doesn\u2019t. There\u2019s no ambiguity about what success looks like.<\/p>\n\n\n\n<p>Unlike a lot of other mushier marketing problems, A\/B testing provides a near ideal environment for AI to learn.&nbsp;<\/p>\n\n\n\n<p>And on top of that, you have these newer generative AI tools that can produce the copy, images, designs, and code you need to run the next test.<\/p>\n\n\n\n<p>You can learn more, and act on what you learned faster.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">AI in A\/B testing is not new<\/h3>\n\n\n\n<p>Most of the <a href=\"https:\/\/www.crazyegg.com\/blog\/best-ab-testing-tools\/\" target=\"_blank\" rel=\"noreferrer noopener\">popular A\/B testing tools<\/a> already incorporate machine learning (ML, a subset of AI) into their products, and have for years. Some examples:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/www.crazyegg.com\/blog\/multi-armed-bandit-vs-ab-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>Multi-armed bandit testing<\/strong><\/a>, which relies on ML algorithms to shift traffic in real-time toward variations that perform better.\u00a0<\/li>\n\n\n\n<li><strong>Predictive targeting<\/strong>, which uses ML to predict which users are most likely to convert when shown a particular variation based on behavioral and demographic signals.\u00a0<\/li>\n\n\n\n<li><strong>Anomaly detection<\/strong>, which figures out the normal patterns in your data so it can flag issues that don\u2019t look right.\u00a0<\/li>\n<\/ul>\n\n\n\n<p>In all of these cases, the platform uses ML to \u201clearn\u201d from incoming data in order to improve results or catch problems early.<\/p>\n\n\n\n<p>This generation of AI-enabled platforms builds on this well-understood and heavily-tested foundation. Unlike many other software products, A\/B testing tools are built by people who have a decade or more experience shipping AI features to their customers.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Where is AI A\/B testing today?<\/h3>\n\n\n\n<p>The clearest way to understand the current landscape is to think about where the human experimenter fits in.&nbsp;<\/p>\n\n\n\n<p>Broadly speaking, AI is being worked into testing programs in three ways:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Human-led with AI assistance<\/strong>: The experimenter drives every decision about what to test, how to set it up, and how to interpret results, using AI to reduce the manual work at each stage.<\/li>\n\n\n\n<li><strong>AI-led with human supervision<\/strong>: The platform generates ideas, builds variations, allocates traffic, and interprets results, with humans reviewing inputs, outputs, and analytics.<\/li>\n\n\n\n<li><strong>Automated testing with human-defined guardrails<\/strong>: The platform runs tests continuously within parameters defined by the human team, optimizing tests without much hands-on management.<\/li>\n<\/ul>\n\n\n\n<p>Most teams operate somewhere between the first and second paradigm today, but the third is where this technology is heading.&nbsp;<\/p>\n\n\n\n<p>I was not able to find anyone who claimed to have taken the human fully \u201cout of the loop,\u201d though it is now possible to do so, in theory.&nbsp;<\/p>\n\n\n\n<p>The products that offer \u201cautomated AI A\/B testing\u201d are generally talking about:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Automating specific parts of the testing process, not the full program.<\/li>\n\n\n\n<li>Running continuous multi-armed bandit testing, where the platform promotes winners and pauses losers, but there is still human supervision.<\/li>\n<\/ul>\n\n\n\n<p>If you are curious about the leading edge of automated experimentation, this <a href=\"https:\/\/www.youtube.com\/watch?v=FXJ_-SGpq1E\" target=\"_blank\" rel=\"noreferrer noopener\">episode of the Outperform Podcast<\/a> is great. The discussion focuses on multi-armed bandit algorithms and considers both <a href=\"https:\/\/www.crazyegg.com\/blog\/multivariate-testing-vs-ab-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">A\/B and multivariate testing<\/a>, but you will get a good sense of both the rewards and risks of letting tests run on their own.<\/p>\n\n\n\n<p>Today, the typical A\/B testing team is still firmly in the driver\u2019s seat. They use AI to automate manual tasks and augment their ability to create impactful tests.&nbsp;<\/p>\n\n\n\n<p>Let\u2019s look at what these tools can help you do.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Existing AI A\/B Testing Capabilities: Benefits and Limitations<\/h2>\n\n\n\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" width=\"1200\" height=\"720\" src=\"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png\" alt=\"AI AB testing workflow diagram showing five key capabilities: idea generation, experimental design, impact forecasting, personalization, and results interpretation.\" class=\"wp-image-108100\" srcset=\"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png 1200w, https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image-300x180.png 300w, https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image-1024x614.png 1024w, https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image-768x461.png 768w\" sizes=\"(max-width: 1200px) 100vw, 1200px\" \/><\/figure>\n\n\n\n<p>These five sections cover what you can do with AI A\/B testing platforms today:<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Generate test ideas and variations<\/li>\n\n\n\n<li>Assist with the design and configuration of experiments<\/li>\n\n\n\n<li>Forecast test impact<\/li>\n\n\n\n<li>Enhance personalization<\/li>\n\n\n\n<li>Interpret and summarize results<\/li>\n<\/ol>\n\n\n\n<p>I\u2019ll walk through these capabilities, highlighting where each one goes beyond traditional tools, how teams are using them, and any important limitations and risks.<\/p>\n\n\n\n<p>We\u2019ll cover a few emerging capabilities separately at the end of the post.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">1. Generate test ideas and variations<\/h3>\n\n\n\n<p>Most AI A\/B platforms include tools to help you generate assets for experiments. You decide what to run, but instead of coming up with ideas or coding the variants manually, you can ask the system to create these for you.<\/p>\n\n\n\n<p><strong>What it can do:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Scan a page and suggest hypotheses based on the content, <a href=\"https:\/\/www.crazyegg.com\/blog\/target-audience\/\" target=\"_blank\" rel=\"noreferrer noopener\">target audience<\/a>, or historic performance.<\/li>\n\n\n\n<li>Propose new layouts or design changes.<\/li>\n\n\n\n<li>Generate variations of headlines, copywriting, images, <a href=\"https:\/\/www.crazyegg.com\/blog\/high-converting-cta-buttons\/\" target=\"_blank\" rel=\"noreferrer noopener\">CTA buttons<\/a>, or offers.<\/li>\n\n\n\n<li>Revise copy and components for different user segments.<\/li>\n\n\n\n<li>Create variants or experiment briefs based on prompts.<\/li>\n\n\n\n<li>Convert natural language requests into simple code or design edits.<\/li>\n\n\n\n<li>Enable brand controls to guide content creation.<\/li>\n<\/ul>\n\n\n\n<p>The obvious benefit is speed. Much of the legwork that went into building an A\/B test can be automated with generative AI.&nbsp;<\/p>\n\n\n\n<p>These capabilities also make it possible for a single person to create tests that would have previously required a copywriter, designer, and developer working together.<\/p>\n\n\n\n<p><strong>Potential risks and limitations:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Testing surface-level variations that don\u2019t produce meaningful outcomes .<\/li>\n\n\n\n<li>Introducing generic \u201cLLM-style\u201d copywriting and images.<\/li>\n\n\n\n<li>Creating variations that optimize for the test at the expense of brand strategy.<\/li>\n\n\n\n<li>Building tests that ignore important context like the funnel stage or buyer personas.<\/li>\n<\/ul>\n\n\n\n<p>By enabling people to do a lot more at a faster pace, there\u2019s a temptation to run more tests rather than better ones. That\u2019s a slippery slope.<\/p>\n\n\n\n<p>I\u2019ve seen claims about AI A\/B testing like, \u201cyou can set up experiments in five minutes!\u201d&nbsp;<\/p>\n\n\n\n<p>That may be true, but the old saying \u201cgarbage in, garbage out,\u201d applies. <a href=\"https:\/\/www.crazyegg.com\/blog\/ai-marketing-assets\/\" target=\"_blank\" rel=\"noreferrer noopener\">Humans hate AI-generated marketing content<\/a>, and letting AI drive the full creative process is likely to lead to poor outcomes.<\/p>\n\n\n\n<p>These capabilities work best when AI handles the manual work and humans handle the thinking. The real value is giving writers, strategists, and designers more time to reinvest in the big-picture goals of the experiment.&nbsp;<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">2. Assist with the design and configuration of experiments<\/h3>\n\n\n\n<p>Before an A\/B test runs, you have to lock in a few key elements. What is the primary metric that defines success? Are there countervailing metrics we should track so we know we didn\u2019t hurt anything we also care about? How long should the test run?<\/p>\n\n\n\n<p>Traditional A\/B testing tools make it fairly easy to select <a href=\"https:\/\/www.crazyegg.com\/blog\/best-website-kpi\/\" target=\"_blank\" rel=\"noreferrer noopener\">key metrics<\/a>, estimate sample sizes, and project test durations. AI A\/B testing tools can use historical data and pattern recognition to provide additional assistance during set up.<\/p>\n\n\n\n<p><strong>What it can do:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Convert natural language prompts into test setups (e.g. \u201cRun a test on returning users optimizing for revenue per session\u201d).<\/li>\n\n\n\n<li>Recommend primary and countervailing metrics based on your goals, page types, and historical data.<\/li>\n\n\n\n<li>Flag tests with a high probability of being underpowered based on historical data.<\/li>\n<\/ul>\n\n\n\n<p>For teams running lots of experiments, AI tools can help standardize and streamline the setup process. It\u2019s going to be easier to run a greater variety of tests on a greater number of pages.<\/p>\n\n\n\n<p>They can also help less experienced teams avoid basic errors that waste valuable testing time or hurt the <a href=\"https:\/\/www.crazyegg.com\/blog\/what-is-good-conversion-rate\/\" target=\"_blank\" rel=\"noreferrer noopener\">conversion rate<\/a>, leading to lost sales.<\/p>\n\n\n\n<p><strong>Potential risks and limitations:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Flawed recommendations stemming from AI reliance on historical data that no longer applies. (e.g. you changed the page last month)<\/li>\n\n\n\n<li>The platform may select metrics that are easy to measure instead of the ones that reflect your business goal, such as optimizing for clicks instead of revenue.\u00a0<\/li>\n\n\n\n<li>Discouraging creative risk, as AI tools may recommend test designs that have worked before rather than bold experiments.<\/li>\n\n\n\n<li>AI tools may hallucinate, picking metrics or scoring test ideas based on fabricated data.<\/li>\n<\/ul>\n\n\n\n<p>The risk is that people treat the suggestions as authoritative without questioning whether they line up with the goals of the test. Can you understand and verify the rationale behind the recommendation?<\/p>\n\n\n\n<p>Treating the AI recommendations as inputs to be considered rather than accepting them without scrutiny is the safer play. This protects teams from running over-standardized, cookie-cutter experiments, or running with test ideas that have no basis in reality.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">3. Forecast test impact<\/h3>\n\n\n\n<p>Some AI tools can help you estimate the likely impact of an A\/B test idea before it launches by analyzing data from past experiments.<\/p>\n\n\n\n<p>They can take into account things like page type, audience, the type of experiment (<a href=\"https:\/\/www.crazyegg.com\/blog\/headline-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">testing headlines<\/a> vs. pricing changes, for example), and then estimate how the new test might behave.\u00a0<\/p>\n\n\n\n<p>It\u2019s not perfect or prophetic, but it can give you a sense of which tests are the most promising.&nbsp;<\/p>\n\n\n\n<p><strong>What it can do:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Estimate probable conversion rate ranges for new experiments<\/li>\n\n\n\n<li>Rank and prioritize test ideas based on predicted value<\/li>\n\n\n\n<li>Project potential revenue under current traffic conditions<\/li>\n\n\n\n<li>Identify test ideas that have underperformed in similar contexts<\/li>\n<\/ul>\n\n\n\n<p>For teams with a lot of historical experimentation data, these tools can help you make evidence-based decisions about which tests are likely to have a meaningful impact on conversions.&nbsp;<\/p>\n\n\n\n<p><strong>Potential risks and limitations:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Teams without a lot of historical data will not get useful guidance from these tools.<\/li>\n\n\n\n<li>If your traffic mix varies or you\u2019ve updated a lot of content, the predictions will be less accurate.\u00a0<\/li>\n\n\n\n<li>For <a href=\"https:\/\/www.crazyegg.com\/blog\/split-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">split testing<\/a> novel test ideas or big structural changes, forecasting is unlikely to provide reliable guidance.<\/li>\n<\/ul>\n\n\n\n<p>These tools are extrapolating from past data, so where the data is thin or the test idea is really different, you can\u2019t expect the predictions to be useful.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">4. Enhance personalization<\/h3>\n\n\n\n<p>A\/B testing tools that use <a href=\"https:\/\/www.crazyegg.com\/blog\/multi-armed-bandit\/\" target=\"_blank\" rel=\"noreferrer noopener\">multi-armed bandit algorithms<\/a> steer traffic to winners based on performance. They can incorporate user attributes like behavior or demographics into the algorithm that decides which version to show which user.<\/p>\n\n\n\n<p>These tools have been around for years, using AI and machine learning to empower teams to find winners faster, or run continuous multivariate tests that show variations to the segments they perform well with.&nbsp;<\/p>\n\n\n\n<p>So what new value do AI A\/B testing tools add? I often see it described by vendors as \u201chyper personalization,\u201d which enables you to reallocate traffic during tests using a much richer set of behavioral and demographic signals.<\/p>\n\n\n\n<p><strong>What it can do:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Automatically define user segments without manual input.<\/li>\n\n\n\n<li>Draw on multiple data sources simultaneously to decide which variation to show (e.g. CRM records, browsing history, or purchase patterns)<\/li>\n\n\n\n<li>Adapt variations and offers in real-time based on user behavior and <a href=\"https:\/\/www.crazyegg.com\/web-analytics\" target=\"_blank\" rel=\"noreferrer noopener\">website analytics<\/a><\/li>\n\n\n\n<li>Enable micro-segmentation, and with some tools, achieve 1:1 personalization at scale<\/li>\n<\/ul>\n\n\n\n<p>Put simply, the newer tools can ingest a lot more information about your users to make decisions about which variation to show them.<\/p>\n\n\n\n<p>Whereas traditional bandits allocated traffic with the sole goal of <a href=\"https:\/\/www.crazyegg.com\/blog\/increase-your-conversion-rate\/\" target=\"_blank\" rel=\"noreferrer noopener\">increasing your conversion rate<\/a>, an AI A\/B testing tool might consider multiple user attributes and behavior patterns to decide which variation to show.<\/p>\n\n\n\n<p><strong>Potential risks and limitations:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Hyper personalization requires a large volume of creative assets, which is challenging to manage, and fully relying on AI to generate them creates its own risks.<\/li>\n\n\n\n<li>You must have high-quality, up-to-date customer data in a centralized source that the AI tool can access, which is difficult to orchestrate and may raise privacy issues.<\/li>\n\n\n\n<li>Results are harder to interpret because there isn\u2019t a clear winner, and it\u2019s not always clear why particular variations performed well.<\/li>\n\n\n\n<li>Privacy regulations (e.g. <a href=\"https:\/\/gdpr.eu\/\" target=\"_blank\" rel=\"noreferrer noopener\">GDPR<\/a>) may not allow you to collect the data such personalization requires.<\/li>\n<\/ul>\n\n\n\n<p>While AI A\/B testing tools can help your teams personalize tests with a greater degree of precision, there is no guarantee that the wins you find are going to be durable. The more granular the user segments, the harder it is to reach statistically significant sample sizes.<\/p>\n\n\n\n<p>What performs well in a particular moment with a specific user may not provide a true win that you can scale out across your site, which undercuts one of the <a href=\"https:\/\/www.crazyegg.com\/blog\/benefits-of-ab-testing\/\" target=\"_blank\" rel=\"noreferrer noopener\">key benefits of A\/B testing<\/a>.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">5. Interpret and summarize results<\/h3>\n\n\n\n<p>Any decent A\/B testing tool makes it fairly easy to see which version performed better. But there has always been a good deal of manual work when it comes to understanding the quality of the results, digging into the segment-level data, and reporting those results in plain language to stakeholders.<\/p>\n\n\n\n<p>AI A\/B testing tools are phenomenal at synthesizing large volumes of structured data and reporting on what they find.&nbsp;<\/p>\n\n\n\n<p><strong>What it can do:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Generate simple summaries of the test outcomes<\/li>\n\n\n\n<li>Highlight meaningful differences about variation performance<\/li>\n\n\n\n<li>Alert you to traffic segments with usually strong or weak performances<\/li>\n\n\n\n<li>Compare results to past experiments<\/li>\n\n\n\n<li>Suggest follow-up test ideas<\/li>\n\n\n\n<li>Draft shareable experiment reports<\/li>\n<\/ul>\n\n\n\n<p>The benefit here is speed but also thoroughness, as the tools can surface patterns that a busy testing team might miss, especially when the top-line result isn\u2019t statistically significant.&nbsp;<\/p>\n\n\n\n<p>For teams that run dozens of concurrent tests, the assistance with analysis and documentation will eliminate a lot of post-test busywork. And for less experienced teams that are still learning what to look for, these features are even more valuable.<\/p>\n\n\n\n<p><strong>Potential risks and limitations:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Summaries may include misinterpretations, hallucinations, or fabricated findings.<\/li>\n\n\n\n<li>Reports don\u2019t necessarily capture the most meaningful business insights, even if they highlight statistically sound patterns.<\/li>\n\n\n\n<li>Failure to account for external factors in its analysis, such as seasonality, ongoing marketing campaigns (yours or competitors), or big industry news.<\/li>\n<\/ul>\n\n\n\n<p>As long as you keep a human \u201cin the loop\u201d when it comes to analysis, most of these potential risks can be avoided. For example, I would double-check results that fall under <a href=\"https:\/\/www.linkedin.com\/pulse\/twymans-law-ronny-kohavi\" target=\"_blank\" rel=\"noreferrer noopener\">Twyman\u2019s Law<\/a> before forwarding an AI-generated experiment report to a stakeholder who will never see the data first-hand.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Emerging AI A\/B Testing Capabilities<\/h2>\n\n\n\n<p>The abilities we just covered are all here today, assisting teams with experimental work on live pages.&nbsp;<\/p>\n\n\n\n<p>What we\u2019ll look at in this section are two emerging capabilities used by frontier experimentation teams, being studied by researchers, and on product roadmaps.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Simulating tests with synthetic users<\/h3>\n\n\n\n<p>What if: instead of running live traffic through your test, you showed it first to AI agents who could interact with the page variations?<\/p>\n\n\n\n<p>Live testing is expensive, time-consuming, and potentially very risky. If you test a new idea on a high traffic site and it completely bombs, you could lose out on significant revenue.<\/p>\n\n\n\n<p>The basic premise for using AI agents instead of live traffic to simulate a test is that it is relatively cheap, very fast, and low-risk.<\/p>\n\n\n\n<p>Under controlled conditions, advanced agents are capable of engaging with websites, entering information into forms, and completing multi-step flows.&nbsp;<\/p>\n\n\n\n<p>These \u201csynthetic users\u201d can also be trained on data from specific <a href=\"https:\/\/www.crazyegg.com\/blog\/buyer-persona\/\" target=\"_blank\" rel=\"noreferrer noopener\">buyer personas<\/a> so that their website behavior is aligned with the real users or shoppers you want to study.\u00a0<\/p>\n\n\n\n<p>Running a simulated test can help you:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Generate insights on pages with limited traffic<\/li>\n\n\n\n<li>Uncover bugs in the design<\/li>\n\n\n\n<li>Identify points of friction in flows<\/li>\n\n\n\n<li>Compare simulated behavior across multiple design options<\/li>\n<\/ul>\n\n\n\n<p>And you would get these insights without ever having to expose real traffic to the variations you want to test.<\/p>\n\n\n\n<p>These capabilities do not fully replace actually running a controlled experiment. No one working on tools for simulated A\/B testing tools claimed that it does.<\/p>\n\n\n\n<p>But they do provide recommendations that can help you prioritize tests and plan ahead. <a href=\"https:\/\/arxiv.org\/html\/2504.09723v3\" target=\"_blank\" rel=\"noreferrer noopener\">Recent research on simulated testing using LLM agents<\/a> showed that, under controlled conditions, these models can pick up on real behavior patterns.\u00a0<\/p>\n\n\n\n<p>The authors of that study write, \u201cOur position is that LLM agents <strong>should not replace real user testing<\/strong>\u201d (emphasis original), but do argue that they can be useful for getting quick, low-risk feedback before running full experiments.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Fully autonomous AI experimentation<\/h3>\n\n\n\n<p>We\u2019ve touched on a few of the different parts that AI A\/B testing tools can automate, but what would it look like to have AI agents take over the entire testing lifecycle?<\/p>\n\n\n\n<p>Some of the leading A\/B testing platforms are starting to offer tools that get close to fully autonomous AI experimentation, and there are some newer AI-native tools working towards this goal as well.<\/p>\n\n\n\n<p>I don\u2019t think it will be long before we have AI agents:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Crawling your site to find pages worth testing<\/li>\n\n\n\n<li>Developing their own hypotheses based on page performance and brand aligned goals<\/li>\n\n\n\n<li>Generating content and code to create variations<\/li>\n\n\n\n<li>Setting up and running tests<\/li>\n\n\n\n<li>Monitoring performance and shifting traffic in real-time according to conversion metrics<\/li>\n\n\n\n<li>Personalizing experiences for specific segments or even individuals<\/li>\n\n\n\n<li>Stopping the test once it hits statistical significance<\/li>\n\n\n\n<li>Generating post-test analysis and reporting key findings<\/li>\n\n\n\n<li>Creating variations for follow-up experiments based on results<\/li>\n\n\n\n<li>Repeating this process over and over, continuously optimizing a site<\/li>\n<\/ul>\n\n\n\n<p>How far are we away from AI-led A\/B testing like this?&nbsp;<\/p>\n\n\n\n<p>Closer than you might think.<\/p>\n\n\n\n<p>Tools today are already capable of automating each step of this process, so what stands between us and fully autonomous testing is a matter of chaining together existing capabilities.<\/p>\n\n\n\n<p>And there are already teams running multi-armed bandit tests for continuous learning, letting ML algorithms shift traffic towards winning variations more or less indefinitely.&nbsp;<\/p>\n\n\n\n<p>But of course there is still human oversight of these processes, something that I don\u2019t see going away anytime soon.&nbsp;<\/p>\n\n\n\n<p>Business ethics, brand strategy, and market forces play a leading role in whether or not a test makes sense to run or the outcomes are desirable.&nbsp;<\/p>\n\n\n\n<p>While you can train agents to understand your strategy and brand, the risk posed by removing human supervision from the full experiment lifecycle is serious. It\u2019s not hard to imagine a scenario where an AI agent generates content that converts very well but harms a brand\u2019s reputation or bottom line.<\/p>\n\n\n\n<p>My main takeaway from researching and thinking about AI A\/B testing tools is that they can help intelligent, hard-working humans get more done. Where they can automate tedious tasks, fantastic, but they have a long way to go before replacing the talented teams that use them.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>A\/B testing tools were some of the first software to incorporate AI into their platforms in a meaningful way. And they did it long before&#8230;<\/p>\n","protected":false},"author":254,"featured_media":108100,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_eb_attr":"","_lmt_disableupdate":"","_lmt_disable":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[4],"tags":[],"class_list":["post-108098","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-conversion-optimization"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.5 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>What Can AI AB Testing Tools Do? Benefits and Limits<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"What Can AI AB Testing Tools Do? Benefits and Limits\" \/>\n<meta property=\"og:description\" content=\"A\/B testing tools were some of the first software to incorporate AI into their platforms in a meaningful way. And they did it long before...\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/\" \/>\n<meta property=\"og:site_name\" content=\"The Daily Egg\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/crazyegganalytics\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-02-27T18:16:22+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-02-27T18:16:23+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"720\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Peter Lowe\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@CrazyEgg\" \/>\n<meta name=\"twitter:site\" content=\"@CrazyEgg\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Peter Lowe\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"16 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/\"},\"author\":{\"name\":\"Peter Lowe\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#\\\/schema\\\/person\\\/526c464e0c3c4513698e8b5823e00989\"},\"headline\":\"What Can AI AB Testing Tools Do? Benefits and Limits\",\"datePublished\":\"2026-02-27T18:16:22+00:00\",\"dateModified\":\"2026-02-27T18:16:23+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/\"},\"wordCount\":3301,\"publisher\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/27101118\\\/AI-AB-Testing_Featured-Image.png\",\"articleSection\":[\"Conversion\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/\",\"url\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/\",\"name\":\"What Can AI AB Testing Tools Do? Benefits and Limits\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/27101118\\\/AI-AB-Testing_Featured-Image.png\",\"datePublished\":\"2026-02-27T18:16:22+00:00\",\"dateModified\":\"2026-02-27T18:16:23+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#primaryimage\",\"url\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/27101118\\\/AI-AB-Testing_Featured-Image.png\",\"contentUrl\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2026\\\/02\\\/27101118\\\/AI-AB-Testing_Featured-Image.png\",\"width\":1200,\"height\":720,\"caption\":\"AI AB testing workflow diagram showing five key capabilities: idea generation, experimental design, impact forecasting, personalization, and results interpretation.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/ai-ab-testing\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Blog\",\"item\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Conversion\",\"item\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/category\\\/conversion-optimization\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"What Can AI AB Testing Tools Do? Benefits and Limits\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/\",\"name\":\"The Daily Egg\",\"description\":\"Conversion Rate Optimization Made Easy\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#organization\",\"name\":\"Crazy Egg\",\"url\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2015\\\/06\\\/Crazy-Egg-logo-small.png\",\"contentUrl\":\"https:\\\/\\\/ceblog.s3.amazonaws.com\\\/wp-content\\\/uploads\\\/2015\\\/06\\\/Crazy-Egg-logo-small.png\",\"width\":191,\"height\":100,\"caption\":\"Crazy Egg\"},\"image\":{\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/crazyegganalytics\\\/\",\"https:\\\/\\\/x.com\\\/CrazyEgg\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/crazy-egg\\\/\",\"https:\\\/\\\/www.youtube.com\\\/channel\\\/UCJNe_xmPi07YezxaqfoRVqg\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/#\\\/schema\\\/person\\\/526c464e0c3c4513698e8b5823e00989\",\"name\":\"Peter Lowe\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g\",\"caption\":\"Peter Lowe\"},\"description\":\"Peter works in marketing and lead generation. He has been involved on the backend of the Crazy Egg blog since 2020. Occasionally, they let him write a post.\",\"url\":\"https:\\\/\\\/www.crazyegg.com\\\/blog\\\/author\\\/peter\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"What Can AI AB Testing Tools Do? Benefits and Limits","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/","og_locale":"en_US","og_type":"article","og_title":"What Can AI AB Testing Tools Do? Benefits and Limits","og_description":"A\/B testing tools were some of the first software to incorporate AI into their platforms in a meaningful way. And they did it long before...","og_url":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/","og_site_name":"The Daily Egg","article_publisher":"https:\/\/www.facebook.com\/crazyegganalytics\/","article_published_time":"2026-02-27T18:16:22+00:00","article_modified_time":"2026-02-27T18:16:23+00:00","og_image":[{"width":1200,"height":720,"url":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png","type":"image\/png"}],"author":"Peter Lowe","twitter_card":"summary_large_image","twitter_creator":"@CrazyEgg","twitter_site":"@CrazyEgg","twitter_misc":{"Written by":"Peter Lowe","Est. reading time":"16 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#article","isPartOf":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/"},"author":{"name":"Peter Lowe","@id":"https:\/\/www.crazyegg.com\/blog\/#\/schema\/person\/526c464e0c3c4513698e8b5823e00989"},"headline":"What Can AI AB Testing Tools Do? Benefits and Limits","datePublished":"2026-02-27T18:16:22+00:00","dateModified":"2026-02-27T18:16:23+00:00","mainEntityOfPage":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/"},"wordCount":3301,"publisher":{"@id":"https:\/\/www.crazyegg.com\/blog\/#organization"},"image":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#primaryimage"},"thumbnailUrl":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png","articleSection":["Conversion"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/","url":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/","name":"What Can AI AB Testing Tools Do? Benefits and Limits","isPartOf":{"@id":"https:\/\/www.crazyegg.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#primaryimage"},"image":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#primaryimage"},"thumbnailUrl":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png","datePublished":"2026-02-27T18:16:22+00:00","dateModified":"2026-02-27T18:16:23+00:00","breadcrumb":{"@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#primaryimage","url":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png","contentUrl":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2026\/02\/27101118\/AI-AB-Testing_Featured-Image.png","width":1200,"height":720,"caption":"AI AB testing workflow diagram showing five key capabilities: idea generation, experimental design, impact forecasting, personalization, and results interpretation."},{"@type":"BreadcrumbList","@id":"https:\/\/www.crazyegg.com\/blog\/ai-ab-testing\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Blog","item":"https:\/\/www.crazyegg.com\/blog\/"},{"@type":"ListItem","position":2,"name":"Conversion","item":"https:\/\/www.crazyegg.com\/blog\/category\/conversion-optimization\/"},{"@type":"ListItem","position":3,"name":"What Can AI AB Testing Tools Do? Benefits and Limits"}]},{"@type":"WebSite","@id":"https:\/\/www.crazyegg.com\/blog\/#website","url":"https:\/\/www.crazyegg.com\/blog\/","name":"The Daily Egg","description":"Conversion Rate Optimization Made Easy","publisher":{"@id":"https:\/\/www.crazyegg.com\/blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.crazyegg.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.crazyegg.com\/blog\/#organization","name":"Crazy Egg","url":"https:\/\/www.crazyegg.com\/blog\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.crazyegg.com\/blog\/#\/schema\/logo\/image\/","url":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2015\/06\/Crazy-Egg-logo-small.png","contentUrl":"https:\/\/ceblog.s3.amazonaws.com\/wp-content\/uploads\/2015\/06\/Crazy-Egg-logo-small.png","width":191,"height":100,"caption":"Crazy Egg"},"image":{"@id":"https:\/\/www.crazyegg.com\/blog\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/crazyegganalytics\/","https:\/\/x.com\/CrazyEgg","https:\/\/www.linkedin.com\/company\/crazy-egg\/","https:\/\/www.youtube.com\/channel\/UCJNe_xmPi07YezxaqfoRVqg"]},{"@type":"Person","@id":"https:\/\/www.crazyegg.com\/blog\/#\/schema\/person\/526c464e0c3c4513698e8b5823e00989","name":"Peter Lowe","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/ee86f41fc6f31ebf4977c622dc46568df1d0ae58dd64405ab97403cb37a04c4e?s=96&d=mm&r=g","caption":"Peter Lowe"},"description":"Peter works in marketing and lead generation. He has been involved on the backend of the Crazy Egg blog since 2020. Occasionally, they let him write a post.","url":"https:\/\/www.crazyegg.com\/blog\/author\/peter\/"}]}},"modified_by":"Lauren Knoll","_links":{"self":[{"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/posts\/108098","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/users\/254"}],"replies":[{"embeddable":true,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/comments?post=108098"}],"version-history":[{"count":0,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/posts\/108098\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/media\/108100"}],"wp:attachment":[{"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/media?parent=108098"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/categories?post=108098"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.crazyegg.com\/blog\/wp-json\/wp\/v2\/tags?post=108098"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}