Every list of CRO tools runs to thirty products sorted by affiliate payout. There are only four jobs in the category, and most stores need two of them.
Search for CRO tools and you'll get a list of thirty products with a paragraph each. The lists are long because they're sorted by affiliate payout, and a longer list means more chances you click something that pays.
The bigger problem is that they're flat. Optimizely and Hotjar and GA4 appear as though they're alternatives to each other. They aren't. They do completely different jobs, and putting them in one numbered list is how people end up paying for three tools that answer the same question and none that answer the one they had.
There are four jobs. Most stores need two of them.
The four categories
Analyzers tell you what's wrong. You give them a URL, they come back with specific failures. No traffic needed, no setup, answer in minutes.
Testing tools tell you which version wins. You build two variants, split traffic, wait for significance.
Behavior tools tell you why. Session recordings, heatmaps, rage clicks. You watch a person fail to find the size guide.
Analytics tools tell you what happened. Sessions, conversion rate, revenue, funnel drop-off, which channel paid for it.
Hold those against the question you actually have. "What should I fix this week" is an analyzer question. "Does the new hero beat the old one" is a testing question. Most people buy the testing tool first and then discover they don't have the traffic to use it.
Analyzers
This is the category nobody explains, partly because it's the newest and partly because it's the hardest to sell as a subscription.
PageSpeed Insights and Lighthouse are free, from Google, and genuinely useful. They audit technical performance: Core Web Vitals, render-blocking resources, image sizing. The limitation is that they only see technical problems. A page can score 100 and still have the price below the fold, no scale reference in the photos, and shipping cost hidden until checkout. Lighthouse has no opinion about any of that.
Shopify's own store speed report does a narrower version of the same thing inside your admin.
That's the free tier, and it stops at technical.
We build Qosmic, so weigh this accordingly. It's an analyzer in the wider sense: it reads your live storefront the way a buyer does, on mobile and desktop, and checks it against five hundred conversion criteria with a specific pass bar for each, then ranks what it finds by what each failure is costing you. Then it does the part analyzers historically don't, which is build the fix inside your own theme using your own components and stage it as a draft you approve.
The reason this category stayed thin for so long is that detection is easy to sell and hard to act on. A report that says "your add-to-cart is 32 pixels" is worth something only if somebody changes it, and the gap between finding and fixing is where most CRO programs quietly die.
Testing tools
Shopify's native A/B testing changed this category in June 2026. Splitting traffic between two versions of a theme or a page used to be the whole product for several companies. Now it ships with the platform. If your testing need is "show half my traffic a different version of this page," start here and spend nothing.
Intelligems is the established Shopify-native option and goes further than the split itself, particularly on price and offer testing, which is genuinely hard to do safely. Shoplift ships a library of theme sections you can test against your current ones, which lowers the build cost of a test.
VWO, Optimizely and Convert are the general-web platforms. More statistical machinery, more integrations, more money, and more than most Shopify stores will use.
Two honest limits on this whole category, and they apply to us as much as anyone.
Testing needs traffic that most stores don't have. At a 2% baseline, detecting a 10% relative improvement takes roughly 80,000 sessions per variant at standard confidence and power. A store doing 20,000 sessions a month spends eight months on one test.
And publishing is not the same as splitting. A tool that lets you push a change live is doing something different from a tool that runs a controlled experiment, and the marketing for both uses the word "test."
Qosmic does both halves of this. It writes the variant into your theme using your own components, stages it for approval, and then adjudicates the result against real revenue once it's live. Building the variant and judging it are normally two tools and two subscriptions. Here they're one thing, which is the difference between a test you can actually run and a test you have to assemble first. And if you have 15,000 sessions a month, fixing what's plainly broken will still beat anything a test could tell you this year, which is why the audit comes first.
Behavior tools
We read from these rather than compete with them, so take the recommendations at face value.
Microsoft Clarity is free, with no session cap, and its rage-click and dead-click detection is the fastest way to find a broken interaction. There is no reason not to install it.
Hotjar is the default for recordings, heatmaps and on-page surveys, and the survey half is underrated: asking exiting visitors one question gets you answers no recording will.
FullStory and Lucky Orange cover similar ground, FullStory at the enterprise end with better search across sessions.
The structural limitation of this category is that it doesn't scale. Watching recordings is the highest-quality signal available about why somebody left, and it costs an hour to learn one thing. You'll do it during a crisis and then never again.
Analytics tools
Also integrations rather than competitors. Short section, because this one has its own post.
GA4 and Shopify Analytics are free and enough for most stores under a few thousand sessions a month. The thing worth learning in GA4 is Explorations, where funnel and path analysis live, and almost nobody opens it.
Triple Whale, Polar Analytics and Northbeam are attribution products for stores spending real money on ads. They answer "which channel paid for this," not "what should I change."
That distinction is the whole reason the analyzer category exists. Conversion funnel analytics will tell you that most of your mobile sessions die at the cart. You still have to go and look at the cart to find out why.
What you actually need, by stage
Under about 10,000 sessions a month. An analyzer and Clarity. Both free to start. Don't buy a testing tool; you can't reach significance and the money is better spent on nothing at all.
10,000 to 50,000. Add GA4 properly configured, and Hotjar if you'll actually watch the recordings. Native Shopify testing when you have a genuine coin flip, one at a time.
Above 50,000. Now a real testing platform earns its place, and attribution starts to matter if you're buying traffic. Keep the analyzer running anyway, because tests only ever answer the question you thought to ask.
At any size: two tools you use beat six you pay for. The most common CRO stack problem isn't a missing tool. It's four subscriptions producing dashboards nobody opens.
The short version
Four jobs: analyzers find what's wrong, testing tools settle disagreements, behavior tools explain why, analytics tools report what happened.
Analyzers are the cheapest place to start and the only category that works with no traffic.
Testing is the most oversold. Below roughly 50,000 sessions a month it's mostly theater, and Shopify's native split covers the real cases for free.
Behavior and analytics are both close to free at the bottom, and both are underused rather than missing.
If you have four tools and still can't say what to change on Monday, a fifth won't help.
