Why the ranked list is the wrong shape
These tools do not compete with each other as often as the listicles imply.
An A/B testing platform and a friction detector are not alternatives. One proves a change worked, the other tells you what to change. Ranking them against each other produces an order that is true for nobody, which is why every list of this kind reads as filler.
The decision that actually matters is which question you cannot currently answer. There are six, and they map cleanly onto six kinds of product.
The six categories
Experimentation and A/B testing
“If we change this, does it do better?”
Examples: VWO, Optimizely, Convert, AB Tasty
You already know what to change and you have the traffic to prove it. Most tests need thousands of conversions per variant to say anything, so on a smaller site this is the wrong first purchase.
Heatmaps and click maps
“Where on the page are people clicking?”
Examples: Hotjar, Crazy Egg, Microsoft Clarity
You want a fast visual read of a page. The limit is that a heatmap shows where clicks landed, not whether any of them worked, so a broken button and a popular button look similar.
Session replay
“What did this one visitor actually do?”
Examples: FullStory, LogRocket, Microsoft Clarity
You have a specific bug or a specific complaint to chase. It is excellent for that and poor as a routine habit, because watching sessions does not scale and you find what you happened to watch.
Surveys and voice of customer
“What do people say is wrong?”
Examples: Hotjar, Qualaroo, Typeform
You need intent or motivation, which behaviour cannot give you. Worth pairing with anything on this list, and worth remembering that people explain their behaviour after the fact rather than report it.
Web and product analytics
“How many, from where, and did they convert?”
Examples: GA4, Plausible, Amplitude, Mixpanel
Always. This is the baseline everything else sits on. It tells you what happened with authority and it does not tell you why.
Behavioural friction detection this is us
“What is broken, and how many people did it cost?”
Examples: Harvv
You can see the drop in analytics and cannot see the cause. This category reads the interactions themselves, clicks that did nothing, repeated clicks, hesitation, comparison shopping, and reports the element responsible with a session count.
How to choose, in one paragraph
Start with analytics, because everything else needs a baseline and you probably already have it. Then ask what is stopping you. If you cannot see why a page underperforms, you want friction detection or session replay. If you know why and want to know what people think, you want surveys. If you know what to change and need to prove it, you want testing. Buy in that order, one at a time, and stop when the question is answered.
Where we fit, and where we do not
Harvv is category six and none of the others. We do not do A/B testing, heatmaps, session replay or surveys, and we are not a replacement for your analytics. We read the interactions themselves and report what is broken with the number of sessions it cost: clicks that did nothing, repeated clicks in frustration, hesitation before a form field, visitors leaving mid-visit to compare a price elsewhere.
That is one useful question out of six. If it is not the one you are stuck on, one of the other five is the better purchase, and we would rather say that here than sell you a seat you stop opening.
Questions
What is the best conversion rate optimization tool?
There is no single best one, and any page that gives you a ranked list is ranking six different jobs against each other. The useful question is which job you are stuck on. If you do not know what to change, a testing tool has nothing to test. If you already know and need proof, friction detection will not give you statistical significance. Pick by the question you cannot currently answer.
How many of these do I actually need?
Most working stacks have two or three: analytics as the baseline, one tool that tells you what to fix, and, once you have traffic, something to test the fixes. Buying all six is a common and expensive mistake, because five of them will go unopened after the first month.
Where does Harvv fit?
Behavioural friction detection, and nothing else on this list. We do not do A/B testing, heatmaps, session replay or surveys, and we are not a replacement for your analytics. We answer one question: what on this site is broken, and how many sessions did it cost. If that is not the question you are stuck on, one of the other five is a better purchase.
Do I need A/B testing to improve conversion rate?
Not at the start, and often not for a long time. Testing proves that a change helped, which matters once the obvious problems are gone. If a button does nothing on iOS, you do not need a test to know it should work. Fix the broken things first, then test the judgement calls.
How much does Harvv cost?
Free on 1 site with 50K events a month, which is enough to see whether there is anything worth fixing. Paid plans start at $19 a month, and agencies pay $49 per client site with unlimited sites.
Will another tag slow my site down?
Fair concern, and the reason CRO stacks get abandoned. Ours is 28 KB gzipped and loads asynchronously so it never blocks rendering. It is worth checking the same number for anything else on this list before you add it.