The best conversion rate optimization software? The assistants had no answer.
For the most direct question we asked,“best conversion rate optimization software”,ChatGPT, Gemini, and Perplexity all returned a readable answer, but none of them named a single product. That was the only query that drew a blank on recommendations out of six we ran. The other five questions each listed between six and eight tools, and overall, 6 of 6 answers came back complete. The gap in the most obvious query is the first thing a vendor in this space needs to know: even when the assistants talk about your category, they can loop right past your product.
Across the remaining five prompts, Crazy Egg and VWO led with 4 appearances each. Hotjar, Microsoft Clarity, Optimizely, and Unbounce followed at 3. Google Analytics turned up twice, and another twenty products showed up only once, including AB Tasty, Amplitude, Contentsquare, FullStory, GA4, Heap, Instapage, Jotform, Kameleoon, Lucky Orange, Mixpanel, Mouseflow, Personizely, Perspective, Plerdy, and Smartlook. The sheer number of single-mention names is worth noticing. Only two tools broke out of the pack, and they still missed a third of the questions.
The surprise runs deeper when you look at the sources. Three domains fed every answer we got: lyssna.com, perspective.co, and wisepops.com. Every single recommendation from every assistant on every question passed through at least one of those sites. monday.com and unbounce.com fed 5 of 6 answers. Together, those five domains supply the vast majority of the material the assistants lean on. If your product page or a profile on one of those sites isn’t part of the picture, you are effectively invisible to these models in this category right now. That narrow sourcing also explains some of the naming chaos: Google Analytics, GA4, and Google Analytics 4 appear as separate entries because the assistants pulled inconsistent naming from different pages.
What we cannot explain is why the plainest query produced zero recommendations. The prompt “best conversion rate optimization software” triggered answers that talked around the category without naming a tool, while prompts like “best conversion rate optimization software for small teams” and “most affordable conversion rate optimization software” delivered full lists. We saw the same assistants handle nuanced prompts with specifics and then choke on the straightforward one. That is not a fluke of a single assistant; it happened across ChatGPT, Gemini, and Perplexity. For a product manager scanning this data, the lesson is that generic category prompts may not bring your name up at all, even if you appear in longer-tail queries.
Crazy Egg and VWO share the top spot on mentions, but neither of their own domains appeared among the frequently cited sources. The assistants found them through aggregator roundups and listicles. That makes the recommendation order fragile: a shift in one or two of those high-frequency source pages could reshuffle the whole list next month. For a seller not named here, the question is less about winning direct mentions and more about getting onto at least one of the domains that already feed all six answers.
Who got named, and how often
Crazy Egg and VWO each appeared in 4 of the 6 answers. That is as close to a default as any tool came in this measurement. No product reached 5 or 6 mentions, and the most direct question,“best conversion rate optimization software”,returned zero product names at all. So even the 4-of-6 leaders were absent from one of the five queries that actually named tools. For a seller, landing in 4 of 6 means you are visible more often than not, but you still miss two out of six chances to be seen.
Hotjar, Microsoft Clarity, Optimizely, and Unbounce followed at 3 of 6. Hotjar and Microsoft Clarity both appeared in the “best free” question, which is not surprising given Clarity’s free-only model and Hotjar’s generous free tier. Optimizely showed up across a mix of paid and free questions, and Unbounce appeared both in the small-teams query and in the free and affordable lists, stretching a landing-page focus across CRO contexts. Three mentions is not dominance, but it is a presence that small shuffles in the underlying sources could easily turn into four.
Google Analytics alone had 2 mentions, then GA4 and Google Analytics 4 each appeared once. Whether the assistants treat these as the same product or as distinct is unclear from the data, but the fragmentation matters. A vendor counting on “Google Analytics” to be a steady entry sees it named in two answers, while the more specific GA4 string pulls only one. Sellers that offer free analytics alongside CRO features might look at this and ask why Clarity got three mentions while the broader Google Analytics ecosystem only reached four across three separate name forms.
The long tail is where the story flips. Eighteen products appeared in exactly one answer. Amplitude, Contentsquare, FullStory, Heap, Glassbox, Kameleoon, and Mixpanel are all serious CRO or analytics platforms that many buyers would expect to see mentioned repeatedly. Instead, each got a single call. FullStory, for instance, emerged only from the “recommendations” prompt, alongside Heap and Kameleoon. AB Tasty and Lucky Orange got one mention apiece as well. Smartlook, Mouseflow, Plerdy, Jotform, Instapage, and Personizely fill out the same single-appearance column. A product named once in six runs is a coin flip. It does not suggest the assistant has chosen it; it signals that a single source page happened to list it, and the assistant grabbed that page for that one query.
What is absent here is almost as loud as what is present. Plenty of known CRO tools,Convert, Omniconvert, SiteGainer, Zarget, and others,did not surface at all. That does not mean they are not in the category; it means our six questions, drawing from the specific source pages the assistants pulled, did not encounter them. A vendor not on this list should not conclude they are invisible to AI overviews. They should conclude that our particular measurement did not catch them, and that running this again next week or with a different prompt wording might well surface them. The list of sources gives a hint: if a tool’s name never appears on perspective.co, wisepops.com, lyssna.com, or unbounce.com, it will struggle to break through, because those four domains fed all six answers. Perspective the product even got a mention itself in one answer, despite the fact that perspective.co as a source was present in every single result. The source and the product share a domain, yet the product barely registered.
For a product trying to move from one mention to three or four, the mechanism is not mysterious. The assistants lean on a small set of articles. Crazy Egg and VWO show up because they are fixtures on those articles. A CRO platform that gets itself included on the high-frequency source pages could shift from the long tail into the small group of twice-or-thrice-named tools in a single content cycle. But the chasm between one mention and four is wide, and no product in this measurement managed to cross it entirely, because even the leaders could not get into the most common question of all.
The full list, counted
Every product serpapi+or:chatgpt,gemini,perplexity named across 6 answers, with how many of those answers named it. The share column is out of the answers we read, not out of the category.
| Product | Named in | Share | Read as |
|---|---|---|---|
| Crazy Egg | 4 of 6 | 67% | 4 answers |
| VWO | 4 of 6 | 67% | 4 answers |
| Hotjar | 3 of 6 | 50% | 3 answers |
| Microsoft Clarity | 3 of 6 | 50% | 3 answers |
| Optimizely | 3 of 6 | 50% | 3 answers |
| Unbounce | 3 of 6 | 50% | 3 answers |
| Google Analytics | 2 of 6 | 33% | 2 answers |
| AB Tasty | 1 of 6 | 17% | one answer only |
| Amplitude | 1 of 6 | 17% | one answer only |
| Contentsquare | 1 of 6 | 17% | one answer only |
The source pages that decided the category
Three domains appear in all six answers: perspective.co, lyssna.com, and wisepops.com. Two more, monday.com and unbounce.com, missed only a single response each. No independent review platform reached even half the queries. The pattern is clear before you look deeper: the assistants built their recommendations almost entirely from pages published by software vendors themselves.
perspective.co is the product site for Perspective, a tool that itself got named in one answer. lyssna.com belongs to Lyssna, a user research and usability testing platform. wisepops.com is the domain of Wisepops, a popup and personalisation tool. All three run content programmes that produce comparison posts and “best CRO software” listicles. Their presence in every single answer says that the retrieval step inside ChatGPT, Gemini and Perplexity latched onto those pages and kept returning to them, regardless of the prompt phrasing.
monday.com reached 5 of 6 answers despite not selling a CRO tool. It sells work management software and publishes a blog that ranks for a wide set of software-category queries. unbounce.com, the landing page builder, also hit 5 of 6. Its content library includes several pieces positioning Unbounce next to other optimisation tools. The assistants clearly treated these sites as authoritative sources on what to recommend, even when the topic falls outside the publisher’s own product.
further out, flippingbook.com, knock-ai.com and personizely.net each fed 3 answers. flippingbook is a digital flipbook maker; knock-ai.com and personizely.net are both vendor domains for products that sit near CRO (personalisation, AI-driven conversion). optimizely.com and whatconverts.com showed up in 2 answers each. The first is a major experimentation platform, the second a call tracking vendor. Then come a set of domains that contributed to only a single answer: beomniscient.com, campaigncreators.com, conversionrate.store, hubspot.com, lancasterchamber.com and trustradius.com.
One review site appears: trustradius.com, and only once. No G2, no Capterra, no Trustpilot, no Reddit thread. The one review platform that surfaced is also the one least like a community-driven aggregator; TrustRadius has always leaned into analyst-style buyer guides, which makes it more similar to the vendor-authored listicle format than to a user review site. lancasterchamber.com, a local chamber of commerce, is the solitary domain I cannot explain. It does not obviously relate to CRO software and appears to have been pulled in by a stray mention or a ranking anomaly.
What this mix says about how a recommendation gets made here matters more than any individual count. The assistants are not comparing product pages or parsing user sentiment. They are summarising listicles and comparison posts, and the bulk of those posts come from companies that sell something else in the neighbourhood of conversion optimisation. That dynamic makes the category self-referential in a narrow way: being named depends on getting included in the content output of a handful of adjacent vendors. For a CRO tool that is not already in those posts, the path to visibility runs through those publishers. For one that is, the risk is that a change to a single high-frequency listicle can drop it out of four or five answers at once. The measurement shows no evidence that the assistants are pulling from product documentation, support forums or customer testimonials directly. They lean on a small set of long-form blog pages, and the blogs belong to tools that do not even compete head-to-head with the CRO tools they are recommending.
The source list decides the shortlist
Crazy Egg and VWO each appeared in four of six answers not because the assistants independently assessed feature sets or customer reviews, but because the pages the assistants leaned on carried them. Three domains appeared in every single answer: perspective.co, lyssna.com, and wisepops.com. None of those three is a comprehensive CRO software review site. perspective.co belongs to Perspective, a tool that itself was named only once in the whole measurement. lyssna.com is the home of Lyssna, a user research and usability testing platform. wisepops.com belongs to Wisepops, a popup and personalisation company. Two more domains, monday.com and unbounce.com, were absent from only one answer each. Across six queries and a total of 24 distinct product names, the list of sources was dominated by vendor content marketing.
What a tool like Crazy Egg needs to do to get named four times, then, is appear consistently on the lists these domains publish. We cannot see which specific page surfaced which specific product, but the 4-of-6 names line up with sources that are themselves in the CRO orbit and running comparison posts. The “best free conversion rate optimization software” query, for example, named Microsoft Clarity, Google Analytics, Lucky Orange, VWO, Hotjar, Optimizely, AB Tasty, and Unbounce. Its sources included hubspot.com and lancasterchamber.com alongside the usual vendor blogs. AB Tasty only turned up in that one answer, suggesting that hubspot.com’s page on free tools was the sole route in for that product. AB Tasty never appeared in answers where hubspot.com was not a source.
The pattern repeats for the long tail of products mentioned just once. Plerdy, Smartlook, Personizely, FullStory, GA4, Heap, Instapage, Jotform, Kameleoon, Mouseflow, Glassbox, Amplitude, Contentsquare, Mixpanel, Lucky Orange , all of these surface only when a query pulls in a source domain where they happen to have a slot. For “most affordable conversion rate optimization software,” Plerdy appears alongside sources like campaigncreators.com and conversionrate.store. For “conversion rate optimization software recommendations,” FullStory, GA4, Heap, Instapage, Jotform, and Kameleoon appear, and the source list includes whatconverts.com and knock-ai.com. Remove those niche publishers from the source list, and those products vanish. The assistants are not assembling an independent market view; they are remixing the comparison pages they can reach.
Unbounce’s own domain, unbounce.com, fed five of six answers, and Unbounce the product was named in three. This is a straightforward case of a vendor using its own site to stay in the conversation. What is harder to explain is why Perspective, whose domain also fed all six answers, got named only once , in the “small teams” query. Either perspective.co does not list itself on its own CRO tool round-ups as aggressively as Unbounce does, or the assistant parsing did not surface Perspective even when it was mentioned. Without the raw answer text and the exact page versions the assistants retrieved, we cannot distinguish between these possibilities. A vendor running a high-frequency source domain does not guarantee product mentions; it guarantees that someone else’s shortlist, filtered through that domain’s editorial choices, is what the assistant will regurgitate.
The same asymmetry shows up in the wildcard sources. monday.com fed five answers but sells work management, not CRO software, so it has no product name in the output. Its inclusion likely means that a monday.com blog post comparing CRO tools was among the top results for multiple queries. Every named product that makes it into such a post gets a free ride across several assistant answers, and every product omitted from that post stays invisible regardless of its market position. The one-answer wonders are not necessarily worse tools than the four-answer leaders; they are simply absent from the three to five pages that drive most of the assistant output.
What this measurement cannot tell us is whether the assistants picked a single page per query and listed whatever it contained, or whether they blended multiple source pages. The source lists for each query suggest blending because no single answer drew from exactly one domain. For “what conversion rate optimization software should I use,” the named products included Optimizely, Unbounce, Contentsquare, Crazy Egg, VWO, Mouseflow, Amplitude, and Glassbox, and the sources spanned perspective.co, unbounce.com, wisepops.com, lyssna.com, optimizely.com, knock-ai.com, and flippingbook.com. It is plausible that multiple comparison posts were combined, which would explain why a product like Mouseflow, which likely appears on knock-ai.com’s list but not on the others, made the cut only here. Still, without access to the assistant’s retrieval logs, the exact mechanism is conjecture. The practical implication for a vendor is unchanged: if your product is missing from the sites that serve as the assistants’ de facto editorial board, you will be missing from the recommendations, no matter how strong your feature set.
Your spot in the assistant’s answer rests on three domains
If you sell a product that got named here, the numbers look reassuring. Four mentions out of six queries. That was the ceiling, and only Crazy Egg and VWO reached it. You might read that and think you own a durable piece of the CRO category inside these assistants. You do not. You are defending a position that exists because a handful of vendor blogs decided to include you, and that inclusion can vanish without warning.
Get specific. Three domains appeared in every single answer: perspective.co, lyssna.com, and wisepops.com. Not one of those is a neutral review platform. perspective.co is the website for Perspective, a tool that itself scraped one mention across the entire measurement. lyssna.com belongs to Lyssna, a user testing platform. wisepops.com is Wisepops, a popup builder. These domains are not running exhaustive market analyses. They are publishing content that happens to make the assistant’s cut right now. If your product got named in four of six answers, chances are high you appear on at least two of those three sites. That is your entire moat in this channel. Remove you from one of those pages and you drop to three mentions or fewer. Remove you from two and you go quiet in half the prompts.
The position is less safe than it looks because the “best conversion rate optimization software” query, the one any buyer types first, returned zero product names. Not one. You could be Crazy Egg with a 4-of-6 score and still be invisible on the highest-intent question in the set. That query alone should terrify any vendor feeling comfortable. A single well-optimized page that the assistants begin to trust, perhaps from a domain like trustradius.com (which already fed one answer) or a new entrant, could populate that empty slot with a list that excludes you entirely. Once that happens, a competitor who was nowhere suddenly owns the top-of-funnel assistant traffic while you are stuck answering only longer-tail variations.
What would cost you the position you have now? An editorial refresh on perspective.co that swaps your tool for a sponsor. A Lysna blog post that drops the CRO listicle format altogether. A shift in Perplexity’s source weighting that elevates a competing list from hubspot.com or optimizely.com, both of which appeared in the source pool but never dominated. The measurement shows that even a single-feature tool like Plerdy or a lightweight heatmapper like Lucky Orange can surface when the source pages emphasize affordability or free tiers. Your current mentions do not reflect a deep product evaluation. They reflect the shelf life of a few blog posts that the assistants cached. When those posts change or get outranked, the assistant’s answer changes with them.
The real threat is the absence of institutional knowledge. The assistants showed no ability to name a canonical leader. They stitched together lists from pages that covered wildly different angles: small teams, free tools, affordabilty, generic recommendations. That means your current visibility is pure surface area. The second a CRO company with a content team secures a mention on all three of the omnipresent domains and also pushes a dedicated “best conversion rate optimization software” page into the assistant’s index, you lose not just a mention but the entire category’s default answer. Defend your spot on perspective.co, lyssna.com, and wisepops.com with your life, but do not assume that defense alone keeps you visible. The unclaimed “best” query is the gap someone will fill, and when they do, the assistant’s picture of the market will shift for every question.
If you sell here and were not named, the fix is a partner list
The assistants did not leave you out because they compared your feature set and found you lacking. They did not evaluate your uptime, your pricing, or your NPS. They left you out because you did not appear on the five or six pages that really matter for this category, and the only way to change that is to show up on those pages. Every product named in this measurement, from the four‑answer leaders to the single‑mention long tail, owes its visibility to a handful of vendor‑published listicles and comparison posts. If your product is absent, the work ahead is not a re‑architecture of your site for crawlers or a prompt‑engineering trick. It is a piece of partnership outreach to the owners of a known set of URLs.
The measurement makes that set unusually transparent. Three domains fed every single answer: perspective.co, lyssna.com, and wisepops.com. Two more, monday.com and unbounce.com, missed only one answer. A small group of additional blogs,flippingbook.com, knock‑ai.com, personizely.net, optimizely.com, whatconverts.com,fed between one and three answers and collectively filled gaps in specific queries. None of these are independent review platforms or analyst reports. They are content hubs run by software companies that sell adjacent, complementary, or even slightly competing products. They publish articles with titles like “Best conversion rate optimization software” to capture search traffic for the same questions we asked the assistants, and their list of recommended tools is an editorial choice, not an automated aggregation.
Getting into those articles means contacting the people who maintain them. The task is straightforward to describe and unglamorous to execute. You find the specific pages on perspective.co, lyssna.com, wisepops.com, monday.com, and unbounce.com that rank for the “best CRO software” queries you care about. You check if those pages have a regular update cadence, an author name, or a visible submissions process. Then you make your case: why your tool belongs in that particular list, what gap you fill that the current selections miss, and what you can offer in return,a reciprocal mention, a data point, a written blurb they can drop in without heavy editing. Sometimes it is as simple as asking the partnership contact at monday.com or Unbounce if their content team accepts nominations for list refreshes. Sometimes it requires waiting until the post is next refreshed and being ready with your pitch.
What this does not involve is anything resembling AI‑specific optimisation. You are not submitting a sitemap to an assistant. You are not writing a prompt‑friendly product description in the hope that ChatGPT or Gemini will synthesise a flattering summary of your tool from thin air. The assistants are surfacing brands because they appear on pages that already rank in the underlying search indices and retrieval pipelines. Change the source page, and you eventually change the answer. The “eventually” matters: a mention added today may not appear in the next answer a user gets, because the assistant may be working from a cached version or a retrieval snapshot that updates on its own timeline. But over a measurement horizon of weeks, being added to perspective.co’s list, for instance, would almost certainly pull a product name into answers for “best conversion rate optimization software” and the other queries where that page appears.
The more uncomfortable reality is that the owners of those source pages have little incentive to add a new tool unless it strengthens their article. perspective.co, the domain of the user‑research tool Perspective, appeared in every single answer yet mentioned its own product only once. Its page exists to pull in readers looking for CRO tools and then introduce them to Perspective’s own offering. Lyssna and Wisepops are doing the same thing. If your product makes their list look more credible or more comprehensive without cannibalising their own position, you have a chance. If your product competes directly with their core offering, you do not. That means the path in is narrowest exactly where your tool might overlap with the publishing vendor’s own product, and widest where you serve a distinct sub‑angle they want to cover,maybe free tools, A/B testing specifically, or heatmapping add‑ons.
For the vendors that landed one mention, like AB Tasty, Contentsquare, or Glassbox, the pattern is just as clear. Their single appearance came from a specific source page. AB Tasty appeared only in the “best free” answer, which drew from hubspot.com among others. If a vendor wants to go from one mention to three, they do not need to “improve their AI visibility” in the abstract. They need to get their name onto the 6-of-6 pages,particularly perspective.co, lyssna.com, and wisepops.com,and then maintain that placement through future updates. That is partner‑facing content work, not technical SEO. The assistants are reading the same pages that any search‑savvy buyer would read; they just do it faster and with less discernment. If you are not listed, the mechanism to fix it is as old as the web: get yourself on the list.
How to replicate this for your own category by hand this afternoon
The six questions we used took under an hour to ask across ChatGPT, Gemini, and Perplexity. The counting that came after took four times that long. If you want to run the same measurement for your own product category, you will spend most of the afternoon not querying the assistants but staring at a spreadsheet, trying to decide whether “Crazy Egg” was named as a product or just mentioned in a blog’s footer link.
Open three browser tabs: one for each assistant. Pick six questions that mix broad and narrow intent, exactly the way we did. You need at least one blank-slate query like “best [category] software” and one that adds a modifier like “for small teams” or “most affordable.” Paste each question into all three tabs, one at a time. Copy every answer in full into a single document. Do not summarise or skip the source links at the bottom. That is the part you will hate later, because Perplexity lists sources inline, Gemini appends them as a bulleted tile strip, and ChatGPT varies. You will have to read each answer three times just to pull out the domain names consistently.
Now the real work starts. Highlight every product name you find. This is the tedious part: some assistants weave product mentions into sentences with no formatting, and you have to distinguish a product name from a generic term like “analytics tool.” A single answer might contain eight products buried in two paragraphs. Write each name in a column alongside the question that generated it. If an answer names no products at all, you mark that down too. That is how we discovered that “best conversion rate optimization software” returned zero product recommendations across all three assistants, a gap you would miss if you only skimmed.
Next, extract every source domain linked from each answer. Some answers will reference the same domain multiple times; you count it once per answer. This step is the slowest, because you have to copy URLs, strip them to their root domain, and reconcile variations like “monday.com” versus “www.monday.com.” The payoff is seeing which domains feed all of your questions. In our run, three domains appeared in every single answer, and none of them were impartial review sites. That pattern is invisible until you tabulate the sources by hand.
When you finish, you end up with two grids: one that lists product mentions per question, and one that lists source domains per question. The first grid tells you which products the assistants recommend, and under what phrasing. The second tells you exactly which pages those recommendations came from. The whole process is monotonous and exacting, but that is the value: the monotony forces you to see the answer not as a consumer would, but as the assistant’s retrieval pipeline actually works.