The only builder every assistant named, every time
Carrd was named in all six answers we received, from every assistant we asked, across every variation of the question. We put six different landing page builder queries to ChatGPT, Gemini, and Perplexity: “best landing page builders,” “best landing page builders for small teams,” “what landing page builders should I use,” “best free landing page builders,” “most affordable landing page builders,” and “landing page builders recommendations.” All six returned a readable set of recommendations. No assistant refused, no answer timed out, and no result came back without a list. But the most surprising thing in the data is not which builder dominated. It is that quora.com fed all six answers, across all three assistants, making it the only domain to appear in the source line of every single response. If you sell a landing page builder and you are not consistently visible on Quora, you are effectively absent from the conversation these assistants are pulling from.
Carrd’s clean sweep is the headline statistic: 6 of 6. Landingi came close behind at 5 of 6, missing only the “best free” variant. But after that, the field fractures. HubSpot, Leadpages, and Unbounce each appeared in four answers. Instapage, MailerLite, Swipe Pages, and Wix showed up in three. Hostinger Website Builder, Kit, and Perspective managed two mentions apiece. Six more products (Canva, Elementor, Google Sites, Involve.me, IONOS Website Builder, Mailchimp, and the combination WordPress + Elementor) scraped a single mention, often in a highly specific query. For instance, Google Sites and IONOS Website Builder only surfaced when we asked for free builders, and Kit and Perspective only appeared in the “small teams” answer. Even a product like Mailchimp, which has a free tier, appeared in only one answer , the one explicitly about free options.
The assistants did not all draw from the same well, but the overlap is stark. Zapier.com, emailtooltester.com, and Reddit each informed five of the six answers. Unbounce’s own domain fed three. Then the list falls off a cliff: a dozen domains, from g2.com to youtube.com, fed just one answer. For a product team watching this space, the practical implication is that ChatGPT, Gemini, and Perplexity treat these handful of comparison and discussion sites as the authoritative map of the category. A builder that is not mentioned on Zapier’s listicles, not featured on emailtooltester.com’s comparisons, not debated on Reddit, and not present on Quora will not appear, regardless of the assistant a shopper uses or how they phrase the need.
The “best free” question exposed the narrowness most clearly. The answer pulled from Reddit, emailtooltester.com, sites.google.com, mailerlite.com, wix.com, websiteplanet.com, and quora.com , a tight circle of seven domains , and produced a list (Mailchimp, HubSpot, Wix, MailerLite, WordPress+Elementor, Carrd, Google Sites, IONOS) that is markedly different from the paid-oriented queries. If you are selling a free builder and your presence lives on a platform outside that circle, you were simply not part of the answer.
One perfect score, then a sharp drop-off
Landingi appeared in five of the six answers, missing only the question that asked explicitly for free builders. That single absence tells you a lot about how these assistants handle cost signals: even when a product offers a free tier, the phrase “free” in the query can still push it out of the list. Carrd went six for six, the only builder to do so, but after that the counts fell fast and then flattened into a long tail of names that showed up once and might never show up again.
Four named in all six, four named in none, and a whole middle that depends on exactly how you ask. HubSpot, Leadpages, and Unbounce each hit four mentions. That puts them in the group a shopper is likely to see regardless of wording, but not in the unshakeable position Carrd occupies. The assistant you ask matters here too: Unbounce appeared in answers from all three assistants, though not always on the same questions, while HubSpot and Leadpages were more concentrated in certain phrasings. Perplexity gave HubSpot in five of its six answers, while ChatGPT brought it in for only two. Gemini recommended Leadpages only when the word “recommendations” appeared.
At three mentions, Instapage, MailerLite, Swipe Pages, and Wix form a set where context flips the outcome. Instapage got named in queries about small teams and affordability, but fell out of the most generic “best landing page builders” list. MailerLite, which is not a dedicated landing page builder, got pulled in because of its free and low-cost tiers. Wix appeared when the question sounded like someone starting from scratch (“what should I use”) or when price was the only visible axis, but not when the question implied a team environment. Swipe Pages kept surfacing in the “best” and “affordable” phrasings, but not in the free or small-team variants.
The twice-named group starts to feel fragile. Hostinger Website Builder and Kit each got two nods. Hostinger appeared once in the straightforward “best” list and once in “recommendations,” both times likely driven by content on emailtooltester.com and zapier.com, which fed five of the six answers. Kit showed up only when the question mentioned small teams or the open-ended “should I use,” never when price was the framing. Perspective, a less well-known name, also landed at two, helped by its own site (perspective.co) feeding one query, but that was enough to pull it into two separate assistant outputs.
Then there is the one-and-done group: Canva, Elementor, Google Sites, Involve.me, IONOS Website Builder, Mailchimp, and the combination WordPress + Elementor. These names are a coin flip. Canva appeared only in Gemini’s answer to “what landing page builders should I use,” alongside a source that included canva.com. Google Sites surfaced solely in the free builder question, likely because sites.google.com was in the source list for that query. Mailchimp got named exactly once, in the free builder response from ChatGPT, and nowhere else. If someone asked these same assistants again tomorrow with a slightly different set of crawled pages, any of these names could drop out entirely.
One surprise in the counts is which products failed to appear at all. A reader who follows landing page tools will notice that ClickFunnels never came up in any answer, from any assistant, on any question. The same goes for GetResponse, Systeme.io, and several others that have high brand recognition in this category. That absence does not mean those products are bad or unpopular. It means the specific mix of questions, the indexed pages each assistant pulled, and the phrasing of the queries left them outside this particular sample. Absence from our six questions is not absence from the category, and a marketer running their own search with longer, more feature-specific prompts might see a completely different shortlist.
The assistants themselves did not converge on a single template. ChatGPT leaned on Zapier, emailtooltester.com, and quora.com for most of its recs. Gemini drew heavily from the same pool but was the only assistant to surface Canva, Elementor, and WordPress + Elementor, all in one query. Perplexity was the most consistent at returning a tight list: in five of six answers, it named five or fewer products, never more than seven, while ChatGPT and Gemini occasionally tossed out longer, noisier sets.
The full list, counted
Every product serper+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 |
|---|---|---|---|
| Carrd | 6 of 6 | 100% | every answer |
| Landingi | 5 of 6 | 83% | 5 answers |
| HubSpot | 4 of 6 | 67% | 4 answers |
| Leadpages | 4 of 6 | 67% | 4 answers |
| Unbounce | 4 of 6 | 67% | 4 answers |
| Instapage | 3 of 6 | 50% | 3 answers |
| MailerLite | 3 of 6 | 50% | 3 answers |
| Swipe Pages | 3 of 6 | 50% | 3 answers |
| Wix | 3 of 6 | 50% | 3 answers |
| Hostinger Website Builder | 2 of 6 | 33% | 2 answers |
The pages the answers were built from
Behind every name these assistants spat out was a URL, and we traced all of them. Six questions, six answers, and the source pool was staggeringly narrow. quora.com fed every single answer, all six. emailtooltester.com, reddit.com and zapier.com fed five each. emailvendorselection.com fed four. After that, the counts fall off a cliff. unbounce.com appeared in three answers; the rest of the domains showed up once or twice. If you want to understand how ChatGPT, Gemini and Perplexity build a recommendation list for landing page builders, you look at those five domains and you stop.
The mix is more forum than formal review. Quora and Reddit combined appeared in 11 of the 18 source slots that five domains would fill; they were in almost every answer even when the question was phrased as a shopping query. The assistants clearly treat a Quora thread where someone asked “best landing page builders” as a relevant source, probably because the surface language matches the prompt exactly. Reddit threads played the same role. A builder that never comes up in those two forums, on the exact phrasing the assistant searches for, has almost no chance of being named.
The actual review and comparison sites that shaped the answers were emailtooltester.com, zapier.com and emailvendorselection.com. emailtooltester publishes side-by-side builder comparisons with pricing tables; it was the single most-used dedicated review domain. zapier’s software roundups, written as long-form buyer guides, were in five answers. emailvendorselection.com, a smaller site, still made it into four. Notice what is missing: g2.com, the major review aggregator, appeared in exactly one answer. That is not an accident of scale; it is a function of which domains the underlying search API returned for these queries. The stack the measurement used (serper+or) fed the assistant a very specific diet, and that diet did not include the largest software review site except once.
Vendor pages crept in, but rarely. unbounce.com was the most recurring, with three appearances, usually when the answer mentioned Unbounce and the source was the company’s own site reinforcing the name. leadpages.com, mailerlite.com, wix.com, canva.com, sites.google.com and perspective.co each appeared just once. In the “best free” query, the assistants pulled from sites.google.com, mailerlite.com and wix.com directly, so a builder’s own landing page about its free tier became the proof that it existed. That is not a broad evaluation; it is scraping the first page that says “free.”
One source should stop you cold: aiaviationacademy.com showed up in an answer for “best landing page builders for small teams,” alongside Perspective. There is no sensible reason a flight school appears in a software recommendation. It is an artifact: some page on that domain mentioned the tool in passing, the search engine indexed it, and the assistant treated it as a source without questioning relevance. That single oddity tells you the system does not vet authority; it pattern-matches text to query.
YouTube, websiteplanet.com, thedoublethink.com each appeared once, rounding out the long tail of one-time sources. So a few well-optimised comparison blog posts and a video roundup can tip an answer, but not consistently.
The practical meaning for a builder on this list is stark. If you are Carrd, Landingi, HubSpot or any of the others that got named multiple times, you likely have a presence on quora, reddit, emailtooltester, zapier or emailvendorselection. That is all it takes. A builder not mentioned does not have an evaluation problem; it has a crawl problem. The assistant never saw a page about it because those five domains did not carry its name for those query strings. Fixing that means getting mentioned on those specific sites, not launching a general SEO campaign. The category’s recommendation landscape is decided by a handful of pages, and now you know exactly which ones.
Exposure is a function of the top five domains, not of the whole web
The assistants did not pull from a broad, representative sample of landing page builder reviews. They pulled from a handful of sites that kept showing up. quora.com was a source in all six answers. emailtooltester.com, reddit.com, and zapier.com each fed five. emailvendorselection.com fed four. No other domain appeared in more than three answers, and most appeared once. That concentration means any builder whose name rarely or never surfaces on those five domains is practically invisible to ChatGPT, Gemini, and Perplexity in this category. Carrd went six for six because it is apparently ubiquitous across that small set. Landingi missed only the “free” question, and that missing domain set tells the story: the free-builder query drew on reddit.com, emailtooltester.com, sites.google.com, mailerlite.com, wix.com, websiteplanet.com, and quora.com. Landingi was absent from the product list that came back, so either those specific pages did not frame it as free, or the assistants’ retrieval and synthesis filtered it out for that prompt. We do not get to see the mechanism, only the output.
The pattern holds when you look at products that appeared just a few times. Swipe Pages earned three mentions across all six questions, and in every case the source list for those answers included zapier.com and emailtooltester.com. Those two domains, plus quora and reddit, form a small clique that largely determines whether a builder like Swipe Pages gets recommended or not. Instapage also hit three mentions, and those answers always drew on zapier, reddit, and quora, with emailtooltester and emailvendorselection appearing in two of the three. The co-occurrence is not a coincidence; it is a picture of what the assistants’ retrieval step latched onto.
Self-referential sources add another layer. Unbounce was named in four answers, and unbounce.com was itself a source in three of those four. That does not prove the assistant copied from the builder’s own site, but the correlation is there. Leadpages appeared four times, but leadpages.com showed up only once as a source, so its presence was mostly driven by third-party mentions. Perspective was named twice, and perspective.co appeared in exactly one answer, the one for “best landing page builders for small teams.” That suggests the assistant surfaced Perspective because one of the fetched pages was the product’s own domain, and that source may not have recurred in other question contexts. The data cannot tell us whether that single appearance was enough, or whether the assistant also found Perspective mentioned in a more general source like reddit for that same answer. All we have is the domain list and the product list, not the provenance of each name.
That missing provenance is the largest gap in the measurement. We know that quora.com, reddit.com, and emailtooltester.com were part of nearly every answer, but we do not know which specific threads or articles the assistants drew from, nor how they weighted them. A builder that shows up in a Reddit comment might get pulled in, while a builder with a dedicated review on a site like g2.com might be ignored if the assistant’s retrieval never reaches that page. g2.com fed only one answer, for the broadest query “best landing page builders.” The single mention of Canva (on “what landing page builders should I use”) came from an answer whose sources included canva.com itself. Elementor appeared in that same answer, and WordPress + Elementor showed up in the free builder list where wix.com and mailerlite.com were sources, hinting that promotional pages from platform competitors can introduce adjacent tools.
What we can say with certainty is that the set of recommended products is not a verdict on quality. It is a map of which brands are referenced on the few dozen pages these three assistants happened to retrieve across six generic queries. A builder absent from those pages is absent from the recommendations, regardless of its features or pricing. Without access to the exact source content and the ranking logic inside each assistant, we cannot explain why, say, Hostinger Website Builder got two mentions and Kit got two while Involve.me got one. The difference may be nothing more than a single missing mention on a forum thread that the assistant read for some queries but not others. The assistant is not reviewing builders; it is mirroring, with some recombination, the surface of a tiny slice of the web.
If you're on this list, you're renting space you don't own
The appearance of your product in these answers is not a vote of confidence in your builder. It's a trace of which comparison articles, forum threads, and aggregator lists the assistants scraped when we queried them. For every named product, the mechanics are the same: quora.com, emailtooltester.com, reddit.com, zapier.com, and emailvendorselection.com fed the vast majority of answers, and your inclusion depends almost entirely on whether your name happened to appear on those domains in a way the assistants could parse. That means you are not defending a recommendation. You are defending a position inside a handful of third-party pages you do not control.
If you sell HubSpot, Leadpages, or Unbounce and you see four mentions and think the assistants value your feature set, you're reading it wrong. The four answers that listed you drew from the same narrow source pool that also listed you. One refresh of a roundup on emailtooltester.com or a new top-voted answer on a Reddit thread could drop that count to two or one. The assistants will not test your editor, your load speed, or your pricing page. They will just stop seeing your name because someone else's name has taken the space. HubSpot's four mentions are one editorial decision on a Zapier post away from becoming two. And the Zapier post is not about you; it is about "best landing page builders," a list any competitor can lobby their way onto with a better pitch, a partnership, or a batch of new user reviews.
For Carrd, the perfect six-out-of-six score looks like a moat. It isn't. The six mentions trace back to the same five domains everyone else relies on, plus a few one-offs. A single Quora thread deletion, a rewrite of the emailtooltester.com free builders page, or a shift in which Reddit comment the assistant's summarization model latches onto could erase Carrd from an answer or two without warning. A perfect score powered by a tiny dependency list is not a durable lead; it's a house of cards where one domain's disappearance can break the streak.
Landingi's five mentions illustrate the fragility even more clearly. The one question where it didn't appear was "best free landing page builders." That single absence was not about the product lacking a free tier. It happened because the sources the assistants used for that query (reddit.com, emailtooltester.com, sites.google.com, mailerlite.com, wix.com, websiteplanet.com) didn't name Landingi. The omission is a pure artifact of source selection. If emailtooltester.com adds Landingi to its free builders list, Landingi climbs to six. If not, it stays at five. The product itself is a passenger.
What would actually cost you the spot is simple: an investment by a competitor in getting their name onto the same pages that currently hold yours. Not a better builder. Not a lower price. Just presence. If a competitor runs a campaign to get mentioned on Zapier's roundup, shows up in Quora answers with enough upvotes, and earns a spot in a Reddit thread that gets indexed, they can flip the recommendations for several queries at once because the source overlap is so extreme. The assistant won't weigh your decade of market presence against their week of link building. It will surface what it finds, and what it finds is whatever the five domains currently contain. Nothing in this measurement suggests the assistants have any durable allegiance to a product. They have durable allegiance to the same few URLs. Defending your spot means defending your mentions on those URLs, and that is an entirely different job than building landing pages.
If you were not named, you aren't on the five pages that matter
Seventeen builders were named at least once. Many more were not. No assistant pulled from G2 in most answers, never touched Capterra, and largely ignored the builder's own site unless that site was Unbounce and the question was broad. The difference between being named and being invisible is not quality, feature set, or even existing reviews. It is presence on quora.com, emailtooltester.com, reddit.com, zapier.com, and emailvendorselection.com. Those five domains fed five or six of the six answers. No other domain fed more than three, and almost all appeared once. If you sell landing page builders and did not get a single mention, the reason is mechanical: your name does not appear on those exact source pages in a pattern the assistants extract.
What it involves. Getting named means working backwards from the specific URLs that surfaced in our measurement. For quora.com, that is likely a single thread ranking for generic queries like "best landing page builders." If your builder is not mentioned in that thread, you will not appear. The fix is not a Quora ad buy. It is ensuring an answer exists in that thread, ideally one with upvotes and clear product mentions. For reddit.com, the same logic applies: a handful of threads fed five answers. Identify which ones (the data does not give thread-level URLs, but the domain pattern holds). For zapier.com, the assistants pulled from directory-style guides and integration roundups. Your builder needs a listing on the relevant Zapier page, and the listing must include text that gets indexed and parsed. For emailtooltester.com and emailvendorselection.com, the assistants read comparison tables and "best of" lists. If your builder is not in the table or article that surfaces for these queries, you are out. So the work is: locate the three or four specific pages on each of these five domains that rank for any of our six question variants, then get your product added to them. That might mean submitting to a comparison site, asking a Quora user to include you, or building a Zapier integration so you appear in their app directory. It is all product listing hygiene, not marketing.
What it does not involve. It does not involve generating more backlinks from unrelated sites. The source pool shows that domains outside the top five contributed nothing reliable. unbounce.com appeared three times, but that was Unbounce's own domain, and the assistant was likely using it as corroboration rather than a source of discovery. Leadpages.com appeared once. No other builder's own site appeared at all. So a content strategy built around your own blog will not get you into these answers unless that blog gets quoted by one of the five domains. It does not involve trying to "optimize for AI" with schema markup or branded Q&A pages. The assistants are not crawling your site and forming an opinion. They are repeating what they find on quora.com, reddit.com, and the three other domains. It also does not involve negotiating with the assistant companies. ChatGPT, Gemini, and Perplexity do not accept payments for placement in organic answers (at least in these results), and no measurement here suggests a pay-to-play mechanism exists. The path is simpler and narrower. You get on the five pages, or you stay out.
How to replicate this measurement for your category by hand
Running this by hand means you will end up with 18 individual answer lists if you pick six questions and three assistants the way we did, and the moment you start your third question you will wish you had built a spreadsheet before typing the first one. The process is not complicated but it is repetitive in a way that punishes small distractions. The first step is to write the questions. We used six: “best landing page builders”, “best landing page builders for small teams”, “what landing page builders should I use”, “best free landing page builders”, “most affordable landing page builders”, and “landing page builders recommendations”. You need a set that covers the ways someone actually asks for your category. Write them down verbatim because you will paste them exactly the same into each assistant later and any variation makes the later counting unreliable.
Open three browser windows, one for ChatGPT, one for Gemini, one for Perplexity. Paste the first question into all three, then copy the answer from each. Not just the product names but every source link if the assistant provides them. Perplexity usually does, ChatGPT and Gemini sometimes do, and the format is never identical. You will find yourself cleaning up markdown lists, expanding shortened URLs, and deciding whether a mention inside a paragraph counts the same as a bolded heading. For landing page builders, Carrd appeared in all six answers, but confirming that meant checking answers where it was typed as “Carrd (carrd.co)” and others where it was just a single word in a comma-separated sentence. One missed instance would drop it to five and change the entire finding.
After one question you have three answer sets. After three questions you have nine, and the tallying starts to fray. The real drag is the source domains. quora.com showed up in all six answers across all three assistants, but verifying that means clicking every hyperlink in every answer. Some answers listed sources at the bottom, others embedded them as citations mid-text, and a few linked domains only as anchor text that you have to hover over to read. emailtooltester.com, reddit.com, and zapier.com each fed five answers, and the moment you think you see a pattern you still have to check the single answer that missed it to understand why. The absence is where the insight lives, but it is also where a manual counter is most likely to miscount.
The counting itself is the tedious core. For named products, you go question by question, assistant by assistant, making a mark next to any name you see. Landingi got five mentions, missing only the explicitly free query, but tracking that single omission meant scrolling back through the answer for “best free landing page builders” from each assistant to make sure it was truly absent and not just buried in a paragraph about freemium tools. Multiply that check by the seventeen builders that appeared at least once and you have an afternoon of slow, error-prone tabulation. No assistant gives you a clean export; you are cutting and pasting into a text file, then into a sheet, then back to the browser to verify.
The value of the exercise is exactly the tedium of it. By the time you have recorded every product name and every source domain for six questions and three assistants, you will never again wonder whether a recommendation is editorial judgment or scraped consensus. You will have stared at the same five domains feeding every answer, counted the same single-appearance names that never came back, and felt the weight of a system that collapses broad intent into a tiny pool of pages. That understanding costs nothing but the time it takes to type, paste, and tally. It is simply very boring work, and most people will not do it.