63% of AI citations point at listicles. You are not in a single one. Here is how to get in.

Open ChatGPT. Ask it for the best company in your category, in your town. Read the answer.

Now look at where it got that answer from. It is almost never a company website. It is somebody else's list.

That is the whole game in 2026 and most people are still refusing to play it. They are writing their 400th blog post, buying a few links, and wondering why the machine keeps naming three rivals and not them. The machine is not ignoring you. It is reading a page you are not on.

What a listicle citation actually is

A listicle is any ranked or grouped roundup: "best CRM for small teams", "top 10 damp proofing companies in Newcastle", "7 alternatives to Semrush". A listicle citation is when an AI assistant answers a question by pulling from one of those pages and linking it as a source.

It matters because of how these systems work. When a model does not already know the answer cold, it searches, grabs a handful of pages, and writes its verdict from what those pages say. If the page it grabs is a roundup of ten companies, those ten companies get named. Everyone else in the category simply does not exist for that answer.

So there are two ways to be in an AI answer. Be the page it quotes, or be named inside the page it quotes. Almost everyone chases the first one. The second one is faster, cheaper, and far less crowded.

Why this matters right now, with the numbers

Evertune reviewed the 6,000 most cited URLs per model across ChatGPT, Copilot, Gemini, Google AI Mode, Google AI Overviews and Perplexity for March and April 2026. Roughly 25,000 unique URLs. Half of them were listicles. Across nearly 400 million citations, 63% pointed at listicles. Every single model did it, from 40% of most cited URLs on Copilot up to 65% on Gemini.

It gets more specific. The vast majority of those listicles were ranked lists, the "Top 5 CRM Tools" format, at 71% to 86% depending on the model. Big institutional rankings, the U.S. News style data monsters, were almost nothing: 1.4% to 4.7%. Forbes landed in the top three listicle sources on every single model, largely off the back of Forbes Advisor and Forbes Vetted.

Worth saying plainly: that study was published as a sponsored piece by an AI visibility vendor, and Search Engine Land noted it neither confirms nor disputes the conclusions. Treat the exact decimals as directional. The direction itself is not in doubt, because independent work says the same thing.

Bill Widmer logged 13,184 citations and 1,765 answers across ChatGPT, Claude, Gemini and Perplexity over about three months, published on Orbit Media in September 2026. Two findings from it should reorganise your week.

First, the models barely agree with each other. Across 1,792 question and domain combinations, all four cited the same domain for the same question just 30 times. That is 1.7%. Perplexity averages 19.2 sources per answer, Gemini 8.1, ChatGPT 4.5, Claude 3.6. Claude, in his data, loves review platforms and curated lists, with Clutch as its single most cited third party domain.

Second, and this is the one that should hurt: ranking in Google is not the ticket you think it is. Only 10% to 30% of the domains the models cited also sat in Google's top 10 for the matching query. At exact URL level it falls apart completely, roughly 1% to 5% for Gemini and ChatGPT. You can be number one on Google and absent from the answer.

One more, because it explains why your dashboard looks confusing. Brands get mentioned far more often than they get cited. Claude mentioned brands in 70% of answers and cited them in 40%. Gemini, 72% and 41%. ChatGPT was the tight one at 56% and 47%. Famous brands get recommended from memory and sourced from somebody else's page. That gap is the listicle.

The Reddit correction nobody wants to hear

For two years the entire industry has told you Reddit is the kingmaker of AI search. In Widmer's last run across 72 B2B buyer questions, Reddit was 0% of Claude's citations, 0% of ChatGPT's, 3% of Perplexity's and 1% of Gemini's. Claude never cited Reddit once across hundreds of answers.

That does not make Reddit worthless. It makes it category specific. For consumer questions it still lands. For B2B and most local services, the citation data says niche roundups, review platforms and comparison posts got the slots instead. Check your own category before you spend six months farming karma. We show you how to check it in about twenty minutes inside the Dojo.

Why writing your own "best X" page will not save you

Here is where people get clever and then get binned. They think: fine, lists win, I will write my own list and put myself at number one.

In February 2026, Lily Ray of Amsive published research on SaaS and B2B brands that got hammered after the December 2025 core update and through January's volatility. The pattern: organic visibility down 30% to 50% within weeks, and the losses were not sitewide. They were concentrated in the blog, guide and tutorial folders, the exact folders stuffed with dozens or hundreds of self promotional listicles where the publisher ranked itself first, often just refreshed with "2026" slapped in the title.

Google has not confirmed an update tied to this, and those sites were doing plenty of other risky things at the same time, so treat it as a strong correlation rather than a proven cause. But the direction of travel is obvious, and Ray's own point is that whatever hits Google organic tends to bleed into every model that leans on Google's results.

Then there is the part with actual legal teeth. The FTC's rule on consumer reviews and testimonials prohibits a business from misrepresenting that a website or entity it controls provides independent reviews or opinions about a category of products or services that includes its own. Ranking yourself number one on your own "best in category" page, with no disclosure, no methodology and no independent testing, is not a clever SEO move. It is a filing risk.

So the play is not to fake the list. The play is to get into other people's lists, honestly, and to be so obviously the right inclusion that leaving you out looks like an error.

The five list types worth chasing

Not all lists are equal. Chase in this order.

Notice what is not on that list: general guest posts on unrelated blogs. A link from a page about nothing in particular does very little for AI visibility, which is a big part of why the guest post market prices look the way they do. We broke the real numbers down in what backlinks actually cost in 2026: sellers ask a median of $929, buyers actually pay $207. Spending $207 on a page no model will ever retrieve is the most common waste in this industry.

The play, step by step

Step 1. Write the prompts your buyer would actually type

Not keywords. Prompts. "Who is the best damp proofing company near Newcastle", "best all in one SEO platform for a two person marketing team", "is it worth paying an agency to manage Amazon". Pull them from sales calls and the questions you get asked on the phone, not from a keyword tool. Twenty of them is plenty to start.

Step 2. Run them and log every URL the model cites

Run each prompt through ChatGPT, Gemini, Perplexity and Claude and write down every single source URL. You are not looking at whether you appear. You are building a map of who owns the citations in your category. You do not need a $500 tool for this. A spreadsheet and an hour does it, and we covered the free setup in tracking AI visibility without paying for a tool.

Step 3. Build the gatekeeper sheet

Count repeats. Any page that shows up for three or more of your prompts, or across two or more models, is a gatekeeper. Those are your targets. Sort by frequency, not by domain authority. A scrappy niche blog cited in eight of your twenty prompts is worth more to you than a famous site cited in none.

This is the bit most agencies charge $2k a month to do, and it is one focused afternoon once you know what you are counting.

Step 4. Qualify before you pitch

For each gatekeeper, check three things. Is the list actually maintained, or is it a 2023 relic still getting cited out of habit? Does it have an obvious inclusion path, a submission form, a "suggest a company" line, a named author? And is the list honest, or is it a vendor ranking itself first? Skip that last type. It is on borrowed time.

Step 5. Pitch like a human with receipts

The pitch that works is short, names the specific list, and hands over something the writer cannot easily get elsewhere. Pick one: a number only you have, a client outcome you can evidence, a service or coverage area nobody else on the list offers, or a genuine correction to something on the page.

The pitch that fails is "hi, love your blog, please add us". You are asking a stranger to vouch for you in public. Give them a reason they can defend.

Step 6. Hand them a block the model can lift

This is the move almost nobody makes. Do not just ask to be added. Write the entry for them: your name, one clean sentence on what you do and who for, the differentiator, the price or price range, the coverage area. Forty to sixty words, factual, no adjectives doing the heavy lifting.

Two things happen. The writer's job becomes copy and paste, so your odds go up. And the text that lands on the page is text designed to be quoted, because models retrieve and reuse self contained chunks rather than whole pages, which we explained in how AI actually reads your page.

Step 7. Make sure you survive the fact check

When a model retrieves a list with your name on it, it will often go and look you up to confirm. If your name, service description, address and phone number say different things in different places, you look like a weak entity and you get dropped in favour of the rival that is consistent. Fix that first. It is unglamorous and it is the cheapest win available, and it is the same work we lay out in entity SEO.

Step 8. Track climate, not weather

Widmer re ran the same 72 questions two days apart. Claude kept only 30% of the URLs it had cited 48 hours earlier. Gemini kept 38% of its domains. Perplexity was the steady one at 65%. So a single AI answer tells you nothing. Run your prompt set weekly, judge it on three or more weeks of data, and do not redesign your strategy because of one bad Tuesday.

Timeline: what actually happens when

To be straight with you: nobody can promise you a citation or a date. The mechanism is reliable. The timing is not, because retrieval is non deterministic and every model recrawls on its own schedule.

What this costs in real money

We are deliberately not publishing a price for paid list inclusion. It swings wildly, and if you are paying for a slot it is advertising, which means it should be disclosed and it should not be passing link equity. Anyone selling you a "guaranteed top spot" on an independent roundup is selling you something the FTC has a rule about.

If you would rather not run this yourself at all, that is what Done For You is for, and the link side of it lives in link building.

Who this suits and who it does not

This suits you if you sell something people compare before buying: local trades, agencies, SaaS, professional services, anything where a buyer asks "who is good at this". It suits small brands especially, because small brands get cited almost one for one with being mentioned, so every list inclusion shows up in your data.

This does not suit you if you are in a category nobody writes roundups about, if your product is genuinely unique with no comparison set, or if you are a household name already. Widmer's data showed one large brand getting mentioned in 83% to 100% of relevant answers while its citation rate swung all the way down to 0%. If that is you, the model recommends you from memory and your citation dashboard is measuring the wrong thing.

The tools you actually need

Four AI accounts and a spreadsheet. That is the honest answer for the first month. ChatGPT, Gemini, Perplexity and Claude on free or cheap tiers, plus a sheet with columns for date, prompt, model, cited URL and whether you were mentioned. Add a paid visibility platform later if you are managing several brands and the manual runs get painful. Our current picks sit in the Armory.

What you do not need is a $500 a month dashboard to tell you that you are not in any lists. You can establish that in an afternoon for free.

Mistakes that will get you binned

• Ranking yourself first on your own roundup. Covered above. Visibility risk and a rule with the FTC's name on it.
• Mass pitching the same email to 200 blogs. Gatekeepers talk, and a generic pitch marks you as somebody who did not read the page.
• Chasing domain authority instead of citation frequency. The famous site cited zero times for your prompts is worth zero.
• Refreshing the year in a title and calling it an update. That exact pattern shows up all over the hardest hit sites in Ray's research.
• Assuming Reddit is the answer because everyone said so. In B2B it was near zero across three of four models. Check your category.
• Panicking after one run. Up to 70% of cited URLs can turn over in 48 hours. Weekly tracking or nothing.
• Writing a 4,000 word monster for a model that wants 1,000. Evertune's heavily cited pages typically ran 1,000 to 2,000 words at about 18 words a sentence.

What to do instead if you cannot get into the lists

Sometimes the gatekeepers are closed. Three honest alternatives, in order of how well they work.

Become a gatekeeper yourself, properly. Publish the roundup for your category and leave yourself out of the rankings, or disclose your position clearly with a stated methodology. Harder, slower, and it compounds, because you become the page other people get cited through.

Publish the number nobody else has. Original data is the one thing models cannot generate, so it earns citations directly and earns list inclusions as a side effect. Survey your customers. Publish your real pricing. Count something in your industry that nobody has counted.

Get the review platforms right. Clutch, G2 and the category pages in your niche are structured, comparative and constantly recrawled. Claude's most cited third party domain in that dataset was a review platform. This is the least glamorous item on the list and often the fastest.

If you want the long view on why third party mentions beat links for this, we argued it out with data in backlinks versus brand mentions.

The risks, stated plainly

You are building visibility on pages you do not control. The gatekeeper can delist you, rewrite the list, sell the domain or get hit by an update, and your citations go with it. Diversify across at least ten sources and never let one page carry your category presence.

The second risk is that this window closes. It is wide open right now because most businesses have not noticed that the citation is the prize. When every competitor is pitching the same twenty roundups, the price of inclusion goes up and the honest routes get crowded. That is an argument for moving this quarter, not next year.

The 2027 view

Three things look likely from here. Models will keep tightening on self dealing lists, because Google has already started and the FTC rule is written. Review platforms and structured comparison data will get more weight, not less, because they are machine readable and independently sourced. And the "who owns the citations in this category" question will become the standard first question of any audit, the way "what is your domain rating" was in 2018.

The businesses that win are not the ones with the most content. They are the ones that show up on the fifteen pages their category's answers get built from.

Do this today

Open a spreadsheet. Write down ten questions a buyer would type. Run them through ChatGPT and Perplexity. Log every source URL. Count the repeats.

You will have your target list before lunch, and you will probably be annoyed by what you find, because the pages deciding your category are pages you have never pitched.

The full system, the prompt templates, the gatekeeper tracker, the pitch scripts and the order to run them in are all in the Dojo. Join The Dojo. $67/mo.

If you want us to just do it, that is Done For You, and the wider search side is SEO services and local SEO.

Sources

• Evertune, AI search loves listicles: what 25,000 URLs reveal about citations, Search Engine Land, 19 May 2026
• Bill Widmer, I tracked 13,184 AI citations across 4 LLMs, Orbit Media, 2 September 2026
• Danny Goodwin, Google may be cracking down on self promotional best of listicles, Search Engine Land, 4 February 2026
• Lily Ray, Is Google finally cracking down on self promotional listicles?
• FTC, 16 CFR Part 465, Trade Regulation Rule on the Use of Consumer Reviews and Testimonials
• Google Search Central, reviews system documentation