The Self-Promotional Listicle, Demystified

You’ve seen these articles a thousand times. Here’s what’s actually happening in them.

1. A company writes “The 12 Best [Whatever] Tools,” and surprise, they’re number one. You’ve half-noticed it. What you didn’t notice: they just voted for their competitors.
2. An AI assistant quotes that exact article when you ask “what’s the best.” It names the brand as a source, then recommends four rivals and leaves the author out entirely.
3. The page ranks beautifully, gets cited everywhere, and still loses the customer. Cited is not recommended, and most companies track the wrong one.
4. In January, a wave of these sites lost 30 to 50% of their traffic in weeks. Not a folder. Whole domains. Most never connected the dots.
5. Some of those number one spots were simply bought, from around £200. Openly sold as a shortcut into AI answers. Much of it quietly breaks Google’s rules.
6. The most polished versions, with invented scores and self-awarded badges, can now be fined. Up to $53,000 a time in the US, up to 10% of global turnover in the UK.

Yes, we know what we just did: opened a takedown of listicles with a listicle. That was the point. If two or three of those made you pause, the rest of this article is for you. We are going to take the whole tactic apart: what it is, why it spread, and exactly where it crosses from clever to costly.

You don’t have to take an SEO’s word for any of this

Every one of those six claims is checkable, and not because an SEO asserts it. Each traces back to a primary source: Google’s published spam policies, consumer-protection law on two continents, and the behaviour of the AI answers themselves. This piece walks all three in plain terms, with the sources linked at the foot, so you can verify every line rather than trust ours.

Here is the shift underneath it all. For a few years, publishing your own “best of” list and seating yourself at number one was close to free money: the page ranked, and AI assistants repeated it. That trade has now inverted. The reasons are not one analyst’s hot take. They are written into policy, into law that carries penalties from five figures per violation up to 10% of worldwide turnover, and into the way the AI answers themselves now behave.

Strip away the format and the move at the centre of it is simple: a seller appoints itself the judge of its own category and, predictably, awards itself the win. Everything below is about why that self-appointment, which used to work, now carries three separate risks most people lump together, and where exactly the line into unlawful sits.

Related analyses in this cluster

This is the complete guide. Two shorter companion pieces each go deeper on a single angle:

• The AI Overview Listicle Problem is the current-evidence piece: what AI Overviews are actually doing with self-promotional lists right now.
• The Listicle Question is the forward-looking piece: why a self-promotion filter may be structurally inevitable.

First: what even is a listicle?

Most people use the word without quite defining it, so let us be plain. A listicle is simply an article built as a list: “7 ways to…”, “12 best…”, “5 reasons you should…”. That is all it is. The vast majority of listicles are perfectly honest and genuinely useful, and the format itself is not the problem.

The problem is one specific use of the format. Picture a spectrum:

• Independent comparison. Someone who does not sell a product in the category ranks the options on transparent criteria. Genuine review platforms and independent advisors live here.
• Vendor comparison. A seller compares the field, including itself, but assesses honestly and does not automatically crown itself. Defensible, when disclosed.
• Self-promotional listicle. A seller publishes a “best [category]” list and ranks itself first, usually with a scoring system of its own invention.
• Fabricated independence. The same, but dressed up as neutral editorial: a made-up “scientific” methodology, a self-awarded 9.8/10, sometimes fake review stars in the search result.

The defining feature of the risky end is the conflict of interest baked in: the publisher is a contestant and the referee. When this article says “the tactic,” it means that, not the format. There is an even sharper version doing the rounds: ranking not just your own brand but the individual author personally above named peers in the same field. Same conflict, turned up a notch.

Why it spread (the fair version)

It is worth stating the mechanics honestly, because the tactic spread for rational reasons, not stupidity.

“Best [category]” searches are high-intent and close to the moment of purchase, so the traffic converts. For years there was a content void around them, most brands were uncomfortable publishing openly biased content, so competition was thin. Review platforms had built whole businesses monetising those same queries, and vendors reasonably asked why they were paying someone else to rank for their own category. And crucially, a listicle needs no original research. Then AI search arrived and amplified it: assistants lean heavily on these lists when they answer “what’s the best X,” so the page paid off twice: in rankings and in AI citations.

And it was not a fringe trick. One analysis tracking the pattern found around 146 brands running self-serving listicles, roughly 50 of them starting in 2025 alone, growth of nearly 200% year on year, and that is one analyst’s sample, a sliver of the whole web. Some sites ran not one or two but hundreds or thousands of these pages. The most public example is Shopify, which published a large library of “best ecommerce platform” articles seating itself at number one (and, tellingly, now appears to be culling many of them). For a window, it genuinely worked. That is exactly why being clear-eyed about the pitfalls now matters more than mocking the format.

The three risks: three different wires

This is the part almost everyone runs together. Publishing a self-serving list exposes you to three separate dangers, with three different consequences and three different referees. Keep them apart and the whole topic becomes clear.

Three referees, three different consequences. Only the regulator can fine you.

Risk 1: you can be cited without being recommended

This is the AI-era trap, and it is the finding most worth corroborating from more than one source, because it is the one people most want to wave away.

The most-quoted measurement comes from Lily Ray’s analysis of B2B “best [category]” queries in Google’s AI surfaces: when a brand’s own listicle was cited as a source, that brand was left out of the actual recommendation roughly two-thirds of the time, with the recommendations going to established leaders instead. Treat that figure as one specific data point from one study. A clean illustration from the same research: for “best LMS for selling courses,” one vendor’s page was cited repeatedly across the answer, yet the brand itself was not recommended, while the competitors it had listed were.

But it does not stand on one analyst. From a different direction entirely, measurement not content, Tom Critchlow, working from a Similarweb dataset, argued that clicks and citations are the least interesting metrics, because an AI recommendation made users markedly more likely (around 2.5×) to visit the recommended brand within a week. From a PR perspective, Frank Strong made the trust argument independently: a company calling itself the best persuades no one, where a customer, an analyst or an independent article does. The B2B content agencies watching their own clients’ data reached the same place.

Four vantage points (an AI-search analyst, a measurement specialist, a PR strategist, a content agency) converging on one thing: recommendation is anchored to what the rest of the web says about you, not to what you say about yourself. A self-promotional listicle is pure self-declaration, which is precisely why it can earn a citation and still hand the recommendation to the corroborated competitors it names. Our Best-Of Problem page documents this cited-not-recommended pattern in live AI Overviews.

Risk 2: you can be quietly demoted

The second trap is in ordinary organic search, and here precision matters because it is easy to overclaim.

Ray documented sites leaning heavily on the tactic losing organic visibility (in several cases drops of roughly 30–50% within weeks, around a 20 January 2026 inflection), concentrated at first in the blog, guide and tutorial subfolders where these pages live, with reports of the damage spreading domain-wide in the heaviest cases and accelerating through Google’s May 2026 core update. Independent agency analyses observed the same pattern. The honest statement: a correlation between heavy self-serving-listicle use and visibility loss, consistent across multiple observers, not a confirmed, dedicated penalty, which Google has not announced.

What Google has published supports the direction without anyone speculating. Its spam policies name scaled content abuse and site reputation abuse, the latter violated when third-party content rides a host site’s established ranking signals, which is exactly the model behind renting a high-authority domain to carry a “best [category]” page. Google’s structured-data guidance separately says you should not use self-serving reviews to generate star ratings in search, which catches the fabricated-independence end directly.

There is also a structural reason these pages are fragile: search “best CRM for small business,” or “best MFT software,” and you find a pile of near-identical “best … 2026” pages, all racing on the same freshness hook. That crowded sameness is a footprint, and it is the self-promoters inside it who are most exposed when quality systems start discounting templated content.

Risk 3: you can be fined, but only the deceptive version

This is the legal wire, and the one to draw most carefully, because lumping it in with the others is where people get scared off normal practice. A fine is on the table only for the deceptive end of the spectrum (fabricated independence and fake reviews), not for an honest, clearly-labelled comparison. The next two sections draw that line exactly.

Lawful, risky, unlawful: drawn three times

The same line, across three industries. Note the difference in how people actually search: software is bought on fit, not proximity (you would search “best CRM for small business,” never “best CRM in Manchester”), whereas restaurants and solicitors are local, so geography is the natural modifier. Getting that right matters; mixing them is a tell that content was written by someone who does not use search.

The same line, drawn three times. The detail follows below.

Software (e.g. a CRM): segment modifiers, no geography – ✅ Lawful: “How [our CRM] compares to Salesforce and HubSpot,” plainly your own comparison, honest about trade-offs. Targeting “best CRM for small business” or “for healthcare” is normal segment SEO. – ⚠️ Risky: “The 10 best CRMs for 2026,” an invented scoring system, yourself at 9.8/10 and number one above named rivals, presented as neutral editorial. Google may demote it; dressed as independent, it edges into regulator territory. – ❌ Unlawful: bolting on fake star ratings or reviews from people who never existed, or paying for five-star reviews to prop the page up.

Law firm: local modifiers – ✅ Lawful: “Why choose [our firm] for personal-injury claims,” obviously your own pitch. “One of the leading PI firms in Leeds” as genuine opinion sits here too (with the caveat in the next section). – ⚠️ Risky: “The 7 best personal-injury solicitors in Leeds,” six competitors listed, yourself at number one under a made-up “ranking methodology,” styled as an impartial local guide. (Solicitors also carry SRA rules on misleading publicity on top of everything else.) – ❌ Unlawful: fabricated client testimonials, or running a “Best Solicitors in Leeds” review site you secretly own while presenting it as independent.

Restaurant: local modifiers – ✅ Lawful: “Some of the best Thai food in Bristol” on your own site. Classic puffery. Fine. – ⚠️ Risky: “The 10 best Thai restaurants in Bristol” on your own site, quietly seating yourself at number one above nine real rivals, presented as a neutral local guide. – ❌ Unlawful: fake five-star reviews, planted hit-piece reviews of those nine rivals, or running “BestBristolEats” as a “neutral” site you secretly own that always crowns you.

The pattern is identical across all three, which is what makes it teachable: talking yourself up is fine. Posing as the impartial judge of your whole category is the risk. Faking reviews or independence is the line into unlawful.

“But can I still put ’best’ in my titles?”

This is the question nine in ten readers actually have, and the honest answer is reassuring, with one sharp edge. Weaving “best” into a title tag (“one of the best family law solicitors in Leeds”) is a claim about yourself, not a ranked judgement of competitors. It is a completely different thing from the tactic above, and it carries essentially none of the listicle’s legal risk. But “best” is not a free pass either, and the reason is worth understanding.

The test is simply whether a reader could take it as a checkable fact.

Puffery is vague, subjective boasting too woolly to prove or disprove, so the law treats it as obvious salesmanship rather than a factual claim. The test is simply: could you measure it? “World-class service,” “one of the finest,” “probably the best lager in the world,” unmeasurable and lawful. The moment a reader would take the words as a checkable fact (“rated number one,” “voted best by [body],” “98% success rate”), it stops being puffery and needs evidence.

Both jurisdictions draw the line the same way, and both now have teeth:

• UK (ASA / CMA). The ASA judges a claim by how a consumer is likely to interpret it, not by what you intended, and it has repeatedly treated unqualified “best,” “No. 1” and “leading” as objective claims that require documentary evidence under CAP Code Rule 3.7. “Probably the best lager” survives because it is genuinely subjective; “voted Britain’s best for coverage” did not, because the survey behind it could not deliver objective data. An award does not automatically substantiate a “best.” And the stakes rose sharply: under the DMCC Act, since 6 April 2025 the CMA can directly impose penalties of up to 10% of worldwide turnover for consumer-law breaches.
• US (FTC). The FTC tolerates puffery but acts on specific deceptive claims, with civil penalties (see below) for the review-rule breaches.

Applied to your titles: – “One of the best family-law solicitors in Leeds” → leans on opinion, but because unqualified “best” tends to read as objective, it is safer when you genuinely have a basis for it. – “Leeds’ No. 1 family-law firm” → almost certainly read as an objective claim; risky unless a real, citable ranking says exactly that. – “Voted best by [X]” with a logo → only safe if the award actually says what you are implying.

One line to remember: vague self-praise is puffery and usually fine; a specific claim you cannot back up is misleading and risky, on both sides of the Atlantic.

The legal layer, precisely

Two things make the deceptive end genuinely risky now rather than theoretically.

In the US, the FTC’s Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465) took effect on 21 October 2024, carrying civil penalties of up to $53,088 per violation (adjusted for inflation annually). One provision aims squarely at this tactic: misrepresenting a company-controlled site as independent (§465.6). A vendor’s “best [category]” page with an invented scoring system that crowns the vendor, dressed as neutral editorial, is the shape of thing that describes. And on 22 December 2025 the FTC took its first public enforcement step under the rule (warning letters to ten companies), signalling a move from education to enforcement. In fairness, those particular letters concerned fake and incentivised reviews rather than listicles specifically, and warning letters are not findings of guilt. But the company-controlled-site provision plus an enforcing regulator raise the floor under the deceptive end.

On Google’s side, the line is narrower than “paid placement is forbidden.” Google explicitly accepts that buying and selling links is a normal part of the web’s advertising economy, provided the links are tagged rel="sponsored" or rel="nofollow". So affiliate and sponsored placements are fine when tagged and disclosed; the violation is paying for a followed link that passes ranking credit while posing as ordinary editorial, which puts both the buyer and the seller in breach. That is why the £200 placement market is riskier than it looks: most of it sells exactly the link Google’s policy prohibits, relabelled as “editorial.”

Being our own devil’s advocate

A demystification that only argues one way is just advocacy. So, the honest qualifications:

• It still works for some. Brands that are already dominant and widely recommended can still get both cited and recommended from a self-ranking page, because the corroboration already exists elsewhere. The tactic is not dead; it has stratified by pre-existing authority. If you are already the category leader the page is largely redundant; if you are not, it is increasingly a liability.
• The data is early. The headline figures come substantially from analyst studies with finite samples, and Google has confirmed no dedicated “listicle penalty.” We corroborate the direction across independent sources, but we are reading a trend, not quoting a law. Where that trend is heading, and why a self-promotion filter may be inevitable, is the subject of The Listicle Question.
• Neutrality is hard. A comparison published by an interested party is still published by an interested party. We are not pretending vendor content can be perfectly objective; what separates defensible from deceptive is disclosure and honest, transparent methodology.
• Read us critically too. We are an SEO consultancy writing about SEO tactics, with our own interest in you valuing this kind of work. That is exactly why every load-bearing claim here is tied to a primary document or an independent analyst, linked below, so you can check it rather than take our word.

And apply all of this to the lists that flatter the experts, too. “The best [profession] in [town],” “the top voices in [field],” the influencer round-ups an AI now reads back to you when you ask who to trust: many are built by the same mechanics described here, some with paid inclusion, some self-published. Being on one is often evidence of the tactic, not proof of authority. The same scepticism applies whoever the list crowns, ourselves included.

What to do instead

The constructive version is simpler than the tactic it replaces, and it compounds.

Earn the thing that actually drives recommendation: be the brand other people list and recommend. Independent reviews, genuine third-party comparisons, analyst coverage, real citations across the trusted web. This is not just our reading of it. A Google patent reported in June 2026, describing how an AI system builds what it calls a “deep, holistic characterization” of a business, sets the mechanism out plainly: the system assembles its understanding of an entity from many sources at once, including reviews, maps data, business listings and third-party mentions, and treats that as an interpretation rather than a copy of whatever the business says about itself. Put simply, it is designed so it cannot be talked into a verdict by a page you wrote about you. The patent’s own guidance for businesses is the tell: keep your description consistent across sources, support claims with evidence, and audit how the wider web would describe you. That is the precise opposite of the self-promotional listicle.

If you publish comparisons yourself, make them genuinely useful: transparent criteria, honest acknowledgement of where competitors are stronger, no automatic self-coronation, clear disclosure, and links tagged correctly. That content is harder to fake, which is exactly why it ages well while templated self-praise ages badly.

The shift, in one line: stop telling the web you are the best, and start giving the web reasons to say it for you.

Sources

Primary: Google – Spam Policies for Google Web Search (link spam; scaled content abuse; site reputation abuse). Google Search spam policies – Qualify your outbound links (rel="sponsored", ugc, nofollow). Qualify your outbound links – Link-tagging and link-spam update. the 2021 link-spam update – Google, “Data extraction using LLMs” (WO2025063948A1), PCT application filed 20 September 2023, published 27 March 2025, status pending. the WO2025063948A1 patent filing. First surfaced for a general audience by Search Engine Land (Sanger, June 2026).

Primary: regulators – FTC Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465), in force 21 Oct 2024. ftc.gov – FTC warns 10 companies / first enforcement, 22 Dec 2025; up to $53,088 per violation. ftc.gov – ASA/CAP, Types of claims: Superlative; Substantiation (Rule 3.7); Types of claims: “No. 1”. asa.org.uk – DMCC Act 2024. CMA direct penalties up to 10% of worldwide turnover from 6 Apr 2025

Independent analysis (multi-source corroboration) – Lily Ray: self-promotional listicle study (cited-vs-recommended; organic drops; brand-growth data). lilyraynyc.substack.com / Search Engine Land coverage – Tom Critchlow / Similarweb: recommendation-vs-citation and downstream demand – Frank Strong, “The end of the self-aggrandizing listicle…”. swordandthescript.com – Minuttia / Growthwaves: “Self-promotional listicles are due for a correction” (mechanics; FTC §465.6) – GA Agency: “Google’s Listicle Crackdown” (independent agency observation)

The pay-for-placement market (illustrative) – Productised “best of” listicle placement services advertising paid top spots from ~£200/$250 per placement (cited generically)

Figures supplied as three standalone SVGs (viz-three-referees.svg, viz-lawful-risky-unlawful.svg, viz-puffery-spectrum.svg), ready to embed as wp:html blocks in the scaffold.