A trademark risk comparison blog post carries real exposure: nominative fair use limits, disparagement traps, and the legal check to add before you merge.
Comparison and alternative pages are the highest-converting format on most SaaS blogs, and they're also the format most likely to name a real competitor by name, on purpose, over and over. That combination is why a trademark risk comparison blog post deserves a specific legal check before it merges, not the same fact-check pass you'd run on a tutorial. A tutorial rarely mentions another company's brand at all. A comparison post is built around one.
We've written about what makes comparison and alternative pages convert at over 7.5%, well above general blog traffic, according to Grow and Convert's analysis of comparison-page traffic. That number is also the reason to add a legal review step instead of avoiding the format: the content converts too well to skip, and the legal exposure is manageable if you check the right three or four things before publishing, not after a competitor's counsel emails you. This post is that check.
A how-to guide or an industry roundup rarely names another company as its subject. A comparison or alternative post is built entirely around one, sometimes several, and names it in the headline, the URL, and every section. That structural difference is what creates the exposure a generic post doesn't carry.
Three distinct legal theories can attach to the same comparison post, and they don't require the same proof.
Trademark confusion is the classic infringement claim: using a mark in a way that makes a reader think the competitor sponsored, endorsed, or is affiliated with your content. Disparagement, legally closer to trademark tarnishment, doesn't require confusion at all. It's triggered by associating a competitor's mark with something negative or low-quality, and courts can find it actionable even when no reasonable reader would think the two companies are affiliated. False advertising under Section 43(a) of the Lanham Act covers a third, separate ground: making a false or misleading factual claim about a competitor's product, whether or not their trademark is involved in a confusing way. Only a business with a commercial injury has standing to bring that claim, not an individual reader who felt misled.
A single comparison post can trip more than one of these at once. Naming a competitor in a way that implies partnership is confusion. Calling their product "broken" without evidence is closer to disparagement. Claiming your uptime beats theirs by a specific percentage you never tested is false advertising. They're separate legal theories with separate elements, but they share one root cause: a claim or a presentation choice nobody checked before it shipped.
A model drafting a comparison post carries a risk a human writer doesn't introduce the same way: it can generate a claim, or even a mark, that resembles a real brand without anyone deliberately choosing it. Brian M. Wheeler, partner and chair of the Commercial and Complex Litigation Practice Group at Atkinson, Andelson, Loya, Ruud & Romo, put it plainly: "It is imperative that AI users monitor and review in detail all AI outputs before using them in commerce." His firm's analysis of trademark risk in generative AI content notes that a model can also produce a derogatory association a human writer wouldn't have proposed on their own, simply by pattern-matching toward confident-sounding but unverified phrasing.
The mechanism is the same one behind AI hallucination generally: a model writing from memory instead of a checked source will produce a fluent, specific-sounding claim, "benchmarks show X is 40% slower," with nothing behind it. That's a fact-checking failure in a how-to post. In a comparison post, it's also a legal one, because the claim is about a named competitor, not an abstract fact. We cover the broader mechanics of catching this in our AI content fact-checking workflow; a comparison post needs that same discipline pointed specifically at every sentence that mentions a competitor by name.
Nominative fair use is the doctrine that lets you name a competitor's trademark without their permission, but it's a narrow test with three specific conditions, not a blanket license to say anything about them.
The foundational case is New Kids on the Block v. News America Publishing, Inc., 971 F.2d 302 (9th Cir. 1992). It set out three conditions a use has to satisfy:
That structure comes directly from the case as summarized in the nominative use doctrine's overview: naming the mark to identify the actual product is protected; using it to borrow the mark holder's goodwill is not.
The three-part test is consistent, but circuits disagree on how it fits into a trademark case procedurally, and that matters if you'd ever need to defend a post. The Ninth Circuit, which decided New Kids, later folded the test directly into its likelihood-of-confusion analysis in Toyota Motor Sales v. Tabari, 610 F.3d 1171 (9th Cir. 2010), replacing the standard confusion factors rather than adding to them. The Third Circuit treats nominative fair use as a true affirmative defense: the plaintiff first proves likelihood of confusion under the ordinary test, and the burden then shifts to the defendant to prove the use was fair.
The Second Circuit rejected both approaches in International Information Systems Security Certification Consortium v. Security University, 823 F.3d 153 (2d Cir. 2016). As Arnold & Porter's analysis of the circuit split explains, the Second Circuit requires courts to run the standard confusion factors and separately weigh the three nominative-use questions alongside them, rather than substituting one test for the other or treating nominative use as a distinct affirmative defense. None of this changes what a well-drafted comparison post should look like in practice. It changes how a dispute over one would actually get litigated, and which circuit you're in decides that.
The three-part test is about word marks used in prose. It doesn't extend the same protection to visual branding. Using a competitor's logo, their exact font, or their brand colors reads to a court as a much stronger signal of implied endorsement than typing their name in plain text, and Harrigan IP's breakdown of nominative fair use is direct about it: "nominative fair use is a narrowly tailored defense, not a loophole." Courts are also more sensitive to a mark's placement, not just its presence, weighing repeated use in a headline or a URL slug more heavily than a single mention in body copy. None of that means don't put a competitor's name in your title, comparison-page titles do exactly that. It means the visual and structural choices around the name, not just the words, are part of what a court weighs.
Nominative fair use governs whether you can name a competitor. It says nothing about whether what you say about them is legal, and that's a separate, second check.
Courts draw a sharp line between two kinds of comparative claims, and the line determines what evidence a plaintiff needs to win. A literally false claim, one that's false on its face, is actionable without any survey or consumer evidence at all: the plaintiff just proves it's false. An implicitly misleading claim, one that's technically true but creates a false impression, requires the plaintiff to bring extrinsic evidence, usually a consumer survey, showing a substantial share of the audience was actually deceived. Legal Clarity's breakdown of comparative advertising law lays out both standards.
That distinction is exactly why the two example sentences above sit in different risk tiers. "Their tool is clunky" is a subjective opinion nobody can objectively disprove. "Independent tests prove it's three times slower" is a specific, falsifiable factual claim, literally false if no such tests exist, and it needs a citation before it ships, the same discipline any other statistic in the post requires.
A comparison that's technically accurate in every sentence can still be misleading in aggregate, if the specific metrics chosen are the ones that happen to favor you and the ones that don't are omitted entirely. This is the implicitly-misleading trap: no individual claim is false, but the overall impression a reader takes away, "X wins on everything," is one the full picture wouldn't support. It's also the fastest way for a comparison post to drift from a fair review into something closer to a pitch dressed up as neutral, a pattern this blog has covered from the SEO side in SaaS comparison pages. Google's own scaled content abuse policy defines the violation as pages "generated for the primary purpose of manipulating search rankings and not helping users," and a comparison engineered so you win on every listed metric is one concrete way a page ends up serving the ranking instead of the reader, independent of whether any individual sentence is false. The honest-frame pages that name a competitor's real strengths are the ones that hold up, on both fronts.
Vague, subjective superiority language, "the better choice," "the best alternative," is protected as puffery, because it can't be measured and a reasonable reader doesn't treat it as a factual assertion. The line moves the moment a claim becomes specific and testable. "50% faster" or "tests prove" needs substantiation on file before publishing, not a plan to find the source if someone asks. Legal Clarity's guidance is direct on this split, and it's the single cheapest check to run on a comparison draft: read every number and every superiority claim, and sort each one into "opinion" or "needs a source." Anything in the second bucket without a citation gets rewritten or removed before the post merges.
The stakes for skipping that check aren't abstract. The FTC can pursue penalties for deceptive comparative advertising that reach $53,088 per violation as of 2026, the statute's ceiling after annual inflation adjustments, on top of cease-and-desist orders and mandatory corrective advertising. And trademark litigation itself isn't cheap even when a company wins: citing the American Intellectual Property Law Association's own economic survey, IP Bureau puts the cost of a trademark lawsuit that goes to trial at $350,000 to $600,000. A ten-minute check on a draft is cheap against that number.
None of the theories above require a lawyer to read every comparison post before it ships. They require someone to ask a short, specific set of questions, on the same pull request that already carries the fact-check.
A comparison-post legal check fits in five questions, run as part of the same review pass that already checks facts and links:
| Check | What it catches |
|---|---|
| Is every superiority claim opinion, or does it need a source? | Puffery vs. an unsubstantiated claim that needs substantiation on file |
| Is the competitor's logo, font, or brand color used anywhere? | Visual signals that undercut nominative fair use even when the words are fine |
| Does the post name the competitor more than necessary to identify them? | Overuse in headlines, URL slugs, and repeated mentions that courts weigh against you |
| Does any sentence imply partnership, sponsorship, or endorsement? | Trademark confusion, independent of whether the facts are accurate |
| Is the comparison one-sided, with no acknowledgment of where the competitor wins? | The omission problem that turns an accurate post misleading in aggregate |
This is the same shape of gate this blog has argued for elsewhere: a specific, named check that runs on the pull request, not a general reminder to "review carefully." Our post on GitHub Actions checks for blog PRs covers automating the mechanical version of this, broken links and schema validation, as a required status before merge; the legal check above is the same idea applied to a judgment call a script can't make on its own.
A legal check that lives in a Slack message from three weeks ago isn't a check, it's a memory nobody can point to later. It has to be a named, visible comment on the pull request that added or edited the post, tied to a specific commit, the same way an audit trail needs a record of who reviewed what and when to mean anything under scrutiny. Who runs it depends on team size: at a small company it's whoever already owns the merge button, the same person this blog's ownership model for AI blog automation names as accountable for technical claims generally. At a larger company with in-house or outside counsel, the checklist above is what that reviewer hands to counsel instead of the full draft, so legal only spends time on the claims flagged as needing a source, not the whole post. None of this post is legal advice, and it isn't a substitute for counsel reviewing a specific post: it's the checklist that tells you which posts are worth that call.
Weigh the two numbers from this post against each other. A trademark suit that goes to trial costs $350,000 at the low end, before a single FTC penalty is added on top. The check that prevents most of that exposure is a five-question pass on a pull request that already exists, run by someone who already reads the draft before it merges. As Harrigan IP put it, nominative fair use "is a narrowly tailored defense, not a loophole," and the same is true of the whole legal check: it's not a blanket protection that lets you say anything about a competitor as long as you use their name correctly. It's a specific set of questions that, asked once per post, keeps a comparison page's 7.5% conversion rate from ever meeting a $350,000 legal bill.
Lyra opens every comparison post as a pull request with the fact-check already attached, which is the same surface where a trademark and disparagement check belongs before you merge.
FAQ
Generally yes, under a doctrine called nominative fair use, as long as you can't identify the product without naming it, you use only as much of the mark as necessary (the words, not the logo or stylized font), and nothing implies the competitor sponsors or endorses your content. The doctrine comes from New Kids on the Block v. News America Publishing, 971 F.2d 302 (9th Cir. 1992), and it protects naming a competitor to compare against, not borrowing their goodwill.
It's the legal doctrine that lets you name a competitor's trademark to identify their actual product, rather than needing their permission every time you mention them. The Ninth Circuit's three-part test asks whether the product can be identified without the mark, whether you used only what's necessary, and whether anything suggests endorsement. Circuits differ on how the test is applied procedurally, but the underlying three questions are consistent.
Yes. Section 43(a) of the Lanham Act doesn't ban naming a competitor in advertising or content, it bans false or misleading claims about anyone's goods or services, including a competitor's. A truthful, substantiated comparison is legal. A claim that's literally false, or true but misleading in the impression it leaves, is what creates liability, and only a business with a commercial injury, not an individual reader, can sue over it.
Disparagement (legally, trademark tarnishment or product disparagement) happens when a claim damages a competitor's reputation rather than just noting a difference. Calling a competitor's tool 'clunky' is a subjective opinion, protected as puffery. Claiming 'independent tests prove it's three times slower' without a source is a specific, checkable assertion, and if it's false or unsubstantiated it can support both a false-advertising claim and a tarnishment claim, since tarnishment doesn't require proving anyone was confused about which company is which.
Built by the tool you're reading about
Lyra finds the topics worth ranking for, writes them in your repo's voice, fact-checks every claim, and opens a pull request scored and ready to merge. You review and hit merge. Want to see what she'd write for you? Start free with three posts, no card.
Keep reading
Multi-agent content review pairs a generator with an independent critic and judge so mistakes get caught before a human ever opens the pull request.
AI content governance means proving who wrote a post, who reviewed it, what changed, and when it shipped, an audit trail a Git pull request already gives you.
AI blogs sound generic without a real guide. What belongs in a brand voice AI content style guide, and how to keep it current as your product evolves.