Comparison Pages Are Legal, Cheap and Rare. Here Is the Template

The US Federal Trade Commission has encouraged advertisers to name competitors since 13 August 1979. That policy is still in force, still codified, and still ignored by most B2B software marketing teams. The legal exposure on a comparison page is real, but it sits almost entirely in one place, and it is not the place your legal team is worried about.

What a comparison page is, and what it is not

A comparison page names a specific competitor and sets your product against it on stated criteria. Examples of the search phrasing: "Bullhorn vs Greenhouse", "Zendesk alternatives", "cheaper than Salesforce".

The FTC's own definition is narrower and more useful than the marketing one. Comparative advertising is advertising that compares alternative brands on objectively measurable attributes or price, and identifies the alternative brand by name. The attributes have to be objectively measurable, and the rival has to be identified.

That rules out a large share of what B2B teams publish and call comparison content. A page listing ten tools with a paragraph each is a directory, not a comparison. A page claiming you are "more modern" is a positioning statement with a competitor's name pasted into it. If you are going to spend the meeting, spend it on measurable attributes.

Why most B2B firms still will not publish one

Three objections come up, in this order, and only one of them is serious.

The first is legal fear. Someone says naming a competitor invites a lawsuit. It has the least basis, and the section below dismantles it.

The second is a sales objection. Reps worry that a public page hands the competitor a document to rebut. That is true. It is also true that your reps already make these comparisons verbally, unsourced, with no version control and no legal review. A written page is the safer artefact.

The third objection is the real one. Publishing a comparison forces you to write down where you lose. Most teams have never agreed on that internally. The page is expensive because the meeting is expensive.

My position, stated plainly: the meeting is the deliverable. The page is a by-product. If your team cannot name three segments where the competitor is the better purchase, you do not have a product strategy, you have a feature list. A comparison page will expose that faster than a board deck will.

What the law actually permits when you name a rival

The FTC encourages it, in writing

The relevant text is short. Commission policy "encourages the naming of, or reference to competitors, but requires clarity, and, if necessary, disclosure to avoid deception of the consumer".

Two sentences in that policy are worth quoting to your legal team. First, "disparaging advertising is permissible so long as it is truthful and not deceptive". Second, industry codes that impose a higher substantiation standard on comparative claims than on your own unilateral claims "are inappropriate and should be revised".

The standard for saying "we are faster than Acme" is the same standard as for saying "we are fast". Naming Acme does not raise the bar. It only makes the claim easier for someone to check.

Using the rival's trademark is nominative fair use

The trademark question was settled in the US in 1992. In New Kids on the Block v. News America Publishing, 971 F.2d 302, the Ninth Circuit set a three part test for using someone else's mark to refer to their own product.

The product must not be readily identifiable without the mark. You may use only as much of the mark as is reasonably necessary. And you must do nothing that suggests sponsorship or endorsement by the mark holder.

A comparison page clears all three easily if you follow one rule: use the competitor's name as plain text, not their logo, not their brand colours, and not their wordmark lockup. The logo is where teams drift into the third prong without noticing.

Europe gives you a written checklist

If you sell into the EU or UK, Directive 2006/114/EC on misleading and comparative advertising sets out the conditions under which comparison is permitted. The one that governs a software comparison page is Article 4(c). The advertising must objectively compare "one or more material, relevant, verifiable and representative features" of the goods or services, which may include price.

Verifiable and representative are the two words to design around. Verifiable means a reader can check it from a public source. Representative means the feature you picked reflects normal use, not an edge case you engineered.

Article 4(d) adds that the comparison must not discredit or denigrate the competitor's marks, goods or services. That is a tone constraint with legal force attached, and worth respecting even outside the EU.

None of this is legal advice, and jurisdictions outside the US and EU differ sharply. Germany and France have historically applied the directive more strictly than the UK. Get the page reviewed before it ships.

The real risk is substantiation, not the name

The FTC's 1984 Policy Statement Regarding Advertising Substantiation requires a reasonable basis for a claim before it is disseminated. Not on request. Not when challenged. Before.

The evidence file is part of the draft, not part of the response to a complaint.

The case worth reading in full

On 7 January 2026 the National Advertising Division, part of BBB National Programs, decided a challenge brought by Human Interest against Guideline, Inc. over retirement plan advertising. Among the claims at issue were "Up to 6x less in fees" and a "4x Lower Asset Fees" claim naming Human Interest directly.

NAD found the naming itself unremarkable. What failed was the arithmetic. The "4x Lower Asset Fees" claim "conveys the broad message that customers of all types will realize significant savings by choosing Guideline over Human Interest, but due to different pricing models is not an 'apples-to-apples' comparison". NAD recommended the challenged fee claims be discontinued.

Nobody was sued. There was no injunction and no damages. A competitor filed with a self-regulatory body, the body read the substantiation, and the claim came down. That is the modal outcome, and it is a far cheaper failure than the one legal teams imagine.

What "apples-to-apples" means in software

Software pricing is exactly the trap NAD described. Per-seat against usage-based is not comparable without a stated workload. A starter tier against an enterprise tier is not comparable at all.

The distortions are structural, and we have covered how they arise in the analysis of seat compression in SaaS pricing and in the piece on credits, tokens and billing confusion. A comparison page that ignores those distortions is not being aggressive. It is being wrong.

The eleven block template you can ship this week

This is the structure. It is ordered so that the blocks a buyer needs come before the blocks you want them to read.

Blocks 1 to 4 are the answer. A buyer arriving from a search or an assistant has one question. Which one should I pick. Answer it in the first screen, name the segments, and disclose that you make one of the two products. A reader who finds the ownership halfway down discounts everything above it.

Blocks 5 to 8 are the evidence. Every row in the price table carries the plan name and the date you checked it. Every capability claim links to public documentation, preferably the competitor's own. Block 6 is the one that pays. Naming two segments where the competitor is the correct purchase is the cheapest credibility available to a vendor, and it is also the passage most likely to be cut in review.

Blocks 9 to 11 are the exit. Block 9 disqualifies badly-fitting buyers before they reach a rep, which is a sales efficiency gain rather than a lost lead. Block 11 is unglamorous and load-bearing. Set a calendar reminder at 90 days and treat a missed check as a page you take down, not a page you leave up.

The tone rules that decide whether it converts

Concede something that costs you. A concession only works if it is expensive. "Their onboarding is friendlier" is not a concession, it is a compliment that implies you are the serious tool. A real one names a segment and sends it away. "If you run more than 40 concurrent integrations, their architecture handles that and ours does not yet." That sentence loses deals, and it converts the ones it keeps at a much higher rate.

Never compare against a rival's worst plan. Comparing your top tier against their entry tier is the single most common failure, and it is precisely what NAD struck down in the Guideline decision. Pick the tier a real buyer in your target segment would actually purchase, and say which tier you picked. If the pricing models are structurally different, publish a worked example instead of a ratio. A ratio hides the assumptions. A worked example exposes them.

Then date everything, and say what you did not test. Every claim you could not verify gets a sentence saying so. The instinct is that admitting gaps weakens the page. It does the opposite. A page that says "we could not test their SSO enforcement under SCIM, so we have excluded it" reads as a page written by someone who actually tested the rest.

Writing it so a machine can quote it

The buyer increasingly meets your comparison inside an assistant rather than on your page. Gartner surveyed 646 B2B buyers between August and September 2025 and found 45% used AI tools during a recent purchase, alongside 67% who said they prefer a rep-free experience. The shortlist effect is covered in the analysis of where the B2B shortlist now forms.

The implications are mechanical. Use a real HTML table, not an image or a grid of styled divs. Put the verdict in the first 60 words. Write FAQ questions in the phrasing a buyer types, not the phrasing your brand guide prefers.

There is one piece of peer-reviewed evidence worth acting on. The GEO study presented at ACM SIGKDD 2024 tested nine content strategies across a 10,000 query benchmark. Its top three, citing sources, adding quotations and adding statistics, produced a 30% to 40% relative improvement on the paper's Position-Adjusted Word Count metric. Keyword stuffing performed worse than doing nothing.

That result is a research finding on a benchmark, not a guarantee about any commercial engine in 2026. What it supports is modest and useful: a page dense with dated, sourced, checkable figures is structurally advantaged over one that is not. The broader evidence on citation behaviour is set out in the piece separating GEO evidence from GEO marketing. What it does not show is covered in the analysis of how ChatGPT citations relate to Google rank.

Where this argument is weakest

The citation evidence is mostly vendor research

The claim that comparison pages are disproportionately cited by AI systems is repeated everywhere in 2026. Figures like "85% of B2B AI citations come from third-party sites" and "90% do not come from brand-owned domains" circulate widely.

I went looking for the underlying study and could not find one I would cite. The percentages appear in vendor blog posts that reference other vendor blog posts. One heavily cited source article, published by a search tooling company in April 2026, contains no such percentage in its own text.

Six restatements is one source, and in this case it may be zero sources. Treat those numbers as unverified. The argument for comparison pages in this post rests on FTC policy, EU law, one NAD decision and one peer-reviewed paper, all of which you can open yourself.

A comparison page can genuinely lose you deals

If your product is meaningfully behind on the criteria a buyer weights most heavily, an honest comparison page tells them so and they leave. That is a real cost.

The counterargument is that they were going to discover it during evaluation anyway, later, after consuming sales time. That holds in most cases and not in all. If you are 6 months from closing a genuine capability gap, waiting 6 months is a defensible call.

The format is saturating

Comparison pages are cheap to generate at volume, which means the format is filling up fast. A page that is generated rather than tested will be indistinguishable from the rest within a year. The volume problem is examined in the piece on AI content saturation and search visibility.

The defensible version is the one carrying something only you could produce. A worked pricing example from a real workload. A migration you actually ran. A dated test result. Structure is copyable. Testing is not.

Frequently asked questions

Where to start this week

Pick the single competitor your reps name most often in lost-deal notes. One page, one rival, not a matrix of nine.

Then run the meeting before you run the draft. Agree the criteria and their weights before anyone opens either product. Agree the two segments where the competitor is the correct purchase. If that takes more than an hour, the page is not the problem you have.

Write the verdict paragraph, the criteria table and the concession section first. Those three blocks are the page. If they cannot be written honestly, publishing the rest will not help.

Related reading