How should a SaaS startup build a comparison page against an established competitor that buyers trust and lawyers can live with?

Illustrative situation: a founder publishes “Us vs. Them” the night before a launch. The startup’s column shows green checks. The competitor’s column shows red crosses, including one for a feature it shipped last quarter. A prospect forwards the page to the competitor’s account manager, who replies with a link to the release notes. The prospect stops answering emails. The scene is invented. It shows the mistake this guide helps avoid: a page built to win a glance, not to survive a check.

Humbleteam, a digital product design agency, publishes this guide. It covers who reads the page, choosing rows, the evidence each row needs, showing where the competitor wins, the switching section, page structure and a phone-sized table, upkeep and the main legal limits in the US, UK and EU, with a decision table and an invented worked example. Sources and product behavior are described as of November 9, 2026 and can change. This is not legal advice. Laws differ by country and by claim, so have a lawyer who handles advertising review the page before you publish it.

Who lands on this page, and what are they deciding?

Our reading, not measured data: three people read it.

• The evaluator already uses the established product or has it on a shortlist. The question is whether the gap justifies a switch for their team.
• The defender must justify the choice to finance, security or a boss, and copies rows into an email.
• The checker works at the competitor, or advises it, and opens every source link.

Write for the checker first. A row that survives that reader survives the other two. Our guide to running a UX teardown of an established competitor covers finding where your product differs. This guide covers the page that reports it.

Which rows belong in the table?

Start from the buyer’s job, not your feature list. Write each row as a question the buyer must answer: “Can a manager approve a week of timesheets in one sitting?” beats “Bulk approval.” Our limits: six to eight rows, at least two where you tie or lose, and no row you cannot source or test.

Two legal tests point the same way. The UK’s CAP Code asks that a comparison cover products meeting the same needs and compare material, relevant, verifiable and representative features (rules 3.33 and 3.34). EU Directive 2006/114/EC has both conditions in Article 4. In practice, compare the plan a buyer in your segment would actually buy, and do not keep only the rows you win.

What evidence does each row need?

Our rule: no row goes live without a source you can open, a capture date and the plan compared. The FTC’s substantiation policy says advertisers need a reasonable basis for a claim before it runs. CAP rule 3.7 asks for documentary evidence before publication.

How do you show where the competitor is better?

Name at least two places. Two real pages, both read on November 9, 2026, show the pattern. Plausible’s comparison with Google Analytics 4 has a section titled “What GA4 does better” and another listing needs Plausible does not cover. We saw no date on it and no link to Google’s own documentation. PostHog’s comparison with Mixpanel carries a January 28, 2026 byline date, a price table and a team-type section that recommends Mixpanel for product-management teams. We found no link to Mixpanel’s own site in the page source.

Both pages admit a loss somewhere. Neither shows the evidence standard above. We would add a short “Choose them if” block with two or three conditions, worded the way a buyer would put them. It keeps the table representative. It is not a legal shield: a true row where they win does not make a false row where you win true.

What should the switching section say?

Say what moves, what does not, who does the work and how long it takes. Give a time only if you have run that migration, and say how many times. Promise no more than your import tool does. Our guides cover designing the import, in switching costs and migration UX, and the weeks after it, in keeping power users fast after a switch.

How should the page be built, and how does the table work on a phone?

Our order: a title naming both products in plain text with the review date and owner’s role under it, two sentences on who each product fits, the table, notes per row with source and capture date, “Choose them if,” switching, then method and a way to report an error.

On a phone, Nielsen Norman Group says wordy comparison cells may fit only two columns on a narrow screen, and that a mobile comparison table is unlikely to show more than two items. So compare exactly two products. A row label makes three columns, which wordy cells may not survive, so we keep cells to about ten words. If readers must still scroll sideways, NN/g advises locking the first column and keeping headers in place. A stacked layout, with the label above two short lines, is another option. We have not tested it, and NN/g’s article does not cover it. Show the evidence link under each row, not in a hover tooltip, and check the page at 375 px wide. One competitor per page also keeps legal review manageable.

What are the legal limits?

These are the rules we read. They are not a checklist for your case.

• In US policy, the FTC’s 16 CFR 14.15 encourages naming competitors but requires clarity and, where needed, disclosure to avoid deception. It allows disparagement that is truthful and not deceptive. So state what you compared: plan, date and test.
• On false advertising, Lanham Act section 43(a)(1)(B), 15 USC 1125, makes liable commercial advertising that misrepresents the nature, characteristics or qualities of your goods or another person’s. In Lexmark v. Static Control (2014) the Supreme Court said a plaintiff must allege an injury to a commercial interest in reputation or sales.
• On using their name, comparing by name is called nominative use. For famous marks, the statute’s dilution section excludes it in advertising that lets consumers compare. For confusion claims, courts differ. The Third Circuit’s Century 21 v. LendingTree (2005) asks whether the mark is necessary to describe both products, whether you use only as much of it as necessary, and whether your wording reflects the true relationship. It disagrees with the Ninth Circuit on how the defense works. Our reading: name them in plain text, with no logo, no look-alike styling and no wording that suggests a partnership.
• In the UK, regulation 4 of the Business Protection from Misleading Marketing Regulations 2008 permits comparative advertising only when, among other conditions, it is not misleading, compares products meeting the same needs, creates no confusion, does not denigrate, and takes no unfair advantage of a competitor’s trade mark. CAP Code rules 3.32 to 3.35 and 3.41 to 3.43 are similar, and rules 3.44 to 3.46 cover fake and misleadingly published consumer reviews.
• In the EU, Directive 2006/114/EC, Article 4, sets conditions of the same kind.
• On review sites, G2’s terms cover G2, Capterra, Software Advice and GetApp, and bar scrapers and commercial redistribution of site content without consent. G2’s guidelines list republishing another vendor’s reviews or aggregated star rating as prohibited. Keep competitor ratings and quotes off your page.

Add a short note saying who owns the product names and when you verified the rows. Our view: it helps clarity, and it does not cure a false row.

How do you keep it current?

Put “Last reviewed,” the next review date and the owner’s role at the top. Our defaults: review quarterly, and re-check a row when the competitor changes pricing or plan names, ships a release that touches it, or your own product changes that flow. Name one owner and a backup.

What breaks pages: prices that vary by region or appear only after login, renamed plans, screenshots that age without notice, and your own releases. If a row is not re-verified by its due date, remove the row. Do not just move the date forward. If you receive a complaint or a letter, pause the row, correct or remove it, and call your lawyer.

Illustrative worked example: Ploverfield and Carrowmet

Everything here is invented. Ploverfield is a fictional timesheet-approval startup. Carrowmet is a fictional established time-tracking suite. The buyer is an operations lead at a 40-person agency.

Ploverfield drafts 10 rows from the buyer’s job. The evidence check leaves 7 backed by Carrowmet’s own documentation or pricing page, captured November 2, 2026, and 2 resting on Ploverfield’s own timed test with the method published. One has no public source, so it is dropped. Nine rows stay: Ploverfield ahead on 4, even on 2, behind on 3. Three losses meet the two-loss rule.

The price row needs care. Carrowmet’s Business plan is $12 per user per month and includes single sign-on. Ploverfield’s Team plan is $9 with no single sign-on, and its Scale plan is $14 with it. For 40 users, Team against Business is $3 × 40 × 12 = $1,440 a year cheaper. A buyer who needs single sign-on compares Scale against Business: $2 × 40 × 12 = $960 a year more. The page shows both lines and names the single sign-on condition. A page that showed only $1,440 would compare plans that do not meet the same needs.

The top of the page reads: last reviewed November 9, 2026; next review February 9, 2027; owner, head of marketing.

Where do these rules stop working?

We read the two example pages once on November 9, 2026. We describe what we saw, not whether the pages work. We found no research on how buyers use comparison pages or whether they convert, and we give none.

On law, we read the statutes and FTC statements cited here, the Supreme Court syllabus and one appeals opinion. We read the EU directive through the UK government’s published copy of the original text, not EUR-Lex, so later EU changes are unchecked. We did not read the Ninth Circuit opinion, other circuits, state law, defamation or trade libel, or law outside the US, UK and EU. We could not open Capterra’s own terms, and we did not check whether the CAP Code reaches a given business-to-business page.

The numbers are ours: six to eight rows, two losses, ten words per cell, quarterly reviews, 375 px. NN/g’s mobile article dates from 2017, and we did not test layouts. Usage-based or regional pricing may not fit a one-line row. A two-product table does not serve a buyer comparing three.

What can you produce now?

1. Ten candidate rows, each written as a buyer question.
2. A filled evidence table: source URL, capture date and plan compared for every row.
3. A “Choose them if” block with two or three conditions.
4. A switching paragraph: what moves, what does not, who does the work and how long, or a plain statement that you have not measured it.
5. A page outline and a 375 px draft of the two-product table.
6. A review sheet with the owner, the next date, the triggers and the rule to remove any row that is not re-verified.
7. A question list for your lawyer: the countries you sell in, any logos or review-site content you plan to use, and the exact wording of the rows.

If you want a design partner for the page and the product behind it, our comparison of agencies that publish B2B SaaS platform cases shows what each one publishes.