Skip to content
Chalk Labs / AI SEO
How-to guide

How to Write Comparison Pages That Get Cited

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

A comparison page gets cited when it gives a clear verdict first, compares options on the same stated criteria in a table, shows first-hand evidence, and discloses your own interest. Name your own product honestly, including where it loses. Add a visible last-updated date and refresh the facts on a schedule, because stale tables lose trust.

  • Lead with the verdict, then a table with identical criteria for every option, including your own.
  • Self-ranked lists work in AI answers, but the evidence is thin and the risk of losing trust is high.
  • Show first-hand evidence and a method: what you tested, when, and what you could not test.
  • Disclose your interest in the page, and keep a visible update date with a change log.

Why do comparison pages matter for AI search?

Comparison questions are where buyers decide, and AI answers lean on outside sources for them. In an Ahrefs test of 20,000 US prompts, official sites got 63.4% of citations for fact questions but only 46.4% for comparison questions 1. More than half of comparison citations went elsewhere 1.

Ahrefs also found 'best X' blog lists made up 43.8% of cited page types across 750 top-of-funnel prompts for software, products and agencies 2. In a 1,474-answer ChatGPT sample for running shoes, the word 'vs' appeared in 105 fan-out searches 1. Fan-out searches are the extra queries an AI tool fires off to build one answer. These figures show where citations go and what searches contain, not that a comparison page will be cited. The running-shoes sample covers one category.

Do it step by step

0 of 8 done0%

  1. Write the decision in one sentence: 'Which accounting tool for a 10-person agency?' Everything on the page should help answer it.

  2. List 6 to 8 criteria that matter to that buyer, such as price, setup time, integrations and support. Decide how you will check each one.

    Tip: Publish the criteria on the page so readers can see you did not pick them after the fact.

  3. Take prices and features from each vendor's own page, with the date. Test what you can yourself and note what you did.

    Tip: Screenshot the pricing page on the day you read it.

  4. Open with two sentences on who wins for which buyer. Add one line saying who you are and any interest you hold.

  5. Use the same rows for every option, with short, factual cells. Put your own product in the same order as the rest.

  6. Cover fit, price, strengths and limits in that order. Include at least one real weakness for your own product.

  7. State what you tested, when, and what you could not test. Add the date and a one-line change note.

  8. Put a quarterly reminder in your calendar. At review, re-check every price and feature, then update the log.

Your ticks are saved in this browser only.

Copy-ready templates

Comparison table skeleton

markdown
| Criterion | Option A | Option B | Option C |
|---|---|---|---|
| Best for | [buyer type] | [buyer type] | [buyer type] |
| Starting price (checked [date]) | [price] | [price] | [price] |
| Setup time | [time] | [time] | [time] |
| Key limit | [limit] | [limit] | [limit] |
| Support | [hours, channels] | [hours, channels] | [hours, channels] |
| Tested by us? | [yes/no, date] | [yes/no, date] | [yes/no, date] |

Use the same wording style in every column. Put your own product in the same row format as the rest.

Disclosure line

text
Disclosure: [Company] makes Option A. We wrote this comparison using the criteria below, checked prices on [date], and tested [list] ourselves. We may earn a commission from some links, marked [sponsored].

Place this directly under the headline, not in the footer.

Update log entry

text
Updated [date]: [Option B] raised its starting price from [old] to [new]. Verified on its pricing page on [date]. Verdict unchanged.

Keep the log on the page and change the visible date only when facts change.

Can you put your own product first?

You can, but do it openly and only where you earn it. Ahrefs found that in ChatGPT answers where a publisher was recommended in software, its own list appeared in 34% of them 2. In 169 of 250 Google searches for best software, the list author ranked itself first 2.

Ahrefs describes the study as a benchmark and says it is ongoing, but still concludes that agencies and SaaS firms should publish self-promotional lists, and its own comparison posts are under half a percent of its blog 2. Treat that as one vendor's reading of correlational data, not a guarantee. A second Ahrefs experiment found self-promotional pages helped a new conference brand far more than an established tool, and about a quarter of citations never returned 3. A page that treats rivals fairly is a safer long-term asset.

What structure should the page follow?

Put the answer in the first paragraph, then the table, then the detail. A reader should know who wins for which buyer within ten seconds. Each section should make sense if quoted alone, because AI tools lift passages, not pages.

Use the same order for every option: what it is, who it suits, price, strengths, limits, and one verified fact. Equal treatment is what makes the page read as fair and what makes the table machine-readable.

A page skeleton that works for most comparisons
BlockWhat goes in itWhy it helps
VerdictTwo sentences: best for X, best for YQuotable answer
DisclosureWho wrote it and what interest they holdTrust and legal clarity
TableSame criteria for every optionEasy to scan and extract
Option sectionsFit, price, strengths, limitsFair, complete detail
MethodWhat you tested, dates, gapsEvidence of first-hand work
Update logDate and what changedShows the page is current

What counts as evidence on a comparison page?

First-hand testing and numbers that readers can check. Google's guidance for reviews asks for evidence of your own experience, quantitative measurements, what sets an item apart from competitors, and which alternative suits which use 4. It also asks you to explain why something is best, with first-hand evidence 4.

Google adds that ranked lists should have enough useful content to stand alone, and that quality and originality beat length 4. Its AI search guide says commodity content, such as a list of tips based on common knowledge, adds little unique insight 5. If you did not test something, say so and cite where the claim came from.

How do you keep a comparison fair?

Fix the criteria before you look at the results. Write them down, weight them if you must, and publish them. Then give your own product the same scrutiny as the rest, including at least one real weakness.

State where a competitor wins. A line such as 'choose the other tool if you need offline mode' costs you little and is exactly the sentence a buyer, and an AI summary, will repeat. Ask a competitor to check the facts about their product if you can, and correct errors quickly.

What must you disclose?

Say who you are and what you gain. If you sell one of the options, say so at the top, in plain words. The FTC says a connection that consumers would not expect should be disclosed clearly, and that free products given in the hope of promotion count too 6.

Affiliate links, paid placements and gifts all need a visible note near the claim, not a footer link. Also remember that Google asks about the 'why' of content: pages made mainly to help people, not to attract search visits 7.

How often should you update a comparison page?

Review prices and key features every quarter, and sooner when a vendor announces a change. In Ahrefs' sample of 1,100 dated best lists, 79.1% were last updated in 2025 and 26% in the previous two months 2. Freshness is common among the pages that get cited.

Profound found comparisons, listicles and reviews had a median citation half-life of 29 days, slightly longer than other page types, but still short 8. Treat the page as a living document. Change the visible date only when facts change, and note what changed.

What are the common mistakes, and when should you skip this?

The common mistakes are a verdict hidden below the fold, criteria that only your product meets, claims copied from vendor pages and a date that never changes. Another is making one page per rival pair with nearly the same text, which can look like scaled content; see programmatic pages without spam.

Skip comparison pages when you cannot test or verify the options, or when you would have to hide your own interest. In that case, publish a clear pricing page and let others write the comparison. For the decision itself, see Reddit vs your own blog.

Frequently asked questions

What is a comparison page?

A page that sets two or more options side by side on shared criteria, such as price and fit, and says which suits which buyer. It can be a head-to-head, a best-of list, or an alternatives page.

Should I name competitors on my own site?

Yes, if you can be accurate and fair. Buyers already compare you, and AI tools cite outside pages for comparison questions 1. Name them, cite their pages, and say where they win.

How long should a comparison page be?

Long enough to cover the criteria with evidence. Google says quality and originality matter more than length 4. A clear verdict, a table and short sections usually beat a long essay.

Do I need schema for a comparison table?

No special markup is needed for Google's AI features 5. Use clean HTML tables with clear headers and keep any structured data matching the visible text.

How do I avoid looking biased?

Publish your criteria, disclose your interest at the top, give your own product at least one real weakness, and let a rival check the facts about their product.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. How ChatGPT Does Query Fan-Out (15 Million Queries Studied) · AhrefsOpened October 20, 2026. Supports: 20,000 US prompts: official sites got 63.4% of citations for fact questions and 46.4% for comparison questions; in a 1,474-response running-shoes sample, 'vs' appeared in 105 fan-outs.
  2. Do Self-Promotional 'Best' Lists Boost ChatGPT Visibility? Study of 26,283 Source URLs · AhrefsOpened October 20, 2026. Supports: Published 4 December 2025: best lists were 43.8% of cited page types; self-promotional list shares; 79.1% of 1,100 dated lists updated in 2025; author's own caveats.
  3. Self-Promotional Content Works, Until It Backfires (AI SEO Experiment) · AhrefsOpened October 20, 2026. Supports: 34 self-promotional pages, 9,886 AI answers, Feb to May 2026: helped a new brand far more than an established one; about a quarter of citations appeared once and never returned.
  4. Write high quality reviews · Google Search CentralOpened October 20, 2026. Supports: Guidance: evidence of experience, quantitative measurements, alternatives, why something is best, ranked lists should stand alone, quality over length. Last updated 2025-12-10.
  5. Optimizing your website for generative AI features on Google Search · Google Search CentralOpened October 20, 2026. Supports: Commodity content, such as a list of tips based on common knowledge, adds little unique insight; no special schema required for generative AI search. Last updated 2026-07-10.
  6. FTC's Endorsement Guides: What People Are Asking · U.S. Federal Trade CommissionOpened October 20, 2026. Supports: Connections consumers would not expect should be disclosed clearly; free products with an expectation of promotion require disclosure.
  7. Creating helpful, reliable, people-first content · Google Search CentralOpened October 20, 2026. Supports: The 'why' question: content should be made primarily to help people; product reviews should disclose how products were tested. Last updated 2026-10-05.
  8. The half-life of an AI citation is 11 days · ProfoundOpened October 20, 2026. Supports: 883,000 pages, Sep 2025 to Sep 2026: comparisons, listicles and reviews had a 29-day median half-life among sustained pages versus 27 to 28 days for other types; vendor study.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.