Your "us versus them" page has been one of your better-performing organic assets for two years. Ask an assistant to compare the two products and it cites a forum thread, a review site, and a competitor's page. Yours does not appear. The page is technically fine, indexed, and fast.
The likely reason is what the page is for. It was written to close a deal at the bottom of the funnel, which means it was written to win. A page that only reaches one conclusion is a poor source for anything trying to summarise a genuine trade-off, and it is a poor source for a careful human reader too.
Disclosure, since it is directly relevant: RankRift publishes vendor-neutral guidance on evaluating AI-visibility trackers and does not claim an owned product. That is a structural conflict and we set out how we handle it on our disclosure page. The advice below is the standard we hold ourselves to, and you can hold us to it.
Why a one-sided comparison is a weak source
Think about what a summariser needs. Someone asked which of two options fits their situation. To answer, the system needs statements that survive being quoted: specific, conditional, attributable, and not obviously produced by an interested party trying to reach one conclusion.
A typical vendor comparison page fails on most of those counts:
- Every criterion in the table is one the publisher wins. The selection is the argument, and the selection is unstated.
- The competitor's row is thin, out of date, or describes a plan the reader was not considering.
- There is no condition under which the reader should choose the other option, which is a claim no honest comparison can make.
- Nothing is dated, so a fragment quoted next year is unverifiable.
- The conclusion is asserted rather than derived from the criteria above it.
None of this is about deceiving a machine. A page like that is weak evidence for a human evaluator as well. The overlap between "content a careful reader trusts" and "content that is safe to quote" is most of the surface area.
There is a second, non-technical reason to care. Statements about a named competitor's product carry real exposure. That is a general observation and not legal advice, and anything you publish comparing yourself to a competitor should be reviewed by qualified counsel in your jurisdiction, particularly where advertising rules apply.
The structure of a quotable comparison
A workable shape, in order:
- Scope and date. What is being compared, which versions or plans, and when the information was checked.
- How we compared. The criteria, why those criteria, and how each was assessed. Hands-on, vendor documentation, or not assessed.
- The short answer by segment. Two or three sentences, each naming a reader type and the option that fits it. This is the block most likely to be quoted, so write it as if it will travel alone.
- The comparison table. Stable columns, labelled rows, units in cells.
- Where each option is stronger, in its own section, with the reasoning.
- Who should choose the other one, including yours.
- Who should choose neither, which is a real category more often than vendors admit.
- Disclosure and limits. Who wrote this, what they sell, what was not tested.
- How to verify it yourself. The test a reader can run in an hour.
Each of those sections should hold up if it is extracted alone, which is the general rule covered in chunk-friendly structure. A comparison page is the hardest case for that rule, because a fragment of a comparison quoted without its scope is exactly how a fair page becomes an unfair quote.
Stable columns and why they matter
Pick your criteria once, define them, and keep them identical across every comparison page you publish and across every revision of each page.
Stability matters for three reasons. It lets a reader compare across your pages instead of only within one. It prevents the criterion set from silently becoming whatever the current subject wins on. And it keeps your own fragments consistent with each other over time, so a row extracted from the 2026 version does not contradict a row extracted from the 2027 version.
A criteria set for a software comparison might look like this. Yours will differ; the discipline is that it is written down before you evaluate anything.
| Criterion | Definition used | How assessed | Last checked |
|---|---|---|---|
| Coverage | Which systems or sources the product handles | Hands-on account, documented list | Date |
| Method transparency | Whether the vendor publishes how numbers are produced | Public documentation only | Date |
| Data export | Formats available and whether raw records are included | Hands-on | Date |
| Update cadence | How often data refreshes, as stated by the vendor | Vendor documentation, labelled as such | Date |
| Onboarding effort | Time and access required before first useful output | Hands-on | Date |
| Pricing model | Structure of charging, linked to the vendor's page | Vendor page, no amounts stated | Date |
| Support model | Channels and stated response commitments | Vendor documentation | Date |
Two rules keep this honest. Mark every cell with how it was assessed, because "we tested it" and "the vendor says so" are different claims and readers deserve to know which they are reading. And state pricing as a model with a link rather than an amount, because amounts go stale quietly and a stale price attached to a competitor is the kind of error you do not want to defend.
Writing the "choose the other one if" section
This is the section that determines whether the page is a comparison or a sales asset, and it is the one most teams water down until it says nothing.
A weak version and a real one. The example is illustrative and both products are invented.
Before:
Is Northgate Queue right for everyone?
While Northgate Queue may be a good fit for some smaller teams, most growing organisations find that our platform's depth and enterprise-grade capabilities better serve their long-term needs as they scale.
That is a compliment shaped like a caveat. It names no situation, concedes nothing checkable, and gives a reader no way to disqualify the publisher's product. Extracted alone it is pure marketing, and a system summarising the trade-off has no reason to use it.
After:
Choose Northgate Queue instead of us if any of these apply
- You need on-premise deployment. We are cloud-only and have no on-premise option and no timeline for one. Northgate Queue documents a self-hosted install.
- Your team is two people and this is a side project. A complex vendor may assume someone will own configuration and ongoing measurement. If nobody can, a spreadsheet or simpler tool is the better fit.
- You need coverage of the three regional sources listed below. We do not cover them today. Checked 12 August 2026.
- Your buying process requires a named support contact. The vendor offers email support only at every plan level.
If none of those apply, the comparison comes down to method transparency and export flexibility, which we cover in the two sections above.
The second version costs you deals you would have lost anyway and buys you something durable: a passage that is quotable, checkable, and obviously not written by someone pretending to be neutral. It also gives an evaluating reader a reason to trust the sections where you do claim an advantage.
Three requirements for this section to be worth anything:
- Every item must be a real disqualifier, not a strength in disguise. "Choose them if you want something less powerful" is not a disqualifier.
- Every item must be checkable. Name the missing capability, the deployment model, the specific gap.
- It must be near the top of the page, not buried under the verdict. If it only appears after the call to action, it is decoration.
Disclosure when you're one of the options
If you sell one of the compared options, say so above the comparison, in the page's own words, not only in a footer. State what you sell, what you did not test, and what your interest is. Then keep any commercial call to action below the "choose the other one if" and "choose neither" sections, so that no reader reaches your purchase path without first passing the honest limits.
Additional disclosure and advertising obligations may apply to comparative claims depending on where you operate and how the page is promoted, and those rules can also cover affiliate arrangements and paid placements. This is general information rather than legal advice. Before publishing a comparative claim, obtain current jurisdiction-specific advice where needed and record the authority, scope, and review date; re-check when the claim, promotion channel, or controlling rule changes.
Practical rules we follow and recommend:
- Never place your own option first purely because it is yours. If it wins a criterion, show the reasoning.
- Never claim to be the best or the leading option in your own comparison. Say which segment you fit and which you do not.
- Never affiliate-link a direct competitor inside a page comparing you to them.
- If you are compensated in any way relating to an option on the page, disclose it in the same screen as the comparison.
Third-party coverage carries weight your own page cannot, and the rules for earning it legitimately are a separate topic covered in reviews and third-party coverage. One thing worth stating here: soliciting favourable comparisons you influence is a different activity from publishing an honest one, and it fails for the same reason a one-sided page fails.
Why fabricated or untested claims are a durable liability
A wrong statement about a competitor does not stay on your page. Once published, it can be copied, quoted, summarised, and cached, and corrections do not propagate at the same speed the original did. You may fix the page in an afternoon and still be arguing about the claim months later.
The exposure comes in three forms. There is the commercial cost, when a prospect checks one claim, finds it false, and discounts everything else on the page. There is the legal exposure attached to false statements about a named competitor, which varies by jurisdiction and is exactly the sort of thing to raise with counsel before publication rather than after. And there is the durable source problem: a page found to contain checkable errors is a weaker source for anything that later summarises the topic.
The defensive habits are dull and they work:
- Every competitor claim is either verified hands-on, quoted from their current documentation with a link and a date, or not stated.
- Never infer a limitation from the absence of a marketing page. Absence is not evidence.
- Re-verify on a schedule and show the date. A dated claim that is later superseded is a normal correction. An undated one is an accusation.
- Log corrections in place, with the date and what changed.
- When you are not sure, write the uncertainty. "We could not verify this at the time of writing" is a publishable sentence and a credible one.
How to check whether your page is being used
Do not guess from impressions. Set up something you can repeat.
- Write a small prompt set of the questions a buyer would actually ask, phrased as buyers phrase them, including the ones where you expect to lose.
- Freeze the wording and record the models and versions you run it against, the dates, and how many times you ran each prompt.
- Record the sources cited, not only whether you were named. The citation list tells you which pages the answer was built from, which is more actionable than a mention count.
- Run it several times. Answers vary run to run, and a single run tells you close to nothing.
- Re-run after you revise the page, and compare against your baseline rather than against memory.
Interpret the result carefully. If your page starts appearing after a rewrite, you have a plausible explanation and a documented change, not proof of causation. Say exactly that in the report. The what actually changes recommendations hub sets out how much weight each lever can honestly carry.
In Plain English
What this means: comparison pages written to win are weak sources. Comparison pages written to help someone choose, with stated criteria, dates, disclosure, and a genuine section on when to pick something else, are stronger sources for both readers and systems that summarise them.
Who this helps: product marketers and content leads with "vs" pages that perform in classic search and never appear in AI answers.
When to use it: on your highest-intent comparison pages first, and on any page where you are one of the options being compared.
What to do next: add a real "choose the other one if" section with checkable disqualifiers, move your disclosure above the table, date every competitor claim, and remove any claim you cannot verify today.
Get the prompt-set template pack from our prompt-set library and build the buyer questions you will use to check whether the rewrite made any difference.
Primary-source verification
Checked July 28, 2026. The operational and factual guidance above was reviewed against the primary references below. Provider features, prices, policies, interfaces, jurisdictions, and account-specific results can change; recheck the controlling first-party document and your own account or measured data before acting.
- Google Search: AI features and your website - How Google describes eligibility, preview controls, and appearance in its AI search features.
- Google: common crawlers and special-case user agents - Current Googlebot and Google-Extended identities and controls.
- OpenAI: overview of OpenAI crawlers - Current GPTBot, OAI-SearchBot, and ChatGPT-User descriptions and publisher controls.
- Anthropic: web-crawler controls - Current ClaudeBot, Claude-SearchBot, and Claude-User descriptions and controls.
- Perplexity: crawler and user-agent documentation - Current PerplexityBot and Perplexity-User descriptions and controls.
- FTC: Endorsement Guides Q&A - Current U.S. regulator guidance on endorsements, disclosures, and reviews.
- FTC: Consumer Reviews and Testimonials Rule Q&A - Current U.S. rule guidance on deceptive review practices.