Measurement
Two comparison pages, same template. One landed at position 6, one at 57.
By Arnav Mukherjee, founder of TofuBofu · August 11, 2026
On a call on 10 August I shared my screen and pulled up Search Console, mostly to make a point about how fast a new domain can move. Two rows sat near each other and made a different point entirely. Both were alternatives pages. Same template, same author, published nine days apart, both aimed at a named competitor in our own category.
The page targeting the smaller, newer competitor was sitting at an average position of about 6. The page targeting the better-known brand was at about 57. The person on the call asked the obvious question, and the honest answer is not that one page is better written. They came out of the same process on the same week.
The difference is who else had already written that page. And that turns out to be the single most useful thing to know before you commission a comparison page, because most teams pick their targets in exactly the wrong order.
The order most people pick in
Ask a founder which competitors to write comparison pages about and the list comes back sorted by threat. The biggest name in the category first, then the one that keeps beating you in deals, then the funded newcomer everybody is talking about. It feels like prioritisation. It is actually a ranking of the queries where you have the least chance of appearing.
A well-known brand's alternatives query is contested ground that has been contested for years. The brand often defends it with its own comparison hub. Every competitor in the category has published their version. The established review sites and roundups rank on it because they rank on everything. You are not writing a page, you are entering an auction, and you are entering it last with the weakest domain in the room.
A smaller competitor's alternatives query frequently has nothing serious on it at all. Nobody has bothered, because the brand is not famous enough to be worth the effort. So a genuinely useful page can be the best result available rather than the twentieth reasonable one, and it gets found by people who are already deep in a buying process.
Same page type, opposite competitive position
What the two rows can and cannot tell you
Two pages on one site pulled on one day is an observation, not an experiment, and I would rather hand you that caveat than let you over-read it. The queries have different volumes. The pages have different link profiles. Average position is not traffic, and a good position on a query nobody searches is worth less than a bad position on one they do.
What makes it worth publishing anyway is that the obvious alternative explanation runs backwards. If you assumed the buried page was simply newer and had not settled yet, the opposite is true: the page sitting near 57 went live on 7 July and the one near 6 went live on 16 July. The older page is the buried one. Whatever is holding it down is not time.
The same shape shows up in AI answers
This is not only a ranking effect, and that is why it matters more now than it did three years ago. When we asked four AI engines to name the best managed IT providers across 15 regions and checked every answer against those regions' own directories, 139 of the 228 directory-listed firms were named by no engine at all. Engines crowd around whoever is already most written about, for the same reason page one does.
Which means the strategy transfers. On a broad, famous, high-competition question the answer is settled and you are not in it. On a specific, qualified, less contested question there is often no settled answer at all, and being the clearest available source is enough to get named. Our own measurement of 3,538 question-and-engine cells found brands named on 5.4 percent of buying questions and zero percent of awareness questions, so the specific end of the spectrum is also the commercial end.
A comparison page sits exactly there. It is a buying question with a named product in it, asked by someone already choosing. That is the most valuable question in your category and, for the smaller competitors on your list, one of the least defended.
Find out which competitors AI actually names beside you
Run a free scan across six AI engines on your own buying questions and read every firm the answers name, which is the real comparison set your buyers are handed.
Get your free auditBut nobody searches for my category yet
A founder I spoke to last week runs a product in a category buyers have no name for. He put the objection perfectly: are people even searching for this, because they do not know it exists. It is the right question, and for most content it is a fatal one. You cannot rank for a phrase nobody types.
Comparison pages are the exception, and the reason is worth stating plainly. Their demand is borrowed. It does not come from your category, it comes from the competitor's brand, and the competitor's brand has demand because somebody else has already spent money creating it. A person typing a funded competitor's name plus the word alternatives has the problem, is in market, and is at the bottom of the funnel. None of that required your category to have a label yet.
The trap is that the biggest brands have the most borrowed demand available and the least of it obtainable. So in a young category, the answer to "we have no volume" is not the most famous competitor's alternatives page. It is the four or five near-peers your buyers actually raise on calls, which have real named demand and almost no competition for it.
How to pick your list
Five steps, in this order. The first two are where the leverage is and they are the two people skip.
Get the real comparison set, not the remembered one
Ask an AI engine your buyers' actual buying question and write down every firm it names. That list is what your buyers are handed, and it regularly differs from the list a founder writes from memory. When the two sets diverge, that gap is itself the finding: your content is teaching engines to frame you against companies you do not think you compete with, and buyers are getting the same signal.
Sort that list by how much has already been written, not by threat
Search each competitor's name plus alternatives and count the serious pages already there. Where the first page is full of established roundups and the brand's own comparison hub, deprioritise it however much the name stings. Where there is nothing but the brand's homepage, move it to the top.
Write the page a buyer would actually finish
An alternatives page must genuinely be a list, including options that beat you on some axis, or nobody trusts the entry that happens to be yours. A versus page must reach a verdict and say plainly who should pick the other one. A sales page with a competitor's name in the title fools nobody and gives an engine nothing worth quoting.
Verify every competitor claim at their own page and date it
Pricing and feature claims go stale fast, and quoting a competitor's blog rather than their pricing page has burned us before. Check the vendor's own page, write the date you checked into the post, and re-check when you touch it. A comparison page with a wrong price on it is worse than no page.
Measure the page on both scoreboards
Track its Google position and, separately, whether the engines start naming you on that comparison question. Those move at different speeds and for different reasons, and a page can be earning citations while still sitting on page four. If you only watch one, you will kill pages that are working.
We publish 17 of these pages across both formats, so this is a lesson from our own bill rather than a theory. The ones aimed at the most obvious targets have performed worst, and the quiet ones aimed at competitors nobody has written about yet are the ones doing work. If I were starting the programme again I would write them in reverse order.
The wider point is the one this site keeps coming back to. Search rank is the floor and AI visibility is a distinct layer on top of it, and a comparison page is one of the few assets that works on both at once. G2's 2026 research found 51 percent of B2B buyers now start vendor research on an AI chatbot, up from 29 percent, and 69 percent have switched vendor based on what AI told them. The page that answers "is there something better than X" is the page that gets read at exactly that moment, whichever surface the buyer is standing on.
Frequently asked questions
Which competitors should I write comparison pages about?
Start with the competitors small enough that nobody has already written the page, and specific enough that your buyers genuinely weigh you against them. On a live Search Console pull from our own site on 10 August 2026, our alternatives page targeting a smaller, newer competitor sat at an average position of about 6, while the page targeting the better-known brand in the same category sat at about 57. Same template, same author, same month. The page aimed at the crowded brand query was published nine days earlier, so age is not the explanation. The difference is how many people had already written that page before we did.
Why do alternatives pages for big brands rank so badly?
Because a well-known brand's alternatives query is one of the most contested pieces of real estate in any category. The brand itself often defends it, every competitor has written their version, and the established review roundups already rank. You are entering a page-one auction that has been running for years. A smaller competitor's alternatives query frequently has no serious page at all, so a genuinely useful one can be the best result available rather than the twentieth reasonable one.
Do comparison pages help with AI visibility or only with Google?
Both, and the AI side follows the same logic. Our research across 15 regions found engines crowd heavily around incumbents: of 228 directory-listed firms in those regions, 139 were named by no engine at all. The more established the brand, the more settled the answer to a question about it, and the harder it is to insert yourself into that answer. The queries where a smaller firm can still get named are the specific, qualified ones, which is exactly the same shape as the ranking finding.
My category has no search volume yet. Is a comparison page pointless?
It is often the opposite. A founder asked me this last week about a product in a category buyers do not have a name for yet, and it is a fair question: if nobody searches for the thing you do, what is there to rank for? Comparison and alternatives pages solve exactly that problem, because their demand is borrowed from the competitor's brand rather than from your category. Someone searching a funded competitor's name is a buyer with the problem, in market, at the bottom of the funnel, whether or not your category has a label yet.
Should I write versus pages or alternatives pages?
They answer different questions and both belong in the set. An alternatives page answers a buyer who has decided the named product is not right and is looking for a list, so it must actually be a list, including options that beat you on some axis. A versus page answers a buyer down to two choices who wants the difference stated plainly, so it must reach a verdict and say who should pick the other one. The failure mode for both is the same: a page that is really a sales page with a competitor's name in the title, which buyers detect immediately and engines have no reason to quote.
How do I know which competitors my buyers actually consider?
Ask the engines and read the names in the answers. When a buyer asks an AI engine for vendors in your category, the firms it lists beside you, or instead of you, are the comparison set your buyers are being handed. That list is frequently not the list a founder would have written from memory. We see the mismatch often enough that we treat it as a headline finding rather than a data-quality problem: the competitors your content makes you look like are the competitors your buyers will weigh you against.
How reliable is a two-page comparison like this?
It is an observation, not an experiment, and it should be read that way. Two pages on one site, pulled on one day, with different query volumes, different link profiles and different levels of established competition. Average position is also not traffic, and a strong position on a low-volume query can be worth less than a weak position on a high-volume one. What makes it worth publishing is that the confound most people would reach for runs the wrong way: the page sitting at 57 is the older of the two.
Sources and further reading
- Google Search Console for tofubofu.com, pulled live on 10 August 2026: the average-position figures for our own two alternatives pages, published 7 July and 16 July 2026.
- TofuBofu MSP research 2026: of 228 directory-listed firms across 15 regions, 139 were named by no AI engine at all.
- Our measurement of 3,538 question-and-engine cells: brands named on 5.4 percent of buying questions and zero percent of awareness questions.
- G2 2026 B2B buyer research: 51 percent begin vendor research on an AI chatbot, up from 29 percent, and 69 percent switched vendor based on AI.
- Forrester B2B Buying Study 2026: 94 percent of buyers use AI somewhere in the buying process.