Name what you sell on the page, beside the claim it would benefit

Put the commercial interest in the sentence that needs it, at the same prominence as the claim it sits beside. Sixty four of 102 asked for it. Those conversations ran on 15 October 2026, with ChatGPT and with Gemini. A reader was going to assume the interest in any case, so the only thing a disclosure settles is whether you said it before they worked it out for themselves. Stating it first is read as care rather than as a weakness, and it costs one sentence.

Our own interest here is plain: businesses pay us to take readings like these and to write what comes after them. Every figure below comes from readings of our own that are published with their denominators attached.

What 64 of 102 were asking for

The count matters less than the shape of the request, which came back consistently enough to list. All ten requirements that set produced, and the way the conversations were run, are written up in what the two engines said they need before citing a vendor page. The disclosure row asked for four things.

• What you sell, in the words you sell it in. Not a relationship, not a category. The thing somebody pays you for.
• What you would gain if the page were believed. The reader can then weigh the claim against the incentive without having to guess at the incentive.
• Beside the claim, not under the page. A line at the foot of a document is a legal habit. A line next to the strongest sentence is the thing being asked for.
• Attached to the method as well as the result. Where a design choice could be read as flattering, say who chose it and why.

Three forms fail. A generic statement that commercial relationships may exist discloses nothing. A disclosure that only appears once the reader scrolls past everything persuasive arrives too late to do its job. And an apologetic one invites the reader to discount the page, which is a different outcome from inviting them to check it.

The disclosure that costs something is the one that works

A statement of interest is cheap to write and worth very little on its own. What gives it weight is a figure printed next to it that the interest would rather have left out. For us that figure is how often our own pages turned up in the source lists of our own runs.

Run	Readings	Readings citing our domain
15 October 2026	597 usable of 612 attempted	3, with 0 in the United States on either surface
16 October 2026	1,260, all usable	6, which is 0.5%

One surface accounts for every zero in that table: taken together, the two runs put 921 Gemini readings on record carrying nothing of ours. 68 of the same 102 conversations asked for failures and negative results rather than only wins, and how to write a zero so that somebody else can use it is set out in reporting nought in 921 readings. A disclosure above that table is doing work. The same disclosure above a page of favourable figures is decoration.

Where to put it, with two placements out of three

Three positions carry a disclosure properly and most pages need two of them.

• In the opening paragraph, when the page is about the category you sell into. The reader meets the interest at the same moment they meet the subject.
• Immediately after the strongest claim. This is the placement that changes how the claim reads, and it is the one most often skipped because it is the one that costs something.
• Inside the method. Where you chose a question set, a denominator or a counting rule, say that you chose it and that you sell the result.

A fourth placement, the closing paragraph, is where most pages put their only disclosure and it is the weakest of the four. By then the reader has either accepted the page or left.

Why the self published page is the field, not the exception

Disclosure is sometimes argued against on the grounds that a seller's page is discounted anyway. The measured picture is narrower than that. In our 16 October run, 82.0% of 20,143 citations landed on a page published by a company or an agency selling something, while review directories took 0.7%. Outside the layer of platform documentation, the slots are held by people publishing their own work.

So the competition for a vendor page is other vendor pages, and 41 of the 102 conversations said they would use a seller's page given measurement a reader could check. The disclosure is part of what makes the measurement checkable, because it tells the reader which direction to be sceptical in. Where trust sits in the documented frameworks around this, rather than in the repeated version of them, is worked through in the rater framework that is not a score and what actually changes on a page.

What a disclosure does not buy

It does not make a weak figure usable. A percentage with no denominator stays unusable with an interest statement above it, and 80 of the same 102 conversations asked for the sample size separately.

It does not license an unsourced claim about somebody else. Naming your interest and then restating a competitor's price without a date or a source is two failures rather than one offset by the other.

And nothing here measures what adding a disclosure does to a citation rate, because we have not run that test. What 64 of 102 establishes is demand on one date from two products. Three of the five surfaces we read for a customer were never in the room for these conversations.

Questions about disclosing a commercial interest on a measurement page

How long should the disclosure be?

One sentence in each place it appears. It names what you sell and what you would gain, and then the page carries on. A paragraph about your values is a different document and it is not what was being asked for.

Does a disclosure have to be in the same words every time?

No, and it should not be. The same formula repeated across a set of pages reads as a template to a person and to a model, and the interest is easier to believe when it is stated in terms of the specific claim the page is making.

What if the commercial interest is indirect?

State the mechanism in plain terms: an affiliate arrangement, a partnership, a product you sell that competes with something named on the page. The test is whether a reader who learned it later would feel it had been withheld.

Should the disclosure go in the structured data as well?

Put it in the visible text first, because that is what a reader and a fetcher both receive on a plain request. Structured data describing the publisher is useful alongside it and is not a substitute for a sentence anybody can read.

Is naming our own product on a research page not just advertising?

It is when the page has nothing checkable on it. The distinction is whether the page would survive a reader going to verify it: a named interest beside a figure with a denominator, a date and a published question set is a different object from a named interest beside an unsourced percentage.

The sentence that makes the rest of the page harder to dismiss

A page with a commercial interest and no statement of it leaves the reader doing two jobs at once, working out what the figures mean and working out what the publisher wants. Doing the second job for them costs a sentence and leaves the figures to be argued with on their own terms, which is the only argument worth having. Here is ours, in the form this page has just asked for: money changes hands when a business pays us to take these readings across five surfaces and to write what follows, and the figure above showing our own domain cited six times is the one that argument has to survive. It sits with both Search Console exports and the dated blind scan at aiknowsus.com/proof.