A top source at 19 readings of 291 means no incumbent holds this category

In our 15 October 2026 run, the most cited host on one of the two engines was webfx.com, in 19 of that engine's 291 readings. That is the leader. Nineteen. Nothing in that table holds the category, which makes it the cheapest surface in our whole dataset for a publisher with no history to start on. It is also a surface where our own domain was cited 0 times across those 291 readings, and 0 times across 630 readings on the same engine the following day. Both belong on one page, because without the second the first is just a sales pitch.

Open and unwon is the honest description. The opening is measured. So is our failure to take it.

How far that is from the other engine in the same run

Engine	Readings it gave us	Its most cited host	Readings citing that host
The flatter one	291 usable	webfx.com	19
The concentrated one	306	developers.google.com	176

Same questions, same three markets, same day. One table has a host in well over half its readings and the other has a leader in fewer than one in fourteen. Those are not two versions of one market. They are two markets, and a page written for one of them is not automatically a page written for the other.

Why a flat table is the cheaper place to start

• Nothing has to be displaced. A page can enter the set without anybody leaving it, which is a far lower bar than beating an entrenched source on its own question.
• No single competitor sets the standard. At six to nineteen citations per host, there is no reference page everybody is being compared against.
• The set is already moving. A table that flat changes between readings on its own, which means a change you cause is not fighting inertia.
• Small sites are in it. Several of the hosts below the leader on that engine are purpose built utilities rather than large companies, which tells you the entry requirement is publishing rather than scale.

And the part that argues the other way

We have had two attempts at that surface and taken nothing from it. 0 citations in 291 readings on 15 October 2026, and 0 in 630 readings on 16 October 2026, which is 0 in 921 across both. Nothing else in our records has come back identical on two separately designed banks.

So a flat table is not an easy table. It is an unclaimed one, and those are different claims. What our own zero rules out is the lazy version of this argument, where openness is presented as the same thing as a quick result. Three things are worth taking from the zero rather than from the nineteen.

• Measure presence, not position. On a table this flat the question is whether you are in the set at all and how often, which needs a frequency and a denominator rather than a rank.
• Expect fewer chances per page here. One page buys fewer chances here. Read that as guidance on which surface to measure first rather than as a reason to publish less.
• Keep a zero as a zero. Reported with its denominator and its date, it is a baseline nobody can later restate as progress.

Where to start, given both halves

Pick the questions in your bank whose source tables are flattest on this surface, because those are the readings a page published this month can be part of. Publish one page per question, with the answer in the first two sentences, a date and a last verified line, a figure with its denominator, and a price in the reader's own currency. Then take a second reading and compare it to your first rather than to anybody's benchmark.

That is the work we sell, and the zero above is the number our own version of it has to move. The Google Search Console side of our own record, two domains with their windows and the export date printed on the page, is our proof page.

The edges of this reading

One engine's half of one run, 102 questions per market on 15 October 2026 across the United States, the United Kingdom and India, with one reading per question. The other three surfaces our product covers were asked nothing. The counts are readings citing a host, deduped to one domain per reading, and they record what an answer was assembled from rather than which business the answer named. No per industry figure exists in the bank, and city level numbers come from a separate scan of ten named places.

Sequencing the work when one surface is open and one is not

Having both numbers changes the order of a plan rather than its contents. Four decisions follow from a reading like this one, and none of them requires a bigger budget.

• Write for the open surface first. The same page costs the same to produce either way, and on the flat surface it is not competing against an entrenched source.
• Take the baseline before the first page goes up. Ours is nought in 921 readings on that engine, dated, with both denominators. A baseline recorded later cannot do this job.
• Report presence and frequency, never a rank. On a table whose leader sits at 19 of 291 there is no ladder to climb, so the column that matters is how often you appear at all.
• Leave the concentrated surface to question selection. There, the decision is which questions to ask rather than how well to write, because the leading host is the subject of the question it keeps answering.

Done in that order, a quarter produces a comparison against your own first reading rather than against a published benchmark that was taken on somebody else's questions.

Questions about the flatter of the two engines

What was the most cited host on that engine?

The host was webfx.com and it reached 19 readings out of 291, on 15 October 2026. On the other engine in the same run the leading host reached 176 of its 306.

Is it genuinely easier to be cited there?

Nothing is entrenched, which removes the displacement problem. Our own result on the same engine is zero citations in 921 readings across two runs, so the opening and the difficulty are both measured and both real.

Should I start with the flat surface or the concentrated one?

Start where your own questions are flattest, which is a measurement you take rather than a rule you inherit. The two engines in our run came out at 19 of 291 and 176 of 306 on the same questions, so the answer is category specific.

What should I report on a surface with no leader?

Presence and frequency, with the denominator beside them, per engine, with a date. A rank has nothing to rank against when the leader is at 19 of 291.

Which engines and which date?

ChatGPT and Gemini, read on 15 October 2026. The full question design, the controls and the complete source mix are published as 597 readings across ChatGPT and Gemini in three markets.

An unclaimed table and a zero, on the same page

Anybody selling you the first half of this page without the second half is selling you the easy version. The table is open, our own count on it is nought in 921, and the two together describe a real opportunity with a real cost attached. That is the version worth planning against.