No history, no links, no age: why a brand new domain is the cleaner test

A measurement is worth what it would cost to disprove, and a brand new domain is the cheapest thing in search to disprove a claim on. The aiknowsus.com export on our proof page is that case. The window is 2026-10-04 to 2026-10-12, 9 days with data, exported on 2026-10-14. It records 5,965 impressions and 3 clicks. The domain was registered days before the window opened, so it had no history, no links and no age. That is why the only claim the file supports is a narrow one about how quickly the pages became findable, and why that narrow claim is harder to argue with than a larger one would be.

What a standing start removes from the experiment

On an established site, a change in the numbers has several candidate causes and no way to separate them from an export alone. Existing pages were already accumulating authority. Existing links were already pointing somewhere. The domain already had a crawl history, so Google already had a reason to come back.

A domain registered days earlier has none of those. There is no backlog of indexed pages to carry the new ones, no inbound link equity, no crawl pattern. When impressions appear anyway, the list of things that could have caused it is short, and that short list is the whole value of the exercise.

The 5,965 is a discovery result and it is not traffic

This is the sentence that has to travel with the figure, because the figure invites the wrong reading. An impression means a page was in the results for something somebody typed. It means Google is showing the pages. It does not mean anybody arrived.

The clicks in the same nine days total 3. Position is what converts being shown into being visited, and the best day in this window, 11 October 2026, recorded 2,685 impressions at an average position of 41.6. Anybody holding up a nine day old domain and calling 5,965 traffic is describing something the file does not contain.

The claim the window supports, written as narrowly as it can be

Here is the whole of it. A domain registered days before 4 October 2026, with nothing behind it, was being shown in Google results 5,965 times by 12 October 2026. That is a statement about discovery speed on a new domain over nine days, nothing more, and it is checkable by anybody who runs the same export on their own site.

Three things are deliberately not in that sentence: a traffic figure, a trend, and a projection. The clicks are reported as 3 because that is what they were. Nine days cannot establish a trend and was not asked to. And no rate of change from this window is extended into next month, because a window describes itself.

How to run the same test on a domain you control

The design is simple enough to copy and it is the only version of this test that answers a question about your own method rather than somebody else's site.

• Register the domain and verify it in Google Search Console before you publish anything, so day one of the window is genuinely day one.
• Publish a known number of pages on a known date, and write that date down separately from the export.
• Add no links from anywhere you control. The point of the test is what happens without them.
• Take the performance export on a stated date and report impressions, clicks and days with data, in that order.
• Check what Google actually kept rather than inferring it from the impressions. Reading the page indexing report tells you how many of your pages are in the index and names the reason for the ones that are not.

The part most people get wrong is the page selection, and it decides the outcome more than the publishing date does. A new domain shown for a competitive head term records nothing in nine days. The same domain answering a specific question that few pages answer properly can appear within days, which is why telling a long tail query worth a page from one that is not is the step to take before the first page is written rather than after the first export.

Why this is the shape of result we publish

We sell AI visibility measurement and the work that moves it, so a number we publish about ourselves is a number with our commercial interest attached to it. 64 of 102 build spec conversations run on ChatGPT and Gemini on 15 October 2026 asked for exactly that interest to be stated on the page, and 101 of 102 asked for a before and after with the window stated.

A standing start satisfies both of those more cleanly than anything else we could show. The before is zero, because the domain did not exist. The window is nine days with both dates printed. And the claim stops where the file stops.

Questions about testing a brand new domain over a short window

Why does a new domain get shown in results at all with no links?

Being shown needs the page indexed and judged relevant enough to appear for something somebody typed. That is a lower bar than being chosen, and it is reachable without inbound links. Being chosen is where position and standing start to matter.

What does a window this short establish?

That discovery happened, and the date on which it happened. Establishing a direction of travel is a separate job that needs months, and the published claim stops at the first one rather than reaching for the second.

What would have falsified the claim?

A flat impressions line. If a domain with no history publishes pages and the performance report stays near zero for nine days, the method did not make those pages findable in that period and the export says so in public. That is the property that makes the test worth running.

Why report 3 clicks rather than leave the column out?

Because leaving it out is the thing a reader should distrust. 68 of those 102 build spec conversations asked for failures and negative results to be shown alongside wins, and a click column reading 3 next to an impression column reading 5,965 is the most informative pair on the page.

Does an average position of 41.6 mean the pages are badly written?

It means they are new. Position responds to the things a nine day old domain has not had time to accumulate, so the useful reading of 41.6 is that the pages are in the results and have further to travel, measured on 11 October 2026 and due to be read again rather than assumed.