What our own baseline said on day one

Why publish this at all?

We are new, so we have no client results to show, and inventing some is out of the question. What we can do is run the method on ourselves and print the output, including the parts that are unflattering.

There is also a narrow window for it. A domain registered this morning is genuinely absent from everything, and that zero can be recorded once. Wait a month and it is gone, and any later claim about where we started becomes a story rather than a record.

What was measured, and what was not?

Instrumentation matters more than findings, so it comes first. The run reached two surfaces: search results, and a direct retrieval of the site by a machine reading it cold. Five surfaces were not reached, because the environment the run happened in has no access to them.

The five are recorded as not measured. Recording them as finding nothing would have been a different claim entirely, and the difference between those two is the whole of a report's honesty. A full run across every surface we publish is done by a person, and that is the run scheduled next.

Where did we appear in the category questions?

Nowhere. Asked who does this kind of work, and asked the problem in a buyer's own words, the answers named nine other firms between them and did not name us once. At a few hours old that is the only possible outcome, and it is recorded as the floor rather than as a finding.

One thing in it was genuinely useful. The category question and the buyer's question returned almost entirely different sets of names. Whoever wins the industry term is not the same party who wins the sentence a buyer actually types, which is a reason to ask both rather than assume one predicts the other.

Does our own name resolve to us?

No, and the finding worth the whole run was not the one we expected.

Searched as an exact phrase, our name returned no match for us. Back came the nearest established entities instead: a marketing agency, an advisory business, a property company, and a wealth management group. Searched as a bare word, the name was treated as a misspelling and corrected toward older ones, among them a long-standing piece of mapping software.

So the name sits inside three separate collisions at once: one with a firm in the discipline we define ourselves against, one with an unrelated sector, and one with an established product. Our own about page explains that the name means falcon. The index already agrees about how the name sounds, and has given that space to other people.

Entity disambiguation is the work we sell, and we now have it as our own first problem. The structured data on this site currently offers nothing to anchor us against those older entities.

Can a machine place us?

A machine fetching the homepage cold described the firm correctly: what it does, both audiences it serves, and the four services. Comprehension is working.

The same fetch reported that our location was not stated, which is accurate. Our jurisdiction is written in prose on one interior page and appears nowhere a machine reading the entry page would find it. A reader gets it; a retrieval does not.

What did we get wrong in our own run?

The first version of this baseline concluded that we promise more coverage than the method can deliver, and called it the most significant thing in the run. The conclusion was wrong, and the correction is published rather than quietly removed.

Five surfaces were out of range for the automated environment doing the measuring, and a person can reach every one of them. Every report we produce is produced by a person, which the site said all along. We had measured a subject with an instrument that could not reach part of it, and then concluded the subject was deficient.

Conflating those two is the most likely way for a visibility report to be confidently wrong about a client, so it is now a written rule in our procedure rather than a lesson we happen to remember.

What do we expect to happen?

Stated now so the next run can score these rather than explain them away. We expect the site to be indexed within two to six weeks, and we expect no category ranking to come with it, because a domain with no inbound links does not enter a competitive category on structure alone.

We expect the name collisions to stay exactly where they are. Older, better connected entities hold them in place, and nothing we can change on our own pages reaches them. Worth being precise about why: the reflex fix for an entity problem is structured data, and from our current position structured data cannot get to it. Corroboration somewhere that is not this domain can.

And we expect assistant answers to keep omitting us for longer than search does, because assistants lean on entities that several independent sources agree about, and there is currently one.

Should the first happen while the other two do not, our argument holds for the firm making it. Should the collisions clear on their own, our argument is weaker than we have been stating, and this site will say so.

When is the next run?

Four weeks from the baseline, across every surface we publish, using the same questions in the same order. The questions are frozen, so the two runs can be compared rather than reconciled.