Research Note
The Public-Sourced Read
655 dealers that entered the cohort through search discovery, read once against Google's business listings. 69 have no Google Business Profile whose listed website is their own domain.
In the single partial tranche read, 1 of 8, that reached 740 of that tranche’s 964 members with 739 producing a record, the 655 of the 739 that entered the cohort through search discovery include 69, 10.5% [8.4, 13.1], with no Google Business Profile that resolves from their own website domain; the 84 recorded dealers that entered through a supplier-sourced dealer-network list are excluded by source, and the restriction is by recorded provenance and was decided after the read.
The question asked of each dealer
Does a Google Business Profile exist for this dealer whose listed website is the dealer’s own domain?
That is narrower than asking whether the dealer has a profile. Most do. It is the question the record can answer on its own, because the profile’s website field and the dealer’s domain are both strings the instrument holds, and comparing them needs no judgment about the business.
How resolution works
For each dealer we ran a Google Places search on the business name and location, and a second search on the website domain when the first returned no profile listing that domain; a profile counted only if the website Google lists for it was the dealer’s own domain, and otherwise the closest business-name match was kept and labelled separately.
A profile resolves from the dealer’s own domain when the website on the profile and the dealer’s domain are the same host, or one is a subdomain of the other. A name-only match is kept and labelled, and it is counted here among the dealers with no profile resolving from their own domain, because no profile named their domain. A dealer for whom the search returned a candidate that failed the name test, or returned nothing, is counted the same way.
The figure is a property of the method as much as of the dealers. A profile that lists no website, or lists one on another domain, fails this test while existing, and a name match may have attached the wrong profile entirely. Both are recorded with the method and neither is resolved by it.
Which dealers, and why 655
The cohort is 7,700+ verified dealer sites measured in the flooring, tile, cabinet and countertop trades. It is read in tranches, one of eight, and this read reached 740 of tranche 1’s 964 members before its calls were spent. 739 produced a record.
Each dealer in the cohort carries a source value saying which route brought it in. Two values exist among the 739. Most are recorded as search discovery, the value the discovery pipeline writes for a Google search of a trade term and a city, screened and verified before the site joined the cohort. 84 dealers in this read entered the cohort from a supplier-sourced dealer-network list. They are excluded so the figure rests only on dealers found through public search. Their rate was higher: 17 of 84, 20.2%, with a 95% interval of 13.0 to 30.0. The headline uses the lower figure.
That is a selection rule and it was decided after the read. The same question was first asked of all 739. The provenance split was run afterwards, in a verification pass, and this note states the public-sourced side of it. The rule is stated here so a reader can see what was chosen and when, rather than meeting a subset presented as if it had been the plan.
Two limits on the base sit in the record itself. Search discovery is also the default value the pipeline writes when no other source is given, so the record establishes firmly that these 655 did not come through the supplier list, and establishes how each one was found less firmly than that. And the record cannot go finer: manufacturer locators, directories and other public routes are not distinguished in it, so search discovery is the whole public-sourced bucket and this note does not pretend to split it.
The interval
10.5% carries a 95% Wilson score interval of 8.4 to 13.1, computed in closed form so it reproduces exactly with no seed. That is the width of a reading on 655 dealers from one partial tranche. This is a single-tranche estimate and not a census. A cohort figure needs all eight tranches, states the window they were collected across, and states that the snapshots inside it differ in date. This figure is rechecked when that read exists.
What this does not establish
It does not establish that any dealer is absent from Google. The 69 are dealers for whom no profile resolved from their own domain by this method. Some matched a profile by name alone, some returned a candidate the method rejected, and a dealer whose profile carries no website, or a website on a different domain, fails this test while having a real profile.
It says nothing about rankings. There is no ranking, impression or visibility data for any dealer here, and nothing in this note describes what a person searching would see.
It says nothing about ad delivery. No advertising product, campaign or eligibility rule was read, and nothing here connects a profile’s website field to whether an ad ran or where it went.
It says nothing about outcomes. No visit, call or sale is measured. A profile whose listed website is not the dealer’s own is a fact about Google’s record on one day, read on 2026-09-09.