Frequently asked questions.

Coverage is defined by sector and comparison group, from public sources. No company enters the index by invitation, and none changes its position by paying.

It measures whether a company’s information is interpretable by automated systems: structured data, identity consistency across sources, and permitted access for the crawlers of the AI engines that now recommend businesses.

No. Every measurement comes from an observable, dated source, never from an opinion, and never from values inferred out of other values. Where a reading is sampled, as with the citation rate in generative engines, the sample size is what the published confidence reflects.

Nothing. Consulting the index and the ranking is free and requires no sign-up. What is paid for are tools built on the data, never access to the index.

No. No position in the index is for sale: there are no sponsored slots, no paid placements and no score corrections by contract. A position changes when the measured signals change.

In periodic cycles, not in real time. Every measurement is dated, and the published date is the date of the measurement, not the date you looked.

From public, publicly accessible sources, and from nothing else: the company website read at the addresses any visitor reads, public DNS records and certificates, Google’s performance API, the news feed, profiles and pages the company itself published, and business data providers. Nothing is bought from personal-data brokers, nothing is obtained through restricted access, and no company is measured with information that is not available to anyone who looks for it. No company is invited in, and none pays to enter, leave or rise.

Both can be asked for and both get an answer. A wrong value is corrected through a dispute from the verified profile, or through our contacts. Removal from the index is requested through those same contacts: we review it and always reply, including when the reply is a reasoned refusal. Where removal is granted, it applies to future editions — published editions are not rewritten.

Coverage starts from public sources, with no sign-up. An index that measured only those who applied would be neither comparable nor independent.

Only publicly observable signals: probes to the website and its public files, search engines and local results, generative engines and language-model crawlers, public reviews, press, directories and third-party mentions, and the cadence of owned and social channels. The sector and size that define the comparison group come from company-level firmographic data. No personal data and no privileged access.

Whoever represents the company can claim the record, confirm ownership and see the detail behind each measurement.

Every measurement has a recorded date and source. There is no contestation form: the route is the contact form, stating what was observed. Once an error is confirmed, the record is corrected in the next computation cycle. The value already published is not rewritten.

Removal requests are reviewed case by case under the applicable legal framework. The definitive wording will be published in the methodology.

Yes, free of charge and with no sign-up. The request follows the normal process: identifier validation, measurement by the same criteria, and entry in the next edition. Requesting indexing neither guarantees nor influences any score.

Because the measure is relative. It is enough that the other companies in the comparison group improved, that companies entered or left that group, or that the sector of one of them was corrected to a better fit. What the index measures is a place within the market, and the market moves too.

Each observed metric becomes a number in one of two ways: by the company’s position within the group of comparable companies, when a value only means something compared (22 reviews is neither many nor few until you know the sector), or against a published ruler, when the question has a right answer, such as having HTTPS or the Core Web Vitals thresholds. Those numbers make up each of the six pillar scores (Foundation, Machine Readability, Discovery, Reputation, External Authority and Activity) by weighted average. The six pillars are then aggregated by geometric mean, and it is the geometry that stops a strong pillar from covering a weak one. The weights and the computation version are published in the methodology.

The other companies in the same sector, nationwide. Not the sector within a municipality: a local group is almost always too small for a stable comparison, and the position ends up saying more about how many neighbours are indexed than about the company. When the national sector itself is too small, the position is marked provisional rather than widening the sector: comparing a clinic with a gym because both are «health» is not comparing.

There is no judgement of ours at any step. Most metrics are converted into the company’s position within the group of comparable companies; the rest are measured against a published ruler, the same for everyone. In neither case is there a mark decided case by case.

What a company does not have weighs on the score. What we could not measure is never imputed: it is recorded as unmeasured and lowers the published confidence.

A score comes out as final when it has enough measured pillars and the measurement confidence reaches the published minimum. Below that it comes out as provisional, with its confidence in plain sight. And below a second, lower minimum there is no score at all: the company stays in the index and keeps being measured, but takes no position until there is something to compare it against.

Every score states the methodology version used to compute it, and changes to the method are on the record.

Yes. Every measurement has a recorded date and source, and an error reported through the contact form is checked against that source. Once confirmed, the correction lands in the next computation cycle: a published edition is not rewritten. What changes is the next score.