Changelog

What shipped

Primacy Score is built in the open and changes most weeks. This is the dated record of what moved, written in terms of what it means for a reading rather than what it took to build. The public record starts in September 2026, when the method settled at 1.0 and the product left beta.

Changes to how the Score itself is computed are not in this feed. They are versioned, dated and explained in the methodology version history, because a number you can compare over time needs a record of every change to the instrument behind it, kept separately from feature news.

Last updated September 8, 2026

September 2026

  • Research

    Houston Personal Injury Index, second edition

    The index now runs on all eight engines under methodology 1.1, with 66 firms measured and the top 20 published. Every edition stays readable at its own address, and the current edition shows how each firm moved against the previous one. No firm can buy its way in, up, or out.

  • Research

    Indexes keep their back editions

    A re-run publishes a new dated edition rather than overwriting the last one, so the record of what a market looked like in August is still there in September. Where two editions are comparable, the newer one prints the movement between them.

  • Reports

    A technical worklist, measured rather than asserted

    Every reading now carries a diagnostic layer beneath the Score: whether AI crawlers can reach your site at all, site-wide issues like duplicate titles and competing pages, the quality and internal consistency of your structured data, transport problems such as a firewall that blocks assistants while admitting browsers, your presence in the listings engines actually cite, and which of your sub-brands are discoverable. Each item states what was checked and what was found.

  • Reports

    The worklist remembers what it told you

    Findings are recorded with the reading that produced them, so they carry an age, a prediction of what fixing them should do, and a verification against the next reading. A fix that did not move anything says so. This turns the action plan from a list into a record you can hold us to.

  • Reports

    Read the answer behind any row

    Every scored row opens the full answer the engine actually gave, with each mention of your brand marked and a control to jump between them. Nothing in a report is now a claim you have to take on trust; the evidence is one click beneath it.

  • Reports

    Criticism and mixed framing are no longer the same thing

    Being described with a caveat and being described badly are different problems with different fixes, and the report stopped filing them together. Mixed framing and genuine criticism are now separated and counted apart.

  • Measurement

    Every engine answer is read in full

    A ceiling in the collection path had been truncating longer answers before they were scored, which meant a mention late in a long response could be missed. Answers are now read whole. Collection also got substantially faster and cheaper in the same pass, which is what makes the eight-engine panel affordable to run on every reading.

  • Coverage

    Question sets cap at 300

    Coverage is sold in packs of 30 on top of the 60 included, up to 300 questions. Past that we quote the engagement rather than sell it self-serve, because a self-serve checkout should not promise a turnaround nobody has measured. The ceiling is now enforced in the checkout and in fulfillment, not just described.

  • Platform

    Watch a reading while it runs

    The dashboard now refreshes while a reading is collecting and names what is currently running, instead of showing a static page until everything finishes. A network blip no longer turns into an error screen.

  • Coverage

    Your other names, recognized at setup

    Setup now proposes the sub-brands and alternate names already visible on your site, so a mention of a product line or a legacy name counts as a mention of you. Short or ambiguous names still get a human look before the set is locked.

How the Score is computed

Methodology 1.1 is current

The Score is computed only on the questions that do not name you. Reputation is reported beside it and never blended in. Every certified Score carries the version it was measured under, and every change to the method is dated and kept on the record.

Read the version history

Published by Primacy Score, a product of Catalyst Studio Inc, Houston. Tell us what you want to see next.