Corrections
What was wrong, and when it was put right
Published work carries defects. This page records them, newest first, dated to the day the defect was corrected. An entry stays where it is once written. Nothing here is edited later and nothing is removed.
The log
-
The commercial disclosure described Sandria as a service the founder also operates, kept independent of the Atlas. That understated the relationship. Healthspan Economy is published by Sandria OÜ, so the disclosure named an adjacency where the fact is ownership, and a reader checking who stands behind this record was given the weaker of the two answers. The methodology page and llms.txt now state the publisher, its registry code and its jurisdiction, and the site footer carries the publisher identity on every page. The AI Visibility Report carries its own copy of the superseded wording in both its standing disclosure and its limitations note. That copy is unchanged here and is corrected when the report and its PDF are next rendered together, so that the page and the downloadable file cannot disagree about who publishes this record.
-
The report's Finding 3 was headed "Brand integrity holds, with two exceptions" while the finding's own text, corrected on 14 August, read "Three confirmed exceptions stand out" and went on to describe all three. The heading and the structured-data question that repeated it were left behind by that correction, so the first line a reader saw contradicted the paragraph beneath it, and a machine reading the structured data was given the superseded count as the finding's title. The heading now matches the text. No figure and no part of the finding changed here. This completes the 14 August correction rather than recording a separate defect. The finding's anchor is unchanged, so existing citations still resolve.
-
Added a relationships-with-panel-clinics disclosure to the report method page. The report already disclosed the author's commercial advisory work and that no clinic paid to appear. It did not state that people connected with panel clinics had been in contact with the author after publication. All such contacts began in August 2026, after the panel was fixed on 30 April 2026 and after publication on 19 May 2026.
-
Nearly every page on this site advertised a canonical URL ending in ".html", in both the canonical tag and the Open Graph URL, while the sitemap submitted the same page without that suffix. The two contradicted each other, so a search engine or an AI crawler was given two different answers about which address is the real one for a page. The article pages were unaffected. The canonical is now the bare path everywhere, matching the sitemap.
-
The remedy recorded below on 14 August sent the report's method links to the Atlas methodology page. That page governs what qualifies as an Atlas record and how a record is verified. It does not describe the report's panel, queries, platforms or scoring, so a reader following a link captioned "Read the methodology" reached the wrong method, and the report's own structured data still offered a machine no route to the right one. The report's method is now published in full at /methodology/ai-visibility-report-2026 and the links point there. The report's account of its 15 May scoring correction pass, which had been published on the Atlas methodology page, moved to the same address.
-
The report described one platform conflating Sollis Health with a similarly named company in its cited sources. A review of the retained captures found the same conflation on a second platform. The count of confirmed brand-integrity exceptions is corrected from two to three.
-
Finding 1 described the suppressed queries as practitioner-name queries. The retest they come from tested the bare [practitioner name] clinic template specifically, and queries naming the same practitioners in other templates were answered normally. The wording is corrected to state the template. The figure is unchanged.
-
The methodology page was missing from sitemap.xml and from llms.txt, and the method links on the report page pointed to an anchor inside the report rather than to the methodology page itself. A machine sent to source the report could reach the findings but not the method or the disclosure. Found and fixed the same day.
Reporting an error
Errors in the Atlas, in the research or in an article can be sent to [email protected]. A reported error is checked against its source. Where it is confirmed, the record is changed and the change is entered here on the day it is made.
This page records defects in published work. The separate question of how a record is kept current, the rolling re-checks and the date stamp that says when each source was last read, is answered under how it is verified.