What changed, and what stayed.

September 11: updated local story leads

The article, key findings and media brief now lead with the reviewed 139/244 Google-to-AI gap, contrasting local names across assistants, and restaurant and dentist breakdowns. Directory and AI Overview findings remain supporting research. The selected 6,000 observations, original ten evidence-ledger findings, prompt set and original visibility tables have not changed.

This is an exploratory follow-up and editorial revision, not a new collection or evidence of business improvement over time. The reviewed cohort and sector counts are separate from the original frozen evidence ledger.

Updated story facts and sector denominators · Editorial revision and document checksums

Editorial version: editorial-2026-09-11-v1 · Evidence edition: lubbock-2026-09-descriptive-study-v2

Original v2 article and media brief

These are the documents before the September 11 editorial update, retained byte-for-byte for earlier citations. They include the older lead stories, not the later reviewed citywide statistic.

• Original article (Markdown)
• Original media brief (Markdown)
• Original article (PDF)
• Original media brief (PDF)

Original document checksums

Current study v2: 6,000 usable observations

The September 2026 release uses the completed selection. Six ChatGPT responses were recovered using the same questions. One personal-injury question was replaced across all six configurations, with ten repetitions each. This changes the sampled question set; it does not measure business improvement over time.

Original snapshot v1: 5,984 usable observations

This is the original collection and publication-preparation snapshot, retained byte-for-byte. It was saved before this public launch, not a previously published edition. Its older counts, preparation labels and deployment flags describe that earlier state. Use study v2 for the current findings.

Original checksums · Original verifier · Reproduction functions

Preservation and corrections

Released edition files are frozen with SHA-256 checksums. A correction receives a new version and a change log; it does not overwrite the cited snapshot. Preparation flags in downloaded study metadata record the state before deployment. They are not a live deployment-status feed.

Current release checksums · Count changes · Public page inventory and indexing

Contact caleb.m@growwithmonsoon.com with a correction, evidence key and supporting information. Comparable future collections are planned every three to six months and require a new collection window.