Methods, GBP study public data v1.2

Design and sample

Google Maps local results were pulled for 9 small-business categories in 6 US metros (18 category-metro pairs), at 8 search points per metro spaced 10-12 km apart, on 2026-10-18. Each result was classed (owner-operated, franchise/chain, off-category, health system, lead-gen) with franchise rule v2. Listings for each pair were then pulled to find owner-operated businesses that never appear in the top 20 ("not seen"), and the websites of a 498-business study set were probed (homepage only).

Tiers: top3 (best rank 1-3 at any point), top10 (4-10), top20 (11-20), not seen (in the listings pull, never in a top 20).

Frames and the review-source rule

• Appearance frame (business × point), visible-business frame, listing frame (visible + not seen), study set, website frame.
• Within-tier and visible-only statistics use the review count shown in Maps results (Part 1). Any comparison that includes not-seen businesses uses the listing review count for every tier, so all tiers share one source.
• Service-area claims use the visible-business frame only: the listings pull under-matches service-area businesses (enrichment bias).

Classification

• Franchise rule v2: brand matches plus large multi-trade operators (review volume). Five study rows reclassified to franchise were excluded.
• Live checks (2026-10-18, read-only look at live Maps profiles and homepages): one further study row was a franchise brand's headquarters listing and was reclassed franchise and excluded; the DataForSEO claimed flag was checked against live Google on a 30-listing sample; a sample of duplicate-listing groups and website/listing phone mismatches was hand-checked.
• Primary-category relation (exact / adjacent / generic / other) uses a reviewed adjacency list per category.

Confirmatory and exploratory tests

Eleven confirmatory tests were written and locked before any statistic was computed (confirmatory_tests.md, SHA-256 c8c18f18e23db9bccf1119ccb0505685e43af4f047b7176e3cf4ea5fa0908ef3). CT-11 (lab page speed) was run in v1.1, after the PageSpeed pass finished. Results:

• CT-01: Review count goes with visibility: within a market, businesses with more reviews sit in higher tiers, the strongest association in the study. Grade High (GBP-D5-01).
• CT-02: Star rating does not separate top-3 businesses from not-seen ones in the same market: within-market AUC 0.47, near a coin flip. Grade High (GBP-D4-01).
• CT-03: Service-area businesses are less likely to reach the top 3 than storefronts with similar review standing (odds ratio 0.23). Grade Medium (GBP-B3-02).
• CT-04: At a given search point, businesses farther away tend to rank lower: rank worsens with distance at 74% of points. Grade Medium (GBP-C1-01).
• CT-05: Estate-planning attorneys whose primary category is "Estate planning attorney" are more often top 10 than those using Attorney or Law firm, at similar reviews and distance. Grade Medium (GBP-E1-01).
• CT-06: Listings in duplicate groups are not seen about as often as other listings in their market (65% vs 63% expected). Grade Medium (GBP-K2-01).
• CT-07: Low-review top-3 holders concentrate where competition is thin: 55% of top-3 appearances at thin points have <10 reviews vs 22% at full points. Grade Medium (GBP-D6-02).
• CT-08: Comparing reviews with the category benchmark tracks tier slightly better than comparing with the nearest point's top 3 (AUC 0.73 vs 0.71). Grade Medium (GBP-D3-01).
• CT-09: Homepages naming the city or a street address go with top-3 rather than not-seen status (94% vs 69%), also after reviews, claimed and distance. Grade Medium (GBP-G1-01).
• CT-10: Stale footer copyright years (2023 or earlier) appear on 27% of not-seen sites vs 2% of top-3 sites; not established after adjustment. Grade Low (GBP-G1-02).
• CT-11: Lab PageSpeed scores show no association with map-pack tier once reviews, claimed status and distance are controlled (median 58 top tiers vs 62.5). Grade High (GBP-G3-01).

All other tests are exploratory; Benjamini-Hochberg q is computed within each track. Intervals: stratified bootstrap within category-metro pair (2,000 reps) or cluster-robust standard errors by pair. Every candidate headline was re-estimated under a sensitivity matrix (v1 vs v2 classes, without plumbers, without generic-category estate attorneys, leave-one-pair-out, review source, enriched-only frame, distance control).

Grading

• High: comparison n ≥ 100 (or ≥ 60 per arm), interval excludes the null (null findings: interval inside a stated equivalence band), and the sensitivity matrix shows no flips and at most one weakening.
• Medium: clear direction with one known caveat, n 30-99, or two weakenings; descriptive levels are capped at Medium.
• Low: n < 30 per arm, confounded by design, a flip, or exploratory with q > 0.10.
• This release: 12 High, 63 Medium, 19 Low findings (none held).

Limitations

• One snapshot (2026-10-18): associations, not effects. Nothing here shows that changing a listing changes its rank.
• Search points are 10-12 km apart, so "points appeared" measures grid geometry, and distance is known only for businesses in the listings pull.
• Enrichment bias: the listings pull under-matches service-area businesses.
• Website checks read the homepage only; claimed status is the data vendor's flag, under manual validation.
• Page speed is lab data: one Lighthouse mobile run per homepage via the PageSpeed Insights API. No real-visitor (CrUX) field data was kept, so nothing here describes what real visitors experience; scores vary by several points between runs.
• Two metros per category; metro-level numbers are directional.

How to cite

datanephew (2026). Google Business Profile local visibility study, data release v1.2 (snapshot 2026-10-18). https://datanephew.com/research/local-pack-study/

License and data-source terms

• License: the data files and charts are released under Creative Commons Attribution 4.0 International (CC BY 4.0). Reuse them freely, including commercially, with credit to datanephew and a link to the study.
• Data sources: Maps results and listing data were collected through DataForSEO, which confirmed that publishing these aggregates is permitted. Lab page-speed scores come from the Google PageSpeed Insights API. Every file holds aggregate statistics only: no business names, identifiers, addresses or phone numbers, and any group of fewer than 10 businesses is suppressed.
• Google, Google Maps and Google Business Profile are trademarks of Google LLC. This study is not affiliated with or endorsed by Google.

Data downloads