Public Google Maps listing measurements and US Census tract figures from three scans of a local caterer search in Fayetteville, Arkansas, October 2026

What this is. Public Google Maps search results joined to US Census figures, collected around one food-service business in Fayetteville, Arkansas, to see which listing features move together with map position. It uses three scans of the same local search, with UTC dates 2026-10-10, 2026-10-30 and 2026-10-31. Each scan searched from a 13 by 13 grid of 169 points spread over about 3.7 miles and recorded the top 20 listings at every point, ads left out. That gives 10,111 rows, where a row is one listing seen at one point. The three scans saw 34, 43 and 35 distinct listings, 49 across all three. Listing names are replaced with made-up labels, and Census figures are for the tract holding each listing's map pin. How to read the numbers. Results are Spearman rank correlations from -1 to 1. Position 1 is best, so a positive figure against map position means more of that feature went with a worse position. Each scan is its own result and the median of the three is quoted, with the three results in brackets. Main results (87 comparisons made within these three scans, 57 calculable; with that many, some will look stronger than they are): - Review count, 0.34 (0.33, 0.59, 0.34). Listings with more reviews tended to sit lower in the results. Among the other listings, positions 1 to 3 had a median of 29 reviews and positions 11 to 20 a median of 553. - Photo count, 0.38 (0.39, 0.38, 0.35), the same direction. Medians were 72 photos at positions 1 to 3 and 378 at positions 11 to 20. - Rating, -0.23 (-0.23, -0.12, -0.24). Higher-rated listings tended to sit slightly higher. - Claimed listing, 0.26 (0.27, 0.23, 0.26). Only 3 of the 49 listings were unclaimed, so this result rests on those three. - Extra categories listed, 0.27 (0.26, 0.43, 0.27); days a week with opening hours, 0.27 (0.27, 0.09, 0.27); website shown, 0.17 (not calculable in the first scan, where the value never changed). - Distance from the scan point to the listing's pin, -0.11 (-0.16, 0.04, -0.11). Next to nothing, and the sign changed between scans. - Star shares, price level and online booking were not recorded in these scans, so they could not be calculated. Within one category. Pooled over the three scans, among the 25 distinct listings whose primary category is Caterer (6,112 rows), review count against position was 0.40 and photo count 0.50, in the same direction. So the pattern is not only restaurants and bars with many reviews ranking lower than caterers. The other categories hold one to six listings each and are not interpretable. Across trades. results.csv also holds 690 further comparisons (rows whose section is "comparison") that total every scan saved in the analysis tool by Google category, these three included, so 777 comparisons in all. Those scans cover other searches and trades as well. They name no listing and no business, and they are not part of the three-scan results above. Census. For the 48 distinct listings with Census figures, review count showed no relationship with any tract measure (-0.08 to 0.09). Photo count was associated with tract education (0.31) and home value (0.24). Rating was associated with tract income (0.31), owner-occupied share (0.31) and median year built (-0.28). With 47 or 48 listings these are weak evidence. The business the scans were built around (Rx Catering NWA) was in the top 20 at 166, 169 and 169 of 169 points, with median positions 14, 8 and 13. Time of day. The first and third scans started at about 8 p.m. US Central the evening before their UTC dates, and the second at about 9 a.m. Central. The morning scan saw more distinct listings (43 against 34 and 35) and the largest review-count result (0.59 against 0.33 and 0.34). Three scans cannot separate time of day from anything else that changed. Limits. Only the top 20 at each point is seen. The same listing appears at many points, so the row counts overstate how much independent evidence there is; 34 to 43 listings per scan is the better guide. The grid was centred on this business, so it is not a random sample of any market. A map pin may not be where a business works. One search term, one metro area and three scans over three weeks. The results show things moving together, not one causing another. Authors Jay Ramsey Rx Catering NWA Corresponding contact Jay Ramsey Rx Catering NWA 121 West Township Street Fayetteville, AR 72703 Telephone: (479) 502-9879 Web: https://rxcatering.net/