Original research
20 local SEO statistics for Australia and New Zealand
Almost every local SEO statistic in circulation is American, and most of the rest are a citation of a citation. These are measurements I took myself, in Australian and New Zealand cities, and I have published the method so they can be argued with.
The sample is 480 grids, 231,907 observations, 20 industries, 24 AU/NZ cities. Every figure below links back to the full study, including the parts that did not survive a larger sample and the three predictions I got wrong.
Use these figures
You are welcome to quote, chart or republish anything on this page, commercially or otherwise, with a link back. No permission needed and nothing gated. If you want the underlying per-industry or per-city numbers rather than the summary, get in touch and I will send them over - I would rather this got checked than admired.
Tom McSherry, "Map Pack Visibility Study 2026" (480 grids, 231,907 observations, 20 industries, 24 AU/NZ cities). https://tommcsherry.com/learn/map-pack-visibility-study-2026Competition decides your visibility, not your industry
This is the headline finding, and it is the one that changes what you should do. Sort every grid by how many businesses compete inside the same 8 by 8 km square, and how far a business holds a top-three position falls away cleanly. It is the only cut in the study where the confidence ranges do not overlap.
4.63 km vs 0.73 km
In the least crowded quarter of markets a business holds a top-three Maps position for 4.63 km in every direction. In the most crowded quarter it manages 0.73 km. Same country, same searches, often the same industry.
Median top-3 reach, quartiles by competitor count (5 - 46 against 103 - 374 businesses per 8 x 8 km square). 95% ranges 3.81-5.59 and 0.69-0.78, which do not overlap.
6.3x
The spread between the thinnest and densest markets. Every step down is monotonic and no range overlaps its neighbour, which is not true of any other cut in the study.
4.63 km / 2.78 km / 1.47 km / 0.73 km across the four quartiles.
32.6% to 27.8%
The share of businesses that reach the top three somewhere barely moves as markets get crowded. Competition does not change how many businesses are visible - it changes how much ground each visible business covers. The pie is always cut into about the same number of slices; the slices get smaller.
Contestedness by competitor quartile, against a 6.3x move in reach over the same quartiles.
Every adjacent pair
Cut the same grids by industry instead and every neighbouring pair of confidence ranges overlaps. The industry table supports "a cafe reaches roughly an eighth as far as an arborist" and does not support "a physio beats a psychologist".
20 industries, 24 grids each. Density separates; industry does not.
How far local visibility actually travels
Reach here means the distance from a business’s own address at which it still holds a top-three Maps position. Not where it appears at all - where it appears in the three results people click.
1.87 km
Median top-three reach across the whole sample. Across the original 12-city core panel, which is weighted more towards metros, it is 1.35 km.
480 grids, 231,907 observations, 20 industries, 24 AU/NZ cities. The frame is deliberately weighted to secondary centres rather than population, so absolute distance runs a little wide; both figures are reported for that reason.
6.17 km to 0.46 km
The range across industries, from pest control at the top to cafe at the bottom - a factor of about 13.
Medians per industry, 24 grids each. Read the ends, not the order: adjacent pairs overlap.
67%
Of every appearance a business makes within 1 km of its own front door, this share is NOT in the top three. Standing on your own doorstep, two appearances in three are outside the positions anyone clicks. Proximity is necessary and nowhere near sufficient.
95% range 66 to 69. Worst: mobile mechanic at 79%. Best: locksmith at 26%.
About 7 in 10
Businesses appearing in local packs that are never visible in a top-three position anywhere in their own city. Contestedness clusters near 30% and barely shifts with how crowded the market is.
From locksmith at 46.8% down to mobile mechanic at 18.6%.
How fast you disappear as people move away
Top-three share by distance from the business’s own pin, in exclusive bands. The cliff is in the first kilometre and everything after it is a slow decay.
32.0% to 18.5%
Top-three share nearly halves between the first and second kilometre, then decays slowly. Past about 3 km a business that still appears is appearing around rank 11 or 12, which is not appearing in any sense a customer experiences.
All-industry falloff across the bands: 32.0% / 18.5% / 15.8% / 13.5% / 12.2% / 9.4% / 8.1%.
63.7%
The best doorstep performance in the study, from locksmiths - a thin market, so they win their own area outright, and still a physical call-out business, so they fade with distance.
Falls to 11.0% beyond 8 km.
21.4% to 14.2%
Mobile mechanics have no cliff at all. The only industry measured where distance from the registered address is nearly irrelevant, which is what a genuinely address-less business looks like in this data.
The single exception to the density pattern, and the only industry with a flat falloff curve.
What it would take to cover a city
Reach defines a disc, and discs cannot tile a plane. At hexagonal packing a site usefully covers about nine tenths of the circle around it. Run that over a 400 km2 urban core - roughly a 20 by 20 km serviceable area - and the multi-location question stops being a matter of ambition.
3.7 vs 677
Sites needed to cover a 400 km2 city: about 3.7 for pest control, about 677 for a cafe. For a service-area business blanketing a city is a handful of sites and a real strategy. For a dense premises business it is arithmetically impossible.
Derived from each industry median reach at hexagonal packing density (0.9069). 14 of 20 industries.
Own a suburb
The honest advice for a dense premises business is to stop trying to cover a city. A physio group with four clinics is not covering a metro; it is covering four neighbourhoods, and it should be planned and measured that way.
Follows directly from the sites-to-cover figures above.
Reviews and ratings
Measured the way a factor study should be: Spearman’s correlation computed among the businesses competing at one single grid point, then averaged. Pooling across points would conflate how competitive a point is with how businesses rank within it, which is the error most published factor studies make.
0.194
The whole-sample correlation between review count and Map Pack position. Real, and far more modest than "get more reviews and you will rank" implies. A correlation of about 0.2 is a tendency, not a lever.
95% range 0.174 to 0.213, across 480 grids.
0.428 to -0.099
The spread by industry, from plumber at the top to cafe at the bottom. Cafe is reliably NEGATIVE - in the densest premises markets the businesses with the most reviews sit slightly further down the pack, because the pack is being decided by proximity and the heavily-reviewed places are not the closest ones.
cafe interval -0.152 to -0.046, which excludes zero, negative in 20 of 24 grids.
-0.005
Star rating against Map Pack position: as close to a clean null as observational data produces. Rating remains a conversion factor - being seen at 3.9 stars and being chosen at 3.9 stars are different problems, and this study measures only the first.
95% range -0.02 to 0.009, positive in 230 of 480 grids - direction decided by a coin flip.
0.229
Within an industry, grids where visibility travels further are grids where reviews correlate more strongly with position. Reviews matter more where proximity decides less. The industry-level version of that sentence is NOT established and an earlier write-up was wrong to state it.
95% range 0.093 to 0.363, 15 of 20 verticals. Across industries it is 0.287 [-0.199, 0.713], which includes zero.
The sample, and what it cannot tell you
Published so the figures above can be argued with. A statistic without a method behind it is a slogan.
231,907
Google Maps results captured across 480 usable grids - a 5 by 5 pattern of search points in each of 24 Australian and New Zealand cities, 20 industries, 41,716 business-grid pairs.
Total API cost US$23.95. No grid exceeded the 15% dead-point threshold.
Correlation
None of this is causation. Businesses with more reviews tend to be older, better established and better linked, and those travel together. The study measures what accompanies a good position, not what causes one.
Stated as a limitation in the published method rather than buried.
3 of 12
Pre-registered predictions that failed. Every prediction was recorded in the frame file before a single grid of the new industries was collected, because recording it afterwards would have been worthless. Three were wrong, and all three are published.
Vet, chiropractor and gym were all predicted under 1 km and came in at 2.85, 1.86 and 1.15 km - all premises industries in thin markets, which is the cleanest evidence for the density finding.
Read the study these come from
The full write-up has the method in full, the confidence ranges, the limitations, and the figures that did not survive when the sample doubled.