We Ran 72 Local Searches in One County. Here Is What Ranking Actually Looks Like

We ran the same 72 local searches twice, two days apart, across twelve Cleveland County towns. 43 different businesses held first place in at least one of them. Four packs in ten had changed who was in them by the second run. There is no single local ranking to hold here, and that changes what a local business should be paying for.

Nobody publishes this. Google does not, and a national study of a hundred thousand businesses cannot, because the question is what happens inside one small county rather than what happens on average.

So we counted it ourselves. The result, the method, and the part it cannot answer all follow.

What did we actually count?

Six business categories across twelve towns, run logged out from each town centre. Seventy two searches, zero failed, 576 listings returned. Those resolved to 239 distinct businesses, which we then opened individually on their Google profiles.

This is a census of what we searched rather than a sample of it, so no estimation enters the counts anywhere. The same searches will run again on the same settings, which turns a snapshot into a series.

Who holds first place around here?

43 different businesses. Across twelve towns and six categories, first place is not a position anybody owns. It is 72 separate answers, and 48.7% of the businesses we saw appeared in exactly one town's results.

Almost half of what shows up in local search here is visible in one town and invisible in the other eleven. Three businesses out of 228 appeared in all twelve.

A ranking report with one number on it told you nothing, and this is why. There is no county-wide position to report. A business can be first in Shelby, absent in Kings Mountain, and fourth in Boiling Springs, all in the same afternoon, and every one of those is the truth. The useful question is not where do I rank but which searches, from which towns, and against whom. We wrote about the distance side of this in how far a map ranking travels in a small town.

Does a local ranking stay put?

Mostly it does, at least over the two days we can speak to. We compared 72 packs across two scans 48 hours apart on identical settings. 56.9% had exactly the same businesses in them. Mean membership churn was 5.9%, and the top result changed in 5.6% of packs.

Two findings pull in opposite directions here. Forty three percent of packs were not identical two days later, so the list a customer sees is a moving one. At the same time, the business on top changed in about one pack in twenty. The churn is happening at the bottom of the list, where businesses drift in and out of positions five, six and seven.

For an owner that means getting to the top is the hard part and staying there is not, at least at this timescale. It also means a rank check on a Tuesday and a rank check on a Thursday can disagree without anything having gone wrong.

Do reviews decide who goes on top?

Not inside a pack. Median review count at slot one was 66. At slot seven it was 77. The spread in median reviews across all seven slots was 22, and mean star rating varied by 0.12 from top slot to bottom.

The business in first place was also the most reviewed business in the pack in 18.3% of cases. In a seven result pack, picking at random would land you near 14%. So the most reviewed business wins the top slot a little more often than chance, and nothing like often enough to call review count the thing that orders a pack.

What the flat result does not license is the obvious conclusion. Proximity to the search point is uncontrolled here and plausibly dominates the ordering, which would mask a review effect sitting underneath it. A single snapshot also cannot test recency or velocity, and velocity is the signal the wider research keeps pointing at. Nothing here contradicts it, and we covered the distinction in review velocity versus total count.

How many reviews does a business here actually have?

Across 229 businesses visible in local search here the median was 50 reviews, which is fewer than most owners assume. A quarter sat at 15 or fewer, and 18.8% had fewer than ten.

Star ratings tell the opposite story. The mean rating was 4.68, 83% of businesses sat at 4.5 or better, and only 2.6% were below 4.0.

Review count still separates businesses here, because the distribution is wide and the bottom quarter is genuinely thin. Star rating mostly does not, because almost everybody already has a good one. A business sitting at 20 reviews and 4.7 stars is closer to the middle of this county than it feels, and the cheapest ground to gain is on the count. What ranking factors do and do not carry weight is covered in the 2026 ranking factors guide.

What does the web presence look like?

48.1% of the businesses we surfaced have no website listed on their Google profile. That is 115 of 239, and the rate varies from 45% to 78% depending on the town.

Two cautions travel with that figure. A blank field means no website is listed on the profile, not that no website exists. A business may own one and never have linked it, which is a listing problem rather than an absence, and from the outside those look identical. And because this only sees businesses that already rank, the real county figure is almost certainly higher than 48%.

What can this data not tell you?

Quite a lot, and every limit below changes how a figure above it should be read.

• This is 239 businesses surfaced by six categories across twelve towns. It is not every business in Cleveland County.
• A missing website means none is listed on the Google profile. It does not prove none exists.
• The 48 hour movement figure is short-term flicker. It says nothing about seasonal or year over year change.
• Slot ordering is measured inside packs Google already built. It cannot speak to what gets a business into a pack.
• Proximity to the search point is not controlled for, and it plausibly dominates ordering.
• Nineteen records carry an address that does not resolve to a town, so the per town split covers 220.
• Nothing here is causal. One snapshot cannot support a claim about what causes what.

When does this get updated?

Every quarter, on the same settings, with the change published whether or not it flatters us. Identical method pins are what make two runs comparable, so a re-run that changes the zoom, the viewport or the logged out state is a different instrument and its numbers do not belong beside these.

One scan is a fact and it ages. The same scan repeated is a trend, and only we hold the July 2026 baseline, so the Q4 comparison is not something a competitor can go back and collect. From the third run the movement becomes the story and these levels become the supporting detail.

If you run a business in Cleveland County and want to know what your own map coverage looks like across the towns near you rather than in the abstract, Studio O'Brien runs the same scan against a single business. Ask for a county scan and we will send the raw positions along with what they mean.

Common questions