A business has no single Google Maps rank, only a position at each point a searcher stands

A number that only exists somewhere

I built the geo-grid tracker that HighRegard uses to measure Google Maps visibility, ran live scans around my own clinic, and then did the thing that most map reporting never does. I went outside with a phone and checked the output at grid points. At the centre point the grid and the phone agreed exactly. At a point 232 yards away they differed by one position. The runs, the failures and the costs are written up in HighRegard's geo-grid calibration note ^[16].

I am a prescribing pharmacist, and I built my clinic's website, its Google Business Profile, its analytics and its tracking myself before HighRegard existed. The tracker was pointed at my own front door before it was pointed at anybody else's.

That small disagreement is the whole subject. Two searches for the same phrase, minutes apart and a short walk apart, did not produce the same order of businesses. Neither result was wrong.

A local business does not have a position on Google Maps. It has a position at every point from which somebody might search, and the enquiry goes to whichever profiles hold the top of the pack where that person happens to be. A single tracked number, taken from one place, describes one point on a surface that covers a whole town.

The commercial consequence is blunt. An owner told they sit third is being told something true about one spot, usually the town centre or their own doorstep. The streets where they sit eighteenth never appear in the report, and neither do the competitors who take those enquiries.

What Google says it is doing

Google is unusually explicit about local ranking. Its Business Profile documentation says local results are based primarily on relevance, distance and prominence, and defines each one: relevance is "how well a Business Profile matches what someone is searching for", distance "refers to how far each business is from the customer who's searching", and prominence "means how well-known a business is" ^[1].

There's no way to request or pay for a better local ranking on Google. ^[1]

Distance is not a property of the business. It is a calculation between the business and whoever is searching, made again for every search. Google's own page on the subject says that if a customer does not share where they are, it uses what it knows about their location ^[1].

What it knows comes from several places. Google's search location documentation lists precise device location on phones, computers and wearables, saved home or work addresses used when it is likely you are at one of them, areas you have searched for in the past, and IP addresses, which it says are roughly based on geography ^[4]. Danny Sullivan, Google's public liaison for Search, wrote in 2020 that when a topic has a local aspect the company shows "a mix of local results that are relevant to particular places along with non-local results" ^[5].

Google's explanation of ranking adds that it uses information such as location, past Search history and Search settings to work out what is most relevant for a given person ^[3]. So two people standing in the same doorway can still see different packs.

What Google does not publish is any weighting. It names three factors and refuses to rank them, and it has never said how sharply distance decays or how far relevance and prominence can carry a business against a nearer rival. Everything beyond the three names is inference.

Why one tracked rank tells you very little

Consider what a single tracked position is actually a sample of. One query, one location, one moment, one device configuration. If the quantity being measured varies continuously across the map, one sample cannot describe it, in the same way that one temperature reading cannot describe a county.

The profile's own reporting does not fill the gap. Google's performance data gives the search terms people used, views on Search and Maps, and interactions such as calls, direction requests and website clicks, and it is available only for verified profiles ^[9]. That is a record of the people who did find the business. It says nothing about the streets where the business was never shown, because absence leaves no data.

Industry surveys point the same way without settling it. Whitespark's 2026 Local Search Ranking Factors report asked 47 local search practitioners to score 187 factors, and its local pack list put the primary Business Profile category first, the proximity of the address to the point of search second, and keywords in the business title third ^[10]. BrightLocal's contributor survey similarly attributes about a third of local pack importance to Business Profile signals and about a fifth to reviews ^[11].

Those are weighted expert opinions, not measurements of the algorithm, and they should be read as such. They are useful because a large group of people who watch local results closely broadly agree with what Google says out loud, not because anybody has seen the code.

The size of the competitive area is not constant either. Search Engine Land's guide to proximity bias argues that Google reads some terms as having neighbourhood level intent, which tightens the radius, and that businesses in thinly contested markets can hold the pack across a much wider area than those in dense ones ^[13]. The same profile, unchanged, will therefore produce a broad shape in one town and a narrow one in another.

The weighting is also not fixed. In late 2021 Search Engine Land reported on an update that practitioners called Vicinity, describing it as a rebalancing in which "smaller, less established businesses that are closer to the user are now ranking better than overpowered businesses further away" ^[14]. Search Engine Journal's assessment of the evidence treats physical proximity as a confirmed ranking factor while noting it had slipped down one industry survey's ordering ^[15]. Distance matters, and how much it matters has moved before.

What a geo-grid measurement actually is

A geo-grid scan runs the same query from a lattice of coordinates spread across an area and records where the target profile appears at each one. Instead of a number you get a field of numbers laid over a map, which can be read as a shape: the area where the business is competitive, the edges where it fades, and the points where somebody else holds the pack.

HighRegard's tracker works this way. The calibration note describes scans of five by five grids, one at one mile spacing and one at half mile spacing, recording mobile results and the top twenty positions on google.co.uk, at a cost of roughly $0.015 for a twenty five point scan ^[16]. It also records what went wrong: nine of the twenty five queries failed in the second run, and the note publishes that alongside the averages ^[16].

Publishing the failures matters more than publishing the averages. An average position across a grid is a weak summary, because it mixes a strong home cell with a wide band of points outside the top twenty and lands somewhere in the middle that describes nothing. The more useful reading is the count of points at which the business sits in the top three, tracked against a dated baseline.

The reason I checked the grid on foot is that a tracker nobody has ever tested against a real phone is an untested instrument. It produces confident output regardless of whether it is right. My check was small, and it agreed at the centre and differed by one position a short walk away, which is what I would expect from two genuine searches rather than from a broken measurement.

What a grid point cannot tell you

A grid is a proxy. It is the best available proxy for a quantity that cannot be observed directly, since nobody can stand on every corner of a town at once, but the gap between the proxy and reality is real and should be stated whenever grid results are reported.

• A grid point is a coordinate, not a person. Google's documentation describes estimating a general area rather than a precise point for search, larger than three square kilometres and containing at least 1,000 users, unless someone has granted precise location access ^[4].
• Results vary with the searcher as well as the place. Google says it uses past Search history and Search settings alongside location when deciding what is most relevant ^[3].
• Results vary with time. Whitespark's respondents placed whether a business is open at the time of search among the strongest local pack factors ^[10], so a grid run at ten in the morning and one run at nine at night are not measuring the same thing.
• A grid carries no demand data. Points are spaced evenly across a map; customers are not. A weak cell over farmland and a weak cell over a housing estate look identical on the grid and are worth entirely different amounts.
• A position is not a click. Holding the top of the pack at a point tells you nothing about whether anyone searched there, or chose you if they did.

Used properly, a grid is a comparison instrument rather than a truth machine. Its value is that the same query, the same spacing and the same tool produce comparable pictures a month apart, so movement can be seen. Its value is not that any single cell is a promise about what a particular person saw.

What changes the shape of the area, and what does not

Distance is the factor an owner cannot touch. The building is where it is. Everything that widens the competitive area therefore has to come from the other two factors, and Google's documentation is reasonably clear about where the levers sit.

Categories and the profile itself

Google says plainly that "the categories you select affect your local ranking on Google", and that a primary category of pizza restaurant may cause a profile to appear for searches such as restaurants, Italian restaurants or pizza ^[6]. It advises choosing the most specific category available, giving nail salon rather than salon as the example, and warns against adding a category for every product or service ^[6]. This is the cheapest correction available to most businesses and one of the most frequently wrong.

The profile has to describe a real place. Google's guidelines require a precise, accurate address or service area, exclude PO boxes and mailboxes at remote locations, and require permanent fixed signage at the address ^[2]. A service-area business may not list a virtual office unless that office is staffed during business hours ^[2]. Breaking those rules is not a shortcut with a small penalty attached; it puts the profile itself at risk.

Service areas

Service areas are widely misunderstood as a coverage control. Google's guidance describes them as a description of where a business travels, allows up to 20 of them, and says the overall boundary should not be more than about two hours of driving time from where the business is based ^[7]. The page sets out what a service area is and how to set one; it does not present the setting as a way to rank across those areas ^[7]. Adding towns to a service area does not move the premises closer to anybody.

Reviews and prominence

Prominence is the slower lever. BrightLocal's 2026 Local Consumer Review Survey, a panel of 1,002 US adults, found that 97 per cent of consumers read reviews for local businesses and that 74 per cent look for reviews written in the last three months ^[12]. That is a United States sample and a survey of stated behaviour rather than observed behaviour, so it is evidence about what people say they do, not a measurement of UK enquiry patterns.

Whatever the number, the acquisition method is constrained. Google's Maps user generated content policy prohibits reviews that have been paid for directly or in kind, and specifically forbids offering incentives such as payment, discounts or free goods and services in exchange for posting a review ^[8]. A review programme that quietly selects only happy customers is buying a problem rather than prominence.

What this means for a local business

The practical implications are narrower than the theory, and most of them are about measurement discipline rather than tactics.

1. Ask any provider reporting a map position where it was measured from, across how many points, at what spacing, on what date and on which device type. A position with no location attached is not a finding.
2. Take a baseline grid before work starts, then repeat it with identical settings. Comparisons between two different tools, spacings or query sets tell you about the tools.
3. Report the count of points in the top three rather than an average position, because averages across a grid hide both the strong home cell and the dead outer band.
4. Weight the grid against where customers actually are. A gain over an industrial estate and a gain over a dense residential area are not equivalent, and no grid will tell you which is which.
5. Fix the profile before buying coverage. Category choice, accurate address details and category specificity are documented influences on local ranking ^[6] and cost nothing but attention.
6. Treat a wide service area as a description of where you will travel, within Google's stated limits, rather than as an instruction about where to appear ^[7].

There is a harder implication for anyone selling or buying local search work. If a business is genuinely competitive only within a few hundred metres, the honest report says so, and the honest plan is about widening that area slowly through relevance and prominence rather than about a number that can be quoted in a meeting.

What the evidence does not settle

Several things in this piece are firm and several are not, and they should not be read at the same strength.

Firm: Google names relevance, distance and prominence, says position cannot be bought, and documents how it estimates a searcher's location ^[1][4]. Category choice is stated by Google to affect local ranking ^[6]. Review incentives are prohibited ^[8]. Those are documented policy positions from the company that runs the system.

Not firm: every claim about how much each factor matters. The ranking factor surveys are careful and useful, but 47 practitioners scoring 187 factors are describing a consensus impression, not an audit ^[10], and the same applies to contributor-weighted breakdowns of local pack importance ^[11]. I could not find an independent, peer-reviewed measurement of how local pack composition varies with distance across UK towns, and the trade coverage of past updates, including the 2021 rebalancing towards nearer businesses, is reconstruction from observed rankings rather than confirmation of a mechanism ^[14].

My own check is one comparison, at one pair of points, on one day. The grid and the phone agreed at the centre and differed by one position 232 yards away, which is consistent with a tracker that is working. It is not evidence that every cell in every scan matches what a person standing at that cell would see, and I have not run enough on-foot checks to claim a general error rate.

The ground is also moving underneath the question. BrightLocal's US panel reported Google's use as a review site falling from 83 per cent in 2025 to 71 per cent, with AI assistants climbing the list of places people ask for recommendations ^[12]. If some share of local discovery migrates to systems that answer in prose and cite a handful of sources, a map of pack positions will describe a smaller part of how people actually choose a local business, and nobody yet knows how large that share becomes.

What remains true for now is smaller than the usual claim and more useful than a single number. Maps visibility is a surface, it can be sampled, the sampling can be checked against reality, and the honest way to report it is with its date, its spacing, its failures and its limits attached.