The short version. Before HighRegard sells a street-level Maps map to anyone, the instrument that draws it had to be shown to measure reality and to bill what it says it bills. This note records that calibration: two live scans of 25 points each, run on 30 and 31 August 2026, one of them checked against a phone on the pavement, and three billing reconciliations that agreed with the provider’s ledger to the cent. It is a calibration with n = 2, not a market study, and it is written up so that the limits are as visible as the results.
What a geo-grid measures
Google guidance Google describes local ranking as a blend of relevance, distance and prominence. Distance means that the Maps result a customer sees depends on where they are standing, so a single “rank” for a town is not a measurement of anything. Measured observation A geo-grid puts a lattice of points over an area (here 5 × 5), runs the same search from each point, and records the business’s Maps position at every one. The output is a map of where the business is visible and where it is not.
Method
- Instrument
HighRegard’s own tracker (GridTracker v2.1.17) using a commercial local-search results provider in standard mode, google.co.uk, English, mobile results, top 20 positions recorded.
- Identity
Each subject business was matched to its Google place identifier (CID) and the identity verified before any ranking query was accepted; queries whose result could not be matched to the verified CID would have been held for review. None were.
- Grid
Five by five points centred on Google’s own pin for the business. Run 1 at 1.0-mile spacing (zoom 14); run 2 at 0.5-mile spacing (zoom 15). One keyword per run; 25 queries per run.
- Ground truth
For run 2 the founder stood at a known coordinate 0.63 miles north of the practice and searched the same phrase on a signed-in phone, then compared the phone’s result with the nearest measured grid point.
- Billing
Provider balance read before and after each run and compared with the tracker’s own booked cost and the provider dashboard. Threshold for a pass: agreement within one query’s cost.
Results
Run 1 — a laser clinic in Bradford, “laser hair removal bradford”, 1.0-mile spacing
| Measure | Value |
|---|---|
| Queries completed | 25 of 25 (0 failed, 0 held for review) |
| Average Maps position across the 25 points | 18.9 |
| Points where the clinic was in the top 3 | 1 of 25 (4%) |
| Points in the top 10 | 3 of 25 (12%) |
| Points where it appeared in the top 20 at all | 3 of 25 (12%) |
| At the clinic’s own door | Position 1 |
| 1-mile ring (8 points) | 6 not in the top 20; the two that were: positions 4 and 5 |
| 2-mile ring (16 points) | 16 of 16 not in the top 20 |
| Weakest half | West (average position 21 across that half, where one competitor most often held the top three) |
Measured observation The subject was the founder’s own clinic, Pulse & Clarity, which is why it can be named. For this one head term it is competitive only at its own address and absent from the Maps top 20 across most of the surrounding two miles. Professional judgement That is a real commercial problem for a clinic whose catchment is the city, and it was the first thing the tool found. Working hypothesis The cause is not established by one scan of one term; the next step is to repeat for “near me” and service phrasings and to compare against whoever holds the top three across the city.
Run 2 — a dental practice in Burnley, “dentist burnley”, 0.5-mile spacing
| Measure | Value |
|---|---|
| Queries completed | 16 of 25 (9 failed at the provider; cause not reported by it, see limitations) |
| Average Maps position across the 16 valid points | 17.3 |
| Points in the top 3 | 3 of 16 (18.8%) |
| At the practice’s own door | Position 1 |
| 0.5 miles north | Position 2 |
Measured observation The practice is not named; it has not agreed to be. Its result is the flattering kind, which is worth saying because a grid that only ever reports bad news would be suspect.
Ground-truth check
Measured observation Standing 0.63 miles north of the practice, the founder’s phone returned the practice at position 1 for the same phrase. The nearest successfully measured grid point, 0.5 miles north and 232 yards from where he stood, had recorded position 2. One position of difference across a quarter of a mile, between a signed-in phone at a different time of day with real GPS and a coordinate pin queried through a provider. Professional judgement That is inside the variation we would expect between two real searches, and it is the evidence that the instrument measures the thing customers actually see rather than an artefact of the provider.
Billing reconciliation
| Run | Balance before | Balance after | Provider dashboard | Tracker booked | Difference |
|---|---|---|---|---|---|
| 1 (Bradford) | $1.0000 | $0.9850 | $0.0150 | $0.015 | $0.0000 |
| 2 (Burnley) | $0.9850 | $0.9700 | $0.0150 | $0.015 | $0.0000 |
| Retry of run 2’s failures | $0.9700 | $0.9700 | $0.0000 | unchanged | $0.0000 |
Measured observation Three reconciliations, three exact matches, and a retry that cost nothing because it resumed stored tasks rather than re-posting them. A 25-point scan cost $0.015, or about six hundredths of a cent per point. Professional judgement That cost is what makes street-level Maps visibility something a small business can afford to have measured monthly rather than once.
Limitations
- n = 2 businesses, one keyword each, one day each. Nothing here generalises to a sector or a town.
- One results provider. A second provider has not yet been compared against it.
- Run 2 lost 9 of 25 queries at the provider and the provider returned no status code or message for them, so the cause is unknown. Failed queries were still charged. Until this is diagnosed, a partial grid is reported as partial, never filled in.
- The ground-truth check is a single observation from a signed-in phone at a different time of day. It supports “the instrument measures reality”; it does not establish the size of the variation between a phone and a provider query.
- The Bradford finding is one head term. It says nothing yet about “near me” phrasings, service terms, or the cause of the pattern.
- Both scans use mobile results on google.co.uk in English. Desktop results and other locales were not measured.
What follows from it
Professional judgement The calibration gate is closed: the tracker is fit to produce grids for clients, with partial scans labelled as partial. Working hypothesis The larger question this note cannot answer, whether the businesses Maps ranks are also the ones AI assistants recommend, is the first planned study on the research page, and it needs a panel rather than two scans.
Prefer HighRegard in Google’s AI answers and Top Stories