If you put a number in front of a paying client, you should be able to say where it came from. This is all of it.
Updated 21 October 2026 · 8 min readHarbourside Plumbing & Heating is a fictional business from our demo seed. These generated results explain the report; they are not a customer’s rankings.
We build a square grid of points centred on the business, with an odd number of points along each side so the premises always sit in the middle. Grids run from 3 × 3 up to 15 × 15, depending on your plan.
You can also shape the area yourself in the map designer: a circle, a square, or an outline you draw around the streets you actually serve. Distances can be set in miles or kilometres. What you see in the designer is exactly what the scan checks.
Default spacing is sized for British towns. Setup offers four areas: 250 metres apart for a walk-in shop or café, 500 metres for most clinics, salons and trades, 1 kilometre when customers drive in from across town and 2.5 kilometres for mobile trades covering a county.
Longitude is corrected at every point for how far north the business is, so grids stay square from Cornwall to Caithness.
Map zoom is fixed for each spacing bracket and stored with every scan, so results stay comparable week to week. At each point we read the top 20 results and match the business on Google's own ID for it where we have it, otherwise on its name.
Google shapes its results for whoever is searching. Where you're standing counts most: move a few streets and the order changes. It also weighs what you've searched and clicked before, and whether you're signed in. On your own phone you're often at the premises, where most businesses rank best, and Google knows the listing is yours.
Kerbrank searches from every point on the map as a new visitor: not signed in, with no search history, the same way every week. That's why one reading compares fairly with the next. To see roughly what a customer sees, open a private window and search from where they'd be.
Before a scan, we check every point against UK postcode data. A point with no postcode nearby, such as open water, a harbour or empty country, is left out of the scan and out of every score.
Nobody searches from the Solent, so nobody should be marked down for not ranking there. If the postcode lookup is unavailable, every point is scanned as normal.
The share of grid points where the business appears in the Google Maps top three for that search. The box of three on a normal Google search (the local pack) is a separate list, measured only when the mobile pack check is on, and labelled as such.
The average position across every point. Anywhere the business is not in the top 20 counts as 21, so a business that only shows up near its door can’t look better than it is. Lower is better.
The average across only the points where the business does appear. It flatters: visible at 2 points out of 49, both in first place, scores 1.0. We show it, but never on its own.
The share of points in the top ten. It tends to move before the headline does, so it shows early progress.
Groups of three or more neighbouring points where the business does not appear at all, named by compass direction.
Visible at 2 points out of 49, both in first place. The same scan, three ways:
We keep two kinds of gap apart. Not found means we looked and the business was not in the top 20. No data means we could not get an answer for that point. The first counts against you. The second is shown as missing and is never scored as a bad ranking.
Guide: reading the map and the headline numbersWe ask ChatGPT the kind of questions a customer would, built from the business's category, town and services: “best plumber in Portsmouth”, “who should I call for a boiler repair near Southsea”, and so on.
Language models often give a different answer each time, so a single question is noise. We ask each question three times and report how often the business was recommended, with a 95% confidence range. That is why you will see “8 of 24 answers (33%), likely between 18% and 53%” rather than a bare figure that jumps about every week.
We also record which businesses were recommended instead, and which websites ChatGPT cited. That second list often matters more: if it keeps reading one directory for your trade, being missing from that directory is the real problem.
For each search you track, we show how many people search for it each month in your area, taken from Google Ads data. Being first for a phrase nobody types is worth very little, so where demand is low, Kerbrank suggests searches that are worth tracking instead.
Choosing which searches to trackWe fetch the business's website and read what is really there: its structured data, name, phone number and address. We then check that the phone number and postcode on your Google profile appear on the site. If we cannot load the site, the report says it could not be checked. It does not guess.
Findings come from the business's Google profile, the profiles of the competitors around it, the grid results, the website check and the ChatGPT answers. Each one is ranked by how serious it is, then by likely effect against effort. The report opens with one sentence and the top three jobs.
Likely effects are a guide, not a promise. They are weighted by Whitespark’s Local Search Ranking Factors survey, in which primary category, distance and being open at the time of the search count for most.
If you connect Google (Search Console, and your Business Profile or Google Analytics if you link them), we read clicks, calls and searches from it to add context. The connection is optional. We never change your Business Profile on our own: posts go out only when you press Publish or schedule them.
Every figure carries the date it was measured. We keep the raw results of each scan, so if we add a new measure later we can work it out across your history rather than asking you to start again.
Demo generated Tue 22 Sept 2026, 09:56 · 81 points · 3 searches Working through your job listWe don't sell citation building, we don't keep a database of scraped personal contacts, and we don't write your review replies for you. An obviously automated reply to a customer does more harm than no reply at all.
Can’t tell where a number came from? Tell us, and we’ll explain it or fix the page.
Every map is measured exactly as this page describes. We show your estimated scan time before you start. Timing depends on your map, searches and Google’s queue. Free for 14 days, and you don't need a card.
The score out of 100 combines Google Maps top-three share (50%), a position component (30%) and the percentage of sampled ChatGPT answers naming the business (20%). The position component is 100 × (21 − typical position) ÷ 20. Position 1 scores 100; position 21, including results outside the top 20, scores 0.
All three measured inputs are required. Missing or unavailable answers are not treated as zero.
The trend compares the latest two readings only when the map, searches and measurement method are comparable, the ChatGPT questions match, and the number of sampled answers matches. The score is rounded to one decimal place. A change is an observed difference, not proof that work caused it.
This is Kerbrank’s index, not a Google score, a percentage of customers or a forecast of sales. The separate value-weighted score in Insights weights searches by commercial value; it is not substituted into this formula.