Automatic grid or chosen tracking points: which method for measuring local visibility

One chosen measurement point beats ten generated at random. Here is why, what each method actually costs, and how to build a sampling plan in thirty minutes.

Between an automatic grid that calculates its locations and declared tracking points you choose one by one, the second method almost always produces a more useful measurement for less money. The reason is simple: a grid distributes its points by geometry, a sampling plan distributes them by market. Here is how to decide, and how to build that plan.

What a regular grid does, and what it measures by accident

A grid places its points by calculation: a centre, a radius, a density. Coordinates land where the geometry sends them — on a busy high street as readily as on a roundabout, on a residential estate as readily as on a quarry.

On a dense, homogeneous territory that is acceptable: most points describe something. As soon as the territory is uneven — and it almost always is once you leave a town centre — a significant share of the measurements lands where nobody searches for your trade.

Those points are not neutral. They enter the average, they enter the top 3 share of voice, and they move the trend line for reasons unrelated to your market. A client whose share of voice drops from 38% to 34% deserves an explanation; "four points on an industrial estate changed colour" is not one.

Declared points: real neighbourhoods, competitors, traffic routes

The opposite method is to name the places you want to be visible in, then measure only those. Five families of location cover most needs.

The five useful location families

  1. Real neighbourhoods. Where your customers live or work, named the way they name them.
  2. Competitor addresses. Measuring from a competitor's front door tells you exactly who wins on their doorstep — and whether you exist there at all.
  3. Retail parks and business districts. High daytime search volume, and behaviour quite unlike residential areas.
  4. Traffic routes. Town entrances, stations, motorway junctions: where people search "open now" or "near me".
  5. Neighbouring towns. Often a significant share of revenue, and almost always outside the radius of a grid centred on head office.

The decisive advantage is not technical, it is presentational: every location carries a name. "We are second in Didsbury and eighth in Salford" is understood in one second by a business owner. "Cell D4 is amber" needs explaining at every meeting.

Cost per measurement: the calculation nobody runs

A measurement is a keyword × location pair. A setup's budget is the product of keyword count, location count and reading frequency. The table below compares two setups at roughly equal cost.

7 × 7 gridDeclared points
Locations4912
Keywords tracked310
Measurements per reading147120
Locations with no commercial stake10 to 20 depending on the territory0, by construction
What you learnThe shape of the visibility areaRankings on 10 queries, in 12 places that matter

At near-identical cost, the second setup tracks three times as many keywords. That is where the real difference sits: an untracked keyword is a total blind spot, whereas an untracked location can often be inferred from its neighbours.

The grid keeps one genuine advantage: it reveals the shape of the visibility area, including where you would not have thought to look. That is a discovery expense, worth making once — at the start, or after a major change — not every month.

Building a sampling plan in 30 minutes

The method below works with any tool, including none at all: a map and a spreadsheet are enough to produce the list.

The five-step method

  1. Pull where your customers actually come from. Postcodes from the last ten quotes, the last hundred orders, or the customer file. It is the only step needing internal data, and it decides everything else.
  2. Rank those towns or neighbourhoods by volume. Keep the ones making up 80% of activity. On most accounts that is between four and ten areas.
  3. Add the addresses of your three to five direct competitors. These are the most informative locations in the setup: they measure the balance of power where it is actually contested.
  4. Add two to four target areas. Places you do not sell in yet but want to. Without them, the setup only measures what you already have.
  5. Place one point per area, at a precise address. A square, a junction, a shopping centre entrance — not the administrative centroid of the town, which sometimes lands in a field.

Budget twelve to twenty locations for a single site. Beyond that you add cost without adding decisions: two points four hundred yards apart in the same neighbourhood almost always say the same thing.

Put your tracking points where your market actually is

Pinperf is built on this principle: you place your locations yourself, to the street address, you name them, and every reading returns one trend line per location and per keyword. Not one measurement spent on an area with no customers.

Explore Pinperf

The sampling plan above takes about ten minutes to enter.

Edge cases

The mobile tradesperson. No catchment area around an address, but a service area. Locations go in the towns that generate revenue, not around the home address — which is often the worst possible place to measure, since nobody searches there.

A service area with no visible address. Some profiles hide their address and declare service areas. Tracking is unchanged: you measure from the declared towns, which also lets you verify that the declaration produces real visibility — it does not do so automatically.

The multi-location network. The rule changes scale: four to six locations per site, chosen the same way, plus two or three "boundary" points between neighbouring sites to detect cannibalisation. The subject is covered in the article on running a network.

How many points make a reliable diagnosis

A one-off diagnosis needs little: six to eight well-chosen locations are enough to know whether a profile is visible on its market. Tracking over time needs stability more than volume — twelve points read every month for a year beat forty read twice.

The top 3 share of voice formula does not depend on point count, which is what makes it comparable: measurements ranked 1 to 3, divided by total measurements, times 100. Forty-two top 3 appearances across a hundred and twenty keyword × location pairs is 35%. This formula and the other metrics are set out in the local SEO KPIs.

Stop paying for measurements in the middle of a field

Every Pinperf reading starts from the locations you named — and from no others. At equal budget you track three times as many keywords, and every trend line carries a name your client understands without explanation.

See plans and pricing

The five-step sampling plan above can be entered as it stands.

Twelve to twenty for a single site, four to six per site across a network. Beyond that you add cost without adding decisions: two points a few hundred yards apart in the same neighbourhood almost always return the same ranking.
No, they answer a discovery question: what is the real shape of my visibility area, including where I would not have thought to look. That is worth spending on at the start or after a major change, not every month.
It is one of the most informative locations there is: it tells you whether you exist where the competitor is strongest, and it reveals the real competitors at that exact point, who are not always the ones you assume.
Take the service queries customers actually type, the ones that trigger a call. Three to ten is enough. Tracking thirty variants from a single point tells you less than tracking five queries from twelve places.

Sources

Read next

All posts