Grid rank tracking: why one ranking number is misleading

Local position is calculated partly from where the searcher is standing. So there is no one number to report. There is a surface, and its shape is the actual information.

An operator recognizes this pattern immediately. It is the difference between the daily sales figure and the sales by daypart. One tells you the week was fine. The other tells you Tuesday lunch is bleeding and nobody noticed.

What is grid rank tracking, exactly?

It is the same search, repeated from a grid of coordinates, with the result recorded at each point.

A tool places a set of points across your area, often a square arrangement of nine, forty nine, or more. At every point it asks Google the same question from that location and writes down where your business appeared. The output is a map with a position at every node, usually colored so the strong areas and the dead areas are obvious at a glance.

What you get is a picture of your visibility, not a score. You can see the radius, the direction it leans, and the streets where you disappear.

That picture is what makes decisions possible. A number cannot tell you that you own the east side and are invisible west of the freeway. A grid tells you that in three seconds.

Why is location such a big part of this?

Because two of the four strongest local ranking factors are about geography, and one of them is about the searcher rather than about you.

In Whitespark's 2026 Local Search Ranking Factors report, where 47 local search experts scored 187 factors, proximity of your address to the point of search ranks second at 225 points, behind only primary category at 227. Having a physical address in the city of search sits fourth at 213. Proper placement of the map pin is tenth at 165. Proximity to the city centroid, the older and weaker version of the same idea, is twelfth at 157.

Read the difference between the second and the twelfth entries carefully, because it is the whole argument for grids. Proximity to the searcher outscores proximity to the city centre by a wide margin. The searcher moves. A single measurement taken from one fixed coordinate is measuring the factor that matters less.

Why is one ranking number misleading?

Because the point it was measured from is almost always flattering.

Most single-number reports measure from the business address or the center of the city. Both of those are the places you are most likely to look good, because you are as close to yourself as it is possible to be.

So the report says position two. Your customer, four miles out, sees you nowhere. Both are true. Only one of them describes the business you are trying to grow.

It gets worse over time. A single number is stable enough to look like progress even when your visible area is shrinking. You can hold position one at your own address while losing three miles of radius, and the report will show a flat green line all year.

If your monthly report is a single ranking figure, ask what point it was measured from. If nobody can tell you, the number is decoration.

What does a grid show you that a number cannot?

Four things, all of which change what you do next.

Radius. How far your visibility actually reaches. This is the number that should drive your ad spend boundaries.

Direction. Visibility is rarely a circle. It leans toward or away from competitors, highways, and water. If you are strong in one direction only, you know where new customers can realistically come from.

Competitor overlap. Better tools record who ranked above you at each point. That tells you which competitor owns which neighborhood, which is far more useful than knowing who is generally strong.

Where a change landed. After you fix a category or start collecting reviews, the grid shows you which parts of the map moved. A single number cannot separate a real gain from ordinary noise.

What does a grid look like when it actually moves?

Here is the arithmetic. Run it with your own grid.

Month one, take the baseline. A seven by seven grid is 49 points. Say you appear in the top three at 12 of the 49, your average position across all 49 points is 8.4, and the single point sitting on your front door reads position 2.

Write down the second series at the same time. Pull the trailing 90 days from the Business Profile performance report: say 268 calls, 194 website clicks and 78 direction requests, 540 actions.

Month four, run the identical grid. Same size, same centre, same term, same tool. Now you are in the top three at 21 of 49, average position 6.1, and your front door point still reads position 2.

Read what happened. The number your old report published did not move at all. The grid says you gained nine points of top-three coverage, close to doubling it. Actions over the same window went from 540 to, say, 690, an increase of 150.

Now the discipline that keeps this honest. You have two series that moved in the same direction over the same quarter. Report them as two series that moved together. Do not divide 150 by 9 and publish a value per grid point. You changed a category, three competitors did their own work, and a season happened. Correlation, stated as correlation, is a strong report. A made-up conversion formula is a liability.

Can I turn a grid position into a traffic number?

No, and this is worth saying out loud to anyone who tries to sell you one.

Google exposes no local pack position dimension anywhere. The Business Profile performance report publishes interactions, search terms, views, direction requests, calls, website clicks, messages, bookings, products, menus and offers. There is no rank, no position, and no pairing of impressions to clicks by position. Search Console has no local pack dimension either. That means no vendor, agency, or tool can produce a genuine click-through rate by local pack position, because the raw material does not exist.

The one traceable attempt is a BrightLocal study from October 2018 in which 5,500 Mechanical Turk workers looked at five static San Francisco search results, desktop only, all service-area categories. It found local pack position one taking 16.1% of clicks and the whole three-result pack taking 28.8% with ads present. Every caveat there is load-bearing: paid microworkers clicking a screenshot, five queries, one city, and a results page that has since absorbed local pack ads, Local Services Ads and AI local packs. The page that study measured no longer exists.

Two more reasons to distrust the exercise. The 2026 survey scores click-through rate from the local pack as a ranking input, thirteenth at 156 points, which means clicks and position feed each other rather than sitting in a clean one-way formula. And Sterling Sky has documented that Google retired several older profile metrics and redefined Views as a deduplicated unique visitor count, so even old published conversion benchmarks cannot be compared against a current account.

The sentence to keep: nobody, including Google, publishes what a first position in the local pack is worth in clicks. Measure your own actions, before and after, and report the grid alongside them.

How big should the grid be and how often should I run it?

Size it to the area you can actually serve, and run it monthly.

The common mistake is setting a grid the size of the whole metro because a big map feels impressive. A big grid full of red squares is not a problem list, it is a fantasy. If you would not drive there, do not measure there.

Set the grid to the area you serve today plus a modest ring beyond it, so you can see whether the edge is moving. Then leave the settings alone. Changing the grid size between runs makes the comparison meaningless.

Monthly is the right rhythm. Local results move slowly and the daily variation will drive you to conclusions you should not draw. If a run looks alarming, re-run it a week later before acting.

One thing the grid will increasingly miss. Sterling Sky's State of Local SEO 2026 reports AI-powered local packs on about 7% of tracked keywords, showing one or two businesses instead of three and carrying no call buttons. Across 322 markets studied with Places Scout, 88% surfaced fewer unique businesses in the AI pack than in the ordinary three-pack, with 5,943 unique businesses appearing against 18,330. Your grid measures the three-pack. A shorter list is being drawn beside it, and the tools have not caught up.

One caveat on the other side: if your grid suddenly goes dark everywhere, do not start optimizing. Check whether the profile is still live, because a suspension looks exactly like a catastrophic ranking drop.

What to do this week

Run one grid on your single most valuable search term. Set the area to what you actually serve. Save the image, the settings, and the date.

Write down three numbers from it: how many points put you in the top three, your average position across all points, and what the centre point reads. Then write down your trailing 90 day calls, website clicks and direction requests.

Look at the shape before you look at the numbers. Where is the edge? Which direction does it lean? Is there a neighborhood you assumed you owned that is actually red?

Note which competitor appears at your weakest points. Then check what they are doing that you are not, starting with reviews, since benchmarking your review cadence against theirs is the fastest comparison available.

Also check the boring stuff before you conclude anything. Inconsistent details across the web can hold you back quietly, which is what NAP consistency is and how much of it still matters.

Questions about what your grid is telling you?

Email me at eric@seod.com with a screenshot of your grid, or just your business name and main term if you do not have one. I will read the shape and tell you which neighborhoods you are realistically able to win and which ones you should stop paying attention to.

I answer these myself. If your map looks healthy and the constraint is that people are not calling, I will tell you to go look at the phone instead.

Running a grid yourself once a quarter is genuinely useful. If you want the broader search picture tracked and acted on week to week, Search visibility reports keyword position and calls with the next priority clearly identified. Or keep reading the rest of the local search library.