How to Track Google Maps Rankings for Clients

Tracking Google Maps rankings for a client is mostly decided in the first week, before a single scan runs. Four choices set everything that follows: which keywords you agree to be judged on, how far out from the pin you measure, when you take the baseline, and how often you rescan.

Get those four right and the monthly report almost writes itself. Get them wrong and you spend the retainer arguing about numbers instead of doing work.

The order is fixed. Agree on a keyword list with the client in writing, size the measurement area to the ground they actually cover, run a clean baseline before anyone touches the profile or the site, then pick a cadence you can hold for a year.

This article walks that sequence with one running example: an agency taking on a garage door repair company. How the grid itself works is covered in geo grid rank tracking, and what it costs across a book of clients belongs in agency local rank tracking.

The onboarding sequence, start to finish

Every step below exists because skipping it creates an argument three months later. The whole thing takes an afternoon of client time and about an hour of yours.

1. Pull the keyword list out of what the client actually invoices, then get it approved in writing.
2. Map the real trade area from job history, not from the client’s ambitions.
3. Set the tracking configuration and freeze it, in a document you can reopen in month nine.
4. Run the baseline scan before any optimization work, including the easy fixes.
5. Record the profile state on baseline day so you can prove what changed.
6. Set the cadence in the calendar so it does not depend on someone remembering.
7. Walk the client through the first heatmap live, and say out loud what will not improve.

Step one: track what the client sells, not what they call themselves

Ask a garage door repair owner what they want to rank for and you will hear “garage doors”. That term is dominated by manufacturers, big box retailers, and people browsing styles for a house they have not bought yet.

It is a vanity term. It looks impressive on a report and it does not ring the phone.

What that business actually gets paid for is narrower and uglier: broken spring replacement, opener repair, off track doors, cable snapped on a Sunday morning. Those are the terms worth measuring, because a green square on one of them means a dispatched van.

Getting the list out of a client who says “rank me for everything”

Do not brainstorm. Ask for the last 60 or so invoices and sort them by job type. Then ask three questions: which job type is the most profitable, which do you have capacity for tomorrow, and which jobs do you turn away?

The answers usually reshuffle the list. A garage door company may make more margin on opener installs than on spring calls, in which case the install term deserves a slot even though the emergency term has more volume.

Use keyword research to sanity check volume and to catch the phrasing customers use that the owner never says. The invoice list decides the shortlist, not the volume column. There is more on that in local keyword research.

Terms worth refusing on day one

• The company’s own name. They already rank first for it. Tracking it inflates every average and proves nothing.
• Services they do not fulfill. If they do not touch commercial roll up doors, a commercial term is a lead they will resent.
• Single word head terms. “Garage” is not a query with commercial intent behind it.
• Duplicate near me variants. A grid already runs the query from many locations, which is exactly what “near me” signals. Tracking both the plain term and the near me version usually doubles your scan volume for near identical output. Pick the one the client says out loud and move on.
• City modifiers when the grid already covers the city. Same problem. The grid supplies the geography.

Five to 10 terms is plenty for a single location. Every extra keyword multiplies against every grid point on every scan, so a loose list is the fastest way to make tracking expensive.

Send the final list in an email, get a reply that says yes, and save it in the client folder. That email ends more disputes than any dashboard.

Step two: measure the trade area, not a circle around the office

The grid should cover the ground the client will actually drive to, and nothing further. Ask for job addresses by zip code for the last year, and ask where the vans genuinely go on a Tuesday at 4pm.

Trade areas are rarely circles. A garage door company on a highway corridor often serves 20 minutes north along the freeway and refuses a job eight minutes away across a river with one bridge.

Then apply the rule that saves you from yourself: every point you include becomes a point the client will judge you on. Add the far suburb where they have taken two jobs in three years and you have signed up for a permanently red corner in every report.

Leave it out, or put it in a separate tracking set you review quarterly, rather than one that sits in the client’s monthly headline.

Spacing and dimensions are a mechanics question, covered in the grid article. The only onboarding decision that matters here is honesty about coverage.

If the client is a service area business with no storefront, center on the address they verified and let the grid extend over the service radius. A service area rank tracker handles that shape.

Tell the client this plainly: the leaked API documentation makes the service area list a display and geographic consistency setting, not a ranking signal, so widening it will not widen the green.

Step three: freeze the configuration and write it down

Record these in a single document before the first scan: pin coordinates, the exact keyword strings including spelling and word order, grid dimensions and spacing, language and device settings, cadence, and who receives the report. Then leave them alone.

This is not bureaucracy. A seven by seven grid at one mile spacing and a seven by seven grid at two miles are different questions with different answers. A client comparing month one to month six is then comparing two instruments, not two results.

The most common way agencies destroy their own evidence is quietly resizing a grid mid engagement because the numbers looked better that way.

If a genuine change is needed, for example the client opens a second shop, do not edit the existing set. Start a new one and keep the original running alongside for at least a quarter so the history stays readable.

Step four: baseline before the easy wins, not after

The most expensive mistake in local SEO onboarding is doing an hour of obvious profile cleanup and then running the first scan. That hour is often the single biggest improvement of the engagement, and you just gave the credit away.

Baseline on the day access is granted, before you fix the primary category, before the photo upload, before the review request campaign.

The primary category outweighs every other profile field, so changing it before the baseline erases the clearest before and after comparison you are likely to get all year. The same goes for a name or address correction.

On baseline day, also record everything you might later claim credit for: primary and secondary categories, hours, services listed, description, review count and average, website URL, and which competitors hold the pack at three or four representative points. A quick pass with a profile audit covers most of it.

One more thing to check: Google now writes some profile fields itself, generating service lists and descriptions on the knowledge panel, and entries have turned up that no owner put there. Note what is on the profile rather than assuming the client put it there.

What the baseline can honestly promise

Before the first heatmap goes on screen, the client needs to understand one number. Around 55 percent of a pack ranking is just the gap between the searcher and the business, and Google resolves that gap before it weighs any other signal.

Nothing you sell in a retainer changes it. It is the reason the map is red at the edges, and that is not a failure of the work.

Under proximity, the heaviest factors are the profile itself and then reviews, with page content behind them and several lighter signals below that. Those shares are approximate and never add up to a complete accounting, so the full taxonomy belongs in local SEO ranking factors, not in a kickoff deck.

What the client needs from the call is simpler: a garage door company with a thin review file and half its profile fields left empty has more headroom than a rival who filled everything in years ago.

Why you rank in some areas and not others covers the geography side.

Say the proximity sentence in the kickoff call, not in month four when the client asks why the north end never went green. It reads as context on day one and as an excuse later.

Choosing a cadence you can defend

Cadence is a budget decision dressed as a technical one. Scans cost money and attention, and past a point more frequent scanning does not produce more insight, it produces more chances to overreact. Pick from the situation, write it in the calendar, and stop revisiting it.

Teach the client the interpretation rule at the same time you agree on the cadence: one scan is a data point, three or four in the same direction is a trend. Agreeing this before the first bad week is the difference between a calm call and a panicked one.

Walking the client through their first heatmap

Do this live, screen shared, not as an attachment. It takes six minutes and it sets the vocabulary for the whole engagement. A rough script:

1. Explain what you are looking at. This is not one ranking. It is dozens of separate searches run from dozens of points around your service area, each one showing where you came in the pack from that spot.
2. Defuse the green. The green near the shop is normal. Being visible on your own street is not an achievement, it is proximity doing its job.
3. Point at the boundary, not the colors. The useful line is where you fall out of the top three. That boundary is the thing that will move, and moving it half a mile in the right direction is worth more than any average.
4. Name who is holding the corners. Say the competitor’s name and their review count. It makes the map concrete and it makes the review conversation easier later.
5. State what will not change. The far edge is a proximity problem. We are aiming to push the boundary out, not to turn the whole map green.

Three things not to say: do not promise all green, do not quote the grid average as if it were a ranking position, and do not say “number one on Google” about anything.

How to read a local SEO heatmap is a good link to send afterward for clients who want to study it.

Warn them about their own phone too. Personalization carries a small but real share of pack results, and the client searching from their kitchen will see something different from the report.

Tell them on day one so the difference reads as expected rather than as a broken tool. How to check Google Maps rankings explains why a manual check disagrees.

The numbers to write down at baseline

Record three things, not one:

• Share of grid points in the top three. The closest proxy for pack visibility that exists.
• Share of points where the business appears at all. This catches the difference between ranking badly and being invisible, which are separate problems with separate fixes.
• Where the boundary sits in two or three directions. Written as distance, for example strong to about two miles north and about one mile east.

A single blended grid score is easy to put in a report and easy to inflate by shrinking the grid, which is exactly why you froze the configuration in step three.

Two grids with an identical average can have completely different shapes, and the shape is what tells you what to do next. Heatmap versus rank tracker covers what each format is genuinely good for.

Attribution: everything else that moved

Ranking changes are never cleanly yours. Keep a dated log next to the scan history, and note anything that could explain a shift:

• Seasonality. Garage door demand spikes after the first hard freeze and after high wind events. Volume and behavior change, and the pack can shuffle with them.
• Competitor moves. A rival collecting reviews faster, or opening a second location closer to the corner you were trying to win.
• Client side edits. Owners change hours, add services, or accept a suggested edit without telling anyone.
• Result layout changes. An AI Overview landing on top of a query changes the page without changing the pack, and practitioners watching those queries have reported half their visibility or worse disappearing. A pack position that held steady while clicks fell is a layout story, not a ranking story.

Separating rank movement from traffic claims is the discipline that matters most here. If you do not have analytics proof that calls went up, say the rankings improved and stop the sentence there.

When month two is worse than the baseline

It will happen. Work the checks in order before touching strategy:

1. Is the pin still where it was, and is the profile still verified and unsuspended.
2. Did the primary category change, by the client or otherwise.
3. Did the review profile take a hit, or did a competitor gain a stack of reviews.
4. Is the drop across the whole grid or one corner. Whole grid points at the profile or an algorithm change. One corner almost always points at a specific competitor or at proximity.
5. Only then reconsider the plan.

Then call the client before they open the report, not after. Send the bad month in exactly the same format as the good months, with the same three numbers.

Agencies that change their reporting layout when the news is bad get noticed, and it costs more trust than the dip did.

Onboarding mistakes that keep recurring

• Baselining after the quick wins, and losing the best evidence of the year.
• Letting the client pick keywords from their own marketing vocabulary instead of their invoices.
• Drawing the grid around aspiration, then explaining a red corner every month for a year.
• Tracking 30 keywords because the plan allows 30.
• Resizing or recentering the grid mid engagement, which silently breaks the comparison.
• Reporting one blended number that nobody can act on.
• Sending screenshots with no date on them.

Where GMBTool fits

GMBTool runs the scans, keeps the history tied to a fixed configuration so month one and month nine stay comparable, and produces the client facing report and share link. White label output is there when the report needs your own branding.

The local rank tracker and map pack rank checker handle the measurement. Local SEO reports plus the portfolio report handle the delivery side.

The trial lasts three days, asks for no card, and starts at app.gbppromote.com/register if you want to run a baseline for a prospect before you pitch.

The honest limit is that a tracker measures the boundary. It does not push it. The far edge of a large trade area belongs to whoever is physically nearer, and that single fact is roughly 55 percent of the pack result.

So a corner that stays red out there is reporting geography rather than failure. Any vendor implying otherwise is handing you a promise the client will hold you to. Proof of the boundary moving is what tracking actually buys.

For the reporting layer itself, see local SEO client reports and white label SEO reports.

Frequently asked questions

How many keywords should I track for a new single location client?

Five to 10, tied to specific services on their invoice list. Cost scales with keywords multiplied by grid points multiplied by scan frequency, so a 30 term list on a dense grid gets expensive fast without telling you more than the 10 term list did.

If the client insists on more, add them as a separate low frequency set rather than inflating the main report.

What if the client already has tracking from a previous agency?

Treat it as context, not as your baseline. Unless the previous configuration is documented down to grid spacing and pin coordinates, you cannot compare your scans to theirs, and any trend line spanning both is fiction.

Run your own baseline on day one, keep their exported history in the folder, and say plainly in the kickoff that the clock starts now.

Can I track a client with no storefront?

Yes. Center on the verified address and extend the grid over the area they serve.

Be careful with expectations. The service area list reads as a description of where you work, not a lever that ranks you there, so adding 20 more towns does not create rankings in those towns.

How long before the baseline is worth comparing against?

60 to 90 days, or three to four scans in the same configuration. One follow up scan against a baseline is two data points, and a straight line through them means very little. Normal pack movement alone can produce a flattering or alarming result by accident.

Set that expectation at kickoff, in the same conversation where you agree on the cadence.

Should the client get the raw grid or a summary?

Both, in that order of importance. The heatmap is the evidence and it should be one click away, but the first thing they read should be two or three sentences saying what moved, what did not, and what you are doing next.

Clients who get raw data with no narrative either ignore it or misread it, and misreading usually shows up as a request to change strategy over a single noisy week.

Should I show competitors in the client report?

Yes, named, with their review counts. It turns an abstract map into a specific race and it makes the review conversation land, since reviews carry a large share of what you can actually influence.

It also protects you when a competitor’s growth explains a dip better than your work does. Keep it factual and avoid commentary about anyone else’s tactics.