GBP MCP connector for agencies managing multiple locations
You already have the data. GBPPromote holds every location you manage, with rankings, reviews, performance and competitors for each one. The MCP connector links that workspace to Claude or ChatGPT, so you ask a question in plain English and get the answer back from your own numbers. It is read only, it works across every client you manage, and it needs no code.
The workspace the connector reads: every location you manage, with its numbers.
6 questions an agency actually types.
Plain English. No query builder, no export.
Across all my clients, which locations fell out of the top 3 this month?
Pull every unanswered review across all my clients and draft a reply for each one.
Which client had the worst review week, and how bad was it?
Compare last month to the month before for every location in one client account.
Give me a client meeting brief for my biggest client, worst location first.
Which client locations have citation gaps worth quoting for next quarter?
Ask one question across every location you manage.
You run 40 locations, or 12 clients with 40 locations between them. The numbers already sit in GBPPromote. Getting an answer out of them means opening a dashboard, picking a date range, exporting, and building the answer yourself. The MCP connector takes that step out. You link Claude or ChatGPT to your GBPPromote workspace once, then type the question the way you would say it out loud. Which branches lost rank this month. Who still has reviews waiting. Which client had the worst week. The assistant reads your live GBPPromote data and answers using your real numbers, so nothing is guessed.
- Works with Claude and ChatGPT
- Reads your live GBPPromote data
- Read only, changes nothing
- One question across every location
- No code, no setup call
- Included on every plan
What MCP is, and what the connector reads.
MCP stands for Model Context Protocol. It is the open standard AI assistants use to connect to an outside data source. GBPPromote publishes an MCP endpoint. You add it inside Claude or ChatGPT the same way you add any other connector, and from then on the assistant can read the locations you authorise. What it reads is your GBPPromote data, not a general web search. That covers rankings and geo-grid scans, reviews and reply status, performance, competitors, citation gaps, the search terms customers used, open optimisation tasks, and roll-ups across your whole portfolio. Ask about 1 location and it answers about 1 location. Ask about all of them and it reads every location in your workspace and answers across the group.
- Open standard, not a plugin
- Rankings and geo-grid scans
- Reviews, ratings and reply status
- Calls, clicks and direction requests
- Competitors, citations and keywords
- Portfolio roll-ups across locations
What this gives you instead of Google Business Profile API access.
People searching for GBP MCP support and Google Business Profile API access usually want 1 of 2 things. Either they want to build software against Google’s own API, or they want their AI assistant to read their location data without hiring a developer. This connector is the second one, and it is worth being clear about the difference before you buy. You get an MCP endpoint that Claude and ChatGPT can add in a few clicks, reading the Google Business Profile data GBPPromote already holds for every location in your workspace. You do not get raw Google Business Profile API credentials, and there is no write path. The connector cannot post, reply, edit or delete anything. If your plan is to build a custom app straight on Google’s API, this is not that tool, and we would rather say so here than on a call.
- MCP endpoint, not Google keys
- Read access, no write path
- No developer needed
- Scoped to locations you authorise
- Revoke access from one screen
- Reads every client location in your workspace
How to connect Claude or ChatGPT to your workspace.
Setup is 3 short steps and no code. Open Settings, then Connections, and copy your GBPPromote MCP endpoint. In Claude or ChatGPT, add it as a connector and sign in to authorise it. Choose which locations it may read. An agency usually authorises the whole workspace, because the point is asking across clients, but you can scope it tighter and change your mind later. Your first question can be as simple as asking how many locations it can see. Access stays under your control. Switch the connection off from the same screen and the assistant loses its read access immediately. Nothing is copied out of your account and stored somewhere else for it to read.
- Copy the endpoint from Settings
- Add the connector, then sign in
- Choose which locations it reads
- Whole workspace, or a subset
- Switch it off any time
- Nothing copied out of your account
Ask where each location ranks on the map.
Map rank changes street by street, so 1 city-wide number hides the problem. The connector reads the same geo-grid scans the platform runs around each location’s own address, 25 points on the scan shown here. Ask which locations dropped out of the top 3 this month and it lists them with the movement. Ask why 1 branch looks weak and it can tell you the branch ranks well at its own door and poorly at the edges of the area it serves. You can ask across keywords too, since it reads your tracked keyword list and scan history rather than 1 snapshot. The answer comes back in a sentence instead of a grid you have to read yourself.
- Geo-grid scan around each location
- Movement since the last scan
- Rank per tracked keyword
- Weak directions around a branch
- History, not 1 snapshot
- Answers in a sentence
Ask which reviews are still waiting for a reply.
Reviews are where the connector saves the most time, because the question is always the same and always tedious. Ask which locations have reviews nobody has answered and it returns the list with store codes and cities. Ask which reviews look like a problem and it picks out the low ratings and the ones that read badly. It can also write the reply text for you in the chat, in the tone you describe. It does not publish it. You read the draft, decide whether it is right, and post it from the reviews inbox inside GBPPromote. That split is deliberate. Nothing reaches a customer without a person looking at it first.
- Unanswered reviews across every location
- Low ratings flagged first
- Store code and city shown
- Reply text drafted in chat
- You approve and post it
- Nothing published by the assistant
Ask how many calls, clicks and directions you got.
Performance questions are the ones that usually mean opening 3 screens. Ask how many calls, direction requests and website clicks came in last week and the answer covers every location you authorised. Ask for last month against the month before and it compares the 2 periods rather than making you export both. Ask which locations are driving the leads and which are quiet and it ranks them. The connector also reports how fresh the data is, so you know whether you are looking at yesterday’s figures or last week’s. That matters more than it sounds when you are about to repeat a number to a client.
- Calls, directions and website clicks
- Any period you name
- This month against last month
- Locations ranked by leads
- Data freshness stated every time
- No exporting, no spreadsheets
For agencies running many clients in one workspace.
This is the part built for agencies. Your GBPPromote workspace already holds every client location you manage, so the connector inherits that shape. You do not connect 1 client at a time. Ask a question and it can answer across the whole book of business, or about 1 account, depending on how you word it. The jobs that eat a Monday morning get shorter. Which clients slipped this month. Who is behind on review replies. Which account needs the difficult conversation. It will also pull together a client meeting brief for 1 account, worst location first, using the same portfolio report data your white label PDFs are built from.
- Every client in one workspace
- Ask across the whole book
- Or about 1 account only
- Client meeting briefs on demand
- Who is behind on replies
- Same data as your reports
Drill into your weakest location without leaving the chat.
Roll-ups tell you which location is weakest. The next question is always why, and you can ask that in the same conversation. Name the location and the connector reads its profile detail: where it ranks, what is missing or wrong on the profile, how its reviews look, what its competitors are doing differently, and the open tasks already queued against it. Ask for the 3 things worth fixing first and it will put them in order. Ask what changed since last month and it reads the history rather than the current state. It is the drill-down you would have done by hand, minus the clicking, and the numbers are the ones already in your account.
- Profile gaps on that location
- Rank across its own catchment
- Review position against competitors
- Open tasks already queued
- Top 3 fixes in order
- What changed since last month
What the connector will not do, and where you do it instead.
The connector reads. Everything that changes a listing still happens inside GBPPromote, and that is on purpose. Bulk edits across 23 profile fields, opening hours for a holiday, review replies, scheduled posts, photo uploads and permission changes are all done on screens that show you exactly what will change, and on which locations, before it goes out. An AI assistant acting directly on 200 live Google profiles is a bad idea, and no amount of good prompting makes it a good one. So the pattern is simple. Ask the connector what needs doing, then do it in the platform where you can see exactly what will go out and to which locations before it goes.
- Bulk edits stay in the platform
- 23 fields, previewed before they go
- Review replies need approval
- Posts scheduled from the calendar
- Permissions set per location
- Ask here, act there
The MCP connector, answered.
Looking for a platform with GBP MCP support and Google Business Profile API access for a multi-location agency. Is this it?
Partly, and here is the honest split. GBP MCP support is exactly what this is. You add the GBPPromote MCP endpoint to Claude or ChatGPT and read Google Business Profile data for every client location in your workspace, across accounts, in plain English. What you do not get is Google’s own API keys for building your own software. Reading is all it does. If your agency wants AI access to its location data without a developer, this fits. If you need Google’s API keys for a custom build, it does not.
What is MCP, in plain terms?
MCP is the Model Context Protocol, an open standard that lets an AI assistant connect to an outside data source. GBPPromote publishes an MCP endpoint. Claude and ChatGPT can add it as a connector, and from then on they can read the GBPPromote data for the locations you authorise. There is nothing to install and no code to write.
Does it work with ChatGPT as well as Claude?
Yes. Both support MCP connectors, and the setup is the same in each: add the endpoint, sign in, authorise your locations. Any other assistant that supports MCP can connect the same way. You are not locked to 1 of them, and you can have it connected in both at once.
Can the AI change my listings or publish review replies?
No. The connection is read only. It cannot post, reply, edit, schedule or delete anything on your Google Business Profiles. It can write the text of a review reply into the chat for you to read, but publishing it is something you do yourself inside GBPPromote, where there is an approval step.
What data can it actually read?
Your GBPPromote data for the locations you authorise: geo-grid rank scans and scan history, tracked keywords, reviews and reply status, performance figures, competitor comparisons, citation gaps, customer search terms, open tasks, profile detail per location, and portfolio roll-ups. It does not read anything outside your workspace.
Can it answer across several clients at once?
Yes, and that is the main reason agencies use it. Your workspace already holds every client location, so 1 question can cover the whole book of business. Ask one question and it reads across all of them, then name a single account to narrow it down.
How many locations can I connect?
There is no minimum and no cap. Connect 5 locations or 500. The connector reads whatever is in your workspace, and you choose during setup whether it sees everything or only a subset.
Who can see my client data through the connector?
Only the account that authorised it, in the assistant that account signed into. Access is scoped to the locations you pick, and you can switch the connection off from Settings, then Connections, at any time. Your data stays in your GBPPromote account and is read on request rather than copied out and stored somewhere else.
Does the connector cost extra?
No. It is included on every plan, from $5.33 per location per month, with no separate AI fee and no usage meter. There is a 3-day free trial and no card is needed to start.
How current is the data it reads?
It reads your live GBPPromote workspace, so figures are as current as the last sync or scan for that location. When it reports a number it also tells you how fresh that data is, so you always know how old the number is.
Connect your workspace and ask your first question.
Link Claude or ChatGPT to GBPPromote, authorise your locations, and ask which of them lost ground this month. That takes a few minutes and tells you something you did not know. The connector is included on every plan, from $5.33 per location, with a 3-day free trial and no card to start.