Use case · Reporting & analysis
Connect Claude, ChatGPT or your own reporting pipeline to Grades.com through the MCP server or the REST API. Your agent reads grades, review velocity, ratings by site, reply rates, competitors and AI answers for every location, and writes the monthly report, the location comparison and the notes for the next call.
monthly_report61 new reviews in October, 4.7 average; Google 38, Yelp 14, Facebook 9.competitors_compareYou lead on velocity and reply rate; Sudsy Express leads on rating.review_insightsComplaints about the vacuum stations doubled since September.Report: 6 numbers, 3 wins, 2 things to fix, ready for the client
The report writes its first draftYour agent pulls the month's numbers and comparisons; you spend the time on what they mean.
Every location, one connectionHeadquarters reads the whole group, and each location still has its own numbers.
Numbers you can traceCounts come from stored reviews, dated AI answers and monthly competitor snapshots.
What your agent can do here
Each one is a sentence typed into Claude, ChatGPT or Cursor. The agent calls the Grades.com tools named under it and answers from your data.
"Grade, review velocity, rating by site, reply rate, versus last month and versus the four competitors AI names next to us." The monthly_report prompt loads the month through reviews_summary_period and review_stats, then competitors_compare for where you lead and trail. You edit the story, not the spreadsheet.
"Rank our 12 locations by grade and 90-day velocity. What does the top one do that the bottom one does not? Which complaints are specific to one location?" locations_compare and locations_summary give the numbers; review_insights, read for the whole group, adds each location's grade and its most used review tags. The multi_location_report prompt runs it in one go.
"Anything move among our competitors this month?" alerts_list returns the alerts Grades.com raised: a new name in your top five, a competitor passing you on rating, review velocity or grade. competitors_list shows who is new this month and who dropped out.
"How often does each assistant name us for our category in our city? Who else, and which sites do they cite?" ai_perception returns the dated answers of ChatGPT, Gemini, Perplexity and Claude with the sites they cite; competitors_compare adds which competitor they name most.
"Put every location's October reviews in a Google Sheet and keep it in sync." reviews_export_sheets writes a sheet in your own Drive that syncs hourly; reviews_export_csv gives a file. Dashboards and pipelines read the same data from the REST API.
Connect it once
Add the MCP server to Claude, ChatGPT or Cursor and sign in, or give your reporting pipeline a read-only API key. One sign-in at headquarters reads every location you have access to. No AI client? Ask Grades, the assistant in your dashboard, answers the same questions.
MCP server https://mcp.grades.com/mcp
REST API https://app.grades.com/api/v1
Scopes reviews:read, companies:read; companies:write only for a synced sheet
ChatGPT custom connectors need a Business, Enterprise or Edu plan
The numbers
Volume, rating, sentiment, reply rate and themes for any period, against the previous one and the median of your category. Read from stored reviews; no model writes the numbers.
Insights →Every month ChatGPT and Perplexity are asked for alternatives to you in your city. The names are matched to real businesses and ranked against you on grade, rating and velocity.
Competitor Map →Four assistants asked the questions a customer would, with your location set. Mentions, sentiment, the ratings they quote checked against your reviews, and the sites they cite, dated.
AI Reputation →Every review is kept, including ones a site takes down. Trends leave removed reviews out, and the archive still shows when and where each one went.
Review Archive →What we can and cannot do. Nobody controls what an assistant says, and we never promise to change an answer. We report what it says, where it got it, and the day it changes.
Many locations
Headquarters sees the group against the previous period, then every location in one table it can sort by any column. Each location keeps its own dashboard.
Multi-location →Admins get one digest a week. For a group it opens with every location at a glance, then a section per location.
Every comparison here is per location. Grades.com does not rank or name individual people, in reports or anywhere else.
What the reporting team gets
Each one runs per location, in the dashboard and through the MCP server and the API.
| Deliverable | Product | MCP tools | Plans |
|---|---|---|---|
| Velocity, rating, reply rate and themes against the previous period | Insights | review_insights, review_stats, reviews_summary_period | Every plan: Google on Starter, every site from Pro |
| The businesses AI names next to you, ranked against you, with alerts when one moves | Competitor Map | competitors_list, competitors_compare, alerts_list | A preview on Pro, the full map and alerts on Premium |
| Dated answers from four assistants, with cited sites and rating checks | AI Reputation | ai_perception, ai_probe_run | Every plan: monthly on Starter, weekly on Pro, daily on Premium |
| Every review kept, including ones a site takes down, with exports | Review Archive | review_list, review_history, reviews_export_csv | Every plan |
| Every location compared, and each one's rank among its own competitors | Multi-location | locations_summary, locations_compare, locations_competitor_ranks | Every plan; group alerts on Premium |
| The REST API, webhooks and a synced Google Sheet for your own dashboards | Developers | reviews_export_sheets | API on every plan; signed webhooks from Pro |
They are counted from the reviews we collect from each site you connect, so they follow the sites closely. A site can hide or take down a review: we keep it in the archive and leave it out of the trend, so the difference is visible, not silent.
Yes, that is the only way it works. Every comparison is per location: grade, velocity, rating, reply rate and themes. Grades.com does not rank or name individual people.
Yes. Your agent reads it through the MCP server, your pipeline through the REST API and webhooks, and reviews export to CSV or to a Google Sheet that keeps itself in sync.
No, and nobody honestly can. We report what each assistant says, which sites it cites, which competitors it names next to you, and the day an answer changes.
14-day free trial of Pro, no card needed.
Product names and logos are trademarks of their owners; no endorsement implied.