Build a lead dashboard with Claude Code Scrape Google Maps, export, then feed your AI

Data comes first, then the dashboard

A raw list of businesses is not a dashboard. It is a spreadsheet you scroll and forget. The value shows up when you can sort by rating, filter by category, spot the leads with no website, and see where your best prospects cluster. That is what a dashboard does, and modern AI tools can build one for you from a single export.

The workflow has two halves. First you need the data: real businesses with names, phones, ratings, categories, and websites. The free Vonsel Chrome extension captures that straight from Google Maps and exports it in the format your AI tool prefers. Second you feed that file to an AI: Claude Code to write and run a real dashboard, or ChatGPT to chart it in seconds without touching code.

The extension gets you the data, and the dashboard is what you actually use. Get the export right and the AI half is fast.

Export a clean file from the extension

Open Google Maps, run a search like "dentists in Austin", and let the Vonsel pop-up capture the results. When the list is ready, export it. The extension writes CSV, Excel (XLSX), and JSON, so you match the format to the tool you plan to use next. For a step-by-step on the capture itself, see export Google Maps data to CSV.

Format matters more than people expect. A clean export that follows the standard CSV structure means the AI spends its effort on analysis, not on guessing your columns.

Rule of thumb: reach for CSV when you want a fast answer inside ChatGPT, and reach for JSON when Claude Code will build something that reads each business as an object. More on the trade-off in JSON vs CSV for AI analysis.

Build the dashboard with Claude Code

Claude Code runs in your terminal, reads files in a folder, and writes and executes code for you. That makes it a natural fit for a lead dashboard, because the whole thing is: read the export, generate an HTML page with charts and filters, and open it in the browser. You describe the outcome in plain English and review what it builds.

Drop your export into a project folder and start with a prompt like this:

"Read leads.json in this folder. Build a single-page HTML dashboard with a table of businesses, a bar chart of leads by category, a filter for rating above 4.0, and a highlight for businesses with no website. Then open it in my browser."

Claude Code will parse the file, write the page, and run it. From there you keep going in the same conversation: "add a column that flags businesses with fewer than 20 reviews", or "sort the table by rating descending and color the no-website rows". Each request is a plain sentence, and the tool edits the code and reloads. This is exactly the loop covered in analyze leads with Claude Code.

Because Claude Code writes real files, the dashboard is yours to keep. You can refresh it next week with a new export, and ask for a map view that plots each business by latitude and longitude, which is walked through in map visualizations from lead JSON with Claude Code.

Or ask ChatGPT to chart it for you

If you do not want a terminal, ChatGPT is the fastest route. Upload the CSV or Excel file into a chat and ask questions in normal language. Its data analysis feature reads the file, runs the math, and returns tables and charts inline, with no code from you.

Prompts that work well on a Google Maps export:

ChatGPT and Claude Code cover most people. If you live in a different stack, the same file works elsewhere: OpenAI Codex can wire the export into a small app, and a local model through a tool like Ollama or LM Studio keeps the data on your machine when the list is sensitive. The export is portable, so you pick the tool and you are not locked in. Broader patterns live in feed your AI with business data.

When you would rather skip the build

Building your own dashboard works, but it is also work you repeat every time the data refreshes. If your goal is to work leads rather than maintain code, the Vonsel dashboard already is the dashboard. Everything the extension captures lands in a mapped CRM with the leads plotted by location, filters built in, and no export step at all.

On top of the map you get the layers a hand-built page will not have out of the box. Sales Intelligence scores and organizes leads, Reviews Intelligence reads each business's reviews to surface pain points, and Email Intelligence drafts a tailored email per business. The result is a contextualized database instead of a flat capture, the difference explained in Google Maps data vs bought lists.

The extension also lets you reopen a past capture and add more businesses, attach reviews to what you already saved, and keep a history of your scrapes. Many people do both: build a quick AI dashboard for a one-off analysis, and use the Vonsel dashboard for the ongoing pipeline. Compare approaches in the AI Workflows hub, and make sure your collection stays on the right side of the rules with is it legal to scrape Google Maps.

Loading the extension takes three steps

Getting the data source running is quick. Download the extension, load it in Chrome using developer mode (see the Chrome extensions getting-started docs), then open Google Maps and click the Vonsel pop-up. That is it. No account wall before you capture, and the export button is right there when your list is ready.

From that single export you can go two ways in the same afternoon: hand the JSON to Claude Code and describe the dashboard you want, or drop the CSV into ChatGPT and start asking questions. Either way, the extension got real, structured business data out of Google Maps and into a file your AI can read.