Skip to main content

• Problem
• Workflow
• Builder
• Use Cases
• Patterns

Dark

• Problem
• Workflow
• Builder
• Use Cases
• Patterns

01 A Practical Guide — No Code Required

Turn any report into a live client dashboard

Your clients don't need another PDF. They need something they can click, filter, and explore. With Claude or ChatGPT, you can build a bespoke interactive dashboard from an Excel file or PDF report using nothing but plain language.

See the workflow → 1 Prompt to get started ~5 min First version ready 0 Lines of code needed The Problem

Static reports leave clients doing the work

You spend hours building a clean Excel workbook or polished PDF. You send it over. The client opens it, scrolls through it, and then emails you three questions that were already answered in the data — they just couldn't see it. Static deliverables are one-way. They don't let clients filter, explore, or ask their own questions.

An interactive dashboard changes that dynamic entirely. It puts the data in the client's hands in a way they can actually navigate. And building one no longer requires a developer, a BI tool subscription, or any technical skill on your end.

Static report — what you send today – Fixed view — one layout, one perspective – Client can't filter by date, matter, or category – Every new question means a new export from you – Key numbers are easy to miss in a wall of cells – Feels like a data dump, not a tailored deliverable Interactive dashboard — what this guide builds ✓ Client controls the view — clicks to filter and explore ✓ Charts update live with dropdowns and toggles ✓ One shareable file handles most follow-up questions ✓ Key figures are prominently surfaced and labeled ✓ Feels bespoke, built specifically for their situation The Workflow

How to build a dashboard with Claude — four steps, start to finish

This is the complete process. It works with Claude — which is recommended for producing clean, self-contained HTML dashboards — or ChatGPT. Select each step below to see exactly what to do at that stage.

1 Prepare your data Clean and export your source file 2 Upload to Claude Attach the file directly in chat 3 Write your prompt Describe what to show 4 Iterate & share Refine in conversation, then export Interactive Tool

Build your Claude dashboard prompt

Before typing anything into Claude, think through what your client actually needs to see and do. The selections below guide that thinking and assemble a ready-to-use prompt. Copy it directly into Claude alongside your uploaded file.

Prompt Builder — compose your first message to Claude 1. What are you uploading?

Before selecting: what format is your source file, and how clean is the data?

Excel (.xlsx) CSV PDF report PDF with charts 2. What should the dashboard surface?

Before selecting: what are the two or three numbers your client will actually care about?

Totals & KPIs Trends over time Category breakdowns Comparisons 3. What interactivity does the client need?

Before selecting: will your client want to slice the data themselves, or just read the highlights?

Dropdown filter Summary headline row Sortable table Clickable charts 4. Who is the audience?

Before selecting: how technical is your client? Should the dashboard explain itself, or just show the data?

Non-technical client Technical / expert Executive audience Needs narrative context Your generated prompt — copy into Claude alongside your file Make at least one selection above and your prompt will appear here. Copy prompt to clipboard Reset Use Cases

What this looks like in practice

The pattern is the same across professions: upload your data, describe what you want the client to see, and let the AI generate the interface. Here's how it plays out in three common contexts.

01 Law Firm Matter cost breakdown for a client

You have a monthly billing summary in Excel. Instead of a flat spreadsheet, you build a dashboard that lets the client filter costs by matter, attorney, or task type. A headline section shows total fees, outstanding balance, and hours billed this month. The client gets a self-service view without five follow-up calls to you.

02 Engineering Firm Project progress report for a developer

You export milestone and budget data from your project management tool. The dashboard shows completion percentages per phase, flags anything over budget or behind schedule, and includes a sortable table of all line items. What used to be a 12-tab workbook becomes a single-page view the client checks weekly.

03 Financial Adviser Quarterly portfolio summary

You pull account performance data into a CSV. The dashboard shows overall returns, allocation by asset class, and a time-series chart of portfolio value — with a dropdown to isolate any single account. A plain-English interpretation below each chart means you no longer need to write a separate commentary memo.

Pre-flight Checklist

Before you hit send

Work through this list before uploading your file. Check each item off as you confirm it — the progress bar tracks your readiness.

Readiness checklist 0 of 6 complete ✓ My file has clear, descriptive column headers in the first row ✓ I have removed any footer summary rows that aren't raw data ✓ I know the two or three key figures my client cares most about ✓ I know whether the client will filter the data or just read the highlights ✓ I have removed any sensitive data not intended for the client ✓ If using a PDF: the file is text-based, not a scanned image Prompt Patterns

Prompts that reliably work

These are proven prompt structures for common scenarios. Expand any item to see the full prompt text you can adapt and paste into Claude alongside your file.

Starting from scratch with an Excel or CSV file ⌄

Use this when you have raw data in Excel or CSV and want Claude to do the full job: parse the data, design the layout, and add interactivity. The more specific you are about what to show, the better the first version will be.

I've attached an Excel file containing [describe the data — e.g. monthly billing records by matter and attorney]. Create a single-page interactive HTML dashboard that shows: [list 2-3 things — e.g. total fees billed, a breakdown by matter, and average cost per attorney]. Include a dropdown filter so the user can select a specific [category — e.g. matter or month] and all charts update accordingly. Use plain-English labels throughout. Return the complete HTML file as a self-contained artifact. Extracting data from a PDF report ⌄

Claude can read text-based PDFs directly and extract numerical data from them. This is useful for board reports, published analyses, financial statements, or any structured document you receive from a third party.

I've attached a PDF report. Please extract all numerical data from it — any tables, figures, and key metrics mentioned in the text. Then build an interactive HTML dashboard that displays: [describe what you want shown]. Include a summary section at the top with the most important figures highlighted in large text, and charts below. Return the full HTML as a self-contained artifact I can open in a browser. Iterating on an existing dashboard ⌄

Once Claude has produced a first version in the conversation, you can refine it without starting over. Just describe the change you want in plain language. Claude edits the existing code rather than regenerating from scratch, so revisions are fast.

In the existing dashboard, please make these changes: 1) Change the bar chart on the left to a line chart showing the same data as a trend over time. 2) Add a "Total" figure in large bold text at the top of the page. 3) Change the chart fills to dark navy blue. Keep everything else the same and return the updated full HTML. Adding a written narrative alongside the charts ⌄

If your client needs context rather than just charts, you can ask Claude to write interpretive text directly into the dashboard — so it reads like a report and works like an app at the same time.

Below each chart in the dashboard, add a short paragraph (2-3 sentences) that explains in plain English what the data shows and what the key takeaway is. The tone should be professional but not stiff — write as though explaining findings to a senior client who is smart but not a data expert. Do not just describe what the chart shows. Interpret it. Refreshing the dashboard with new data next month ⌄

The dashboard you build this month becomes a reusable template. When the data refreshes, upload the new file with the same prompt and get an updated version in minutes. No rebuilding, no reformatting.

I've attached the updated Excel file for [next month / quarter]. It has the same column structure as before. Please rebuild the dashboard using this new data, keeping exactly the same layout, charts, filters, and interactivity from the previous version. Return the updated full HTML as a self-contained artifact. What's next Put it into practice with En Dash.

The first dashboard won't be perfect. Send it, get feedback, and refine it in the same conversation. That iteration loop — not the prompting itself — is the actual skill. If you want help making it yours, we're here.

Explore our Governance Workshop → Soundboard with Steve →