FlowProofsFrom spreadsheet to live canvas in four steps.
This is how you turn a spreadsheet into a dashboard people can question: upload it, drag it, write the math, share it. The first canvas takes minutes, not a project.
Step 1
Upload a CSV or Excel file.
Drop a .csv, .xlsx or .xls file on the Data page. We read every sheet, detect each column's type, and show you a preview before anything is stored, so you can pick the sheets and the header row.
Already keep the data somewhere else? Connect Google Sheets, PostgreSQL, Stripe, HubSpot, QuickBooks, Notion or Google Analytics instead, and refresh it whenever it changes.
A CSV becomes a table, and the table becomes a number on the canvas. Step 2
Drag columns onto shapes.
Every column shows up in the sidebar. Draw a rectangle, a circle or a diamond on the canvas and drag a column into it. The shape now holds a live number: the sum, average or count of that column.
Arrows between shapes are just arrows until you want them to mean something. Then they carry the arithmetic.
Drag a column onto a shape. The shape now holds a live number. Step 3
Write formulas in plain math.
Type = Revenue - Costs, or Margin = @Profit / @Revenue * 100. Name numbers the way you say them out loud and reference them with @.
There is no DAX and no filter context. SUM, AVERAGE, COUNT and IF cover most of what a monthly report needs, and every formula stays visible to whoever reads the canvas.
Formulas are plain arithmetic on named numbers.
Explain a change
Pick what to compare with in the top bar: last month, the same period last year, a snapshot, or a target. Every number then shows how far it moved. Click that badge, or right-click the number and choose Explain this change, and you get a bridge from the old value to the new one.
- The bars add up. Each driver gets its fair share of the change, worked out with Shapley values, so the pieces always sum to the total, even when drivers multiply each other.
- Where it happened. FlowProofs splits the change by the column that concentrates it most, such as region or channel, and separates a rate change from a shift in mix.
- Is this normal? The change is checked against how much the number usually moves, so a routine wobble is not presented as news.
Contribution shows what moved with the number, not proof of cause. To try it, open the “Why did revenue change?” sample when you sign up.
What happens to your data
- Stored privately. Uploaded files go to private storage under your account. Nobody else can read them, and nothing is published at a public address.
- Never used to train AI. AI features are optional. When you use one, the request goes to Anthropic, whose commercial terms say customer content is not used to train their models.
- Deleted when you say so. Delete a file, a canvas or your whole account from Settings. Deleting the account removes your files and every canvas with it.
The details, including encryption and who our sub-processors are, are on the security page. If you would rather read about the problem first, start with what a metric tree is. For every feature, see the product page; for the details of each step, the docs.