FlowProofsData analysts
Build the metric tree. Skip the 40-slide deck.
Revenue = Traffic × Conversion × AOV, drawn as a live tree from your own tables, for the people who never open the BI tool.
The meeting you know too well
You have spent two days on the analysis. Revenue fell because conversion fell, and conversion fell on mobile after the checkout change. The SQL is clean and the charts are right.
Then you build the deck. Thirty-eight slides, because every stakeholder wanted their cut. In the meeting, the VP skips to slide 31 and asks a question the deck answers on slide 12.
Nobody sees the shape of the argument: one number, made of three others, one of which broke. That shape was the whole finding.
What data analysts build in FlowProofs
01
A metric tree from real tables
Write Revenue = @Traffic * @Conversion * @AOV and draw it as a tree. Each leaf reads from your data, so the tree is a model, not a picture of one.
02
What moved, by how much
Take a snapshot, refresh the data, and the why panel holds every input at its old value except one, to show how much each input contributed to the change.
03
Breakdowns without a new chart
Split a changed number by any category column, like device, region or channel, to find where the movement is concentrated.
What it replaces
Read next
- What a metric tree is, and why it beats a KPI dashboardA metric tree breaks one number into the numbers that make it up. A worked example, how to build one in a spreadsheet, and where KPI dashboards fall short.
- Conversion funnel analysis: finding the step that actually brokeConversion funnel analysis, worked through: conversion falls from 3.2% to 2.7% on $120K of ad spend. Find the step that broke and rule out false alarms.
- FlowProofs vs TableauUse Tableau if you have analysts who explore large datasets visually every day, need a wide range of chart types, or run Tableau Prep pipelines into governed data sources.
Questions
Is this a replacement for SQL or our BI tool?
No. Do the analysis wherever you do it now. FlowProofs is for presenting the result as a model people can interact with, built on the same data.
Can I connect to our warehouse?
PostgreSQL connects directly and read-only. For Snowflake, BigQuery and others, export the result set to CSV or Google Sheets and connect that.
How do formulas work?
Plain arithmetic and familiar functions: SUM, AVERAGE, COUNT, IF. Reference other numbers with @Name and columns with table.column. No DAX, no LOD expressions.
Can I join tables?
Yes. Drag a column onto a column in another table to relate them, and formulas that span both tables resolve through the relationship.