Exact attribution. Reconciled to the total. In your warehouse.
New in 5.0: connect a table and get a one-sentence verdict instantly — before you configure anything.
FactorPrism® uses AI to automatically attribute every change across the dimensions you provide—region, product, channel, segment, and their intersections—and shows exactly how much each factor contributed, at the right level, in seconds, not weeks. Running natively in Snowflake, your data never leaves your secure environment.
Free during early access · 30-day Marketplace trial · No card required · Built-in demo — see a full analysis before connecting any data.
Not on Snowflake? Download the FactorPrism® app for Windows, Mac, or Linux — same analysis engine, runs entirely on your computer.
Behind every metric shift are hidden causes: seasonal patterns, market trends, operational changes, competitive moves—all overlapping and compounding in ways that are impossible to untangle manually. These drivers often hide at the intersection of dimensions—a specific region-by-product combination—where one-dimension-at-a-time reports never look.
FactorPrism® uses AI to automatically attribute your metric change across the dimensions you provide. It separates overlapping effects, isolates each factor's contribution, and quantifies exactly how much each one moved your result—work that would otherwise take analysts weeks to assemble by hand.
The result? You see not just what changed, but exactly which factors—and which segment intersections—account for the change, and by how much.
All the power of advanced root-cause attribution with the security and governance your enterprise demands.
Stop guessing. Automatically identify the specific segments driving changes in your metrics.
Analysis runs entirely in your Snowflake warehouse. Your data never leaves your environment, with full governance controls.
See exactly how much each driver contributed with intuitive visualizations and AI-powered insights.
From install to your first verdict in minutes — the answer appears the moment you connect.
Grant FactorPrism® read-only access to a table. We only need SELECT permissions—your data stays exactly where it is.
The moment a table is connected, FactorPrism® leads with the answer: a one-sentence verdict of what moved and why, with the top factors ranked—before you configure anything.
Refine dimensions and time windows, drill into the full reconciled bridge, or schedule the same analysis headlessly via SQL.
Every factor sums to the total change—no plug
Across 280 blind benchmark runs, in its operating regime
Pinpoints the segment and intersection, not just the headline
In our NYC case study
When we analyzed NYC's public 311 service data, the headline looked simple: 11% growth in complaints. But FactorPrism® automatically attributed this across category and time to surface the real story:
FactorPrism® surfaced these patterns automatically from the public 311 dataset—no manual segment-by-segment hypothesis testing required.
See How We Did ItOne number (11% growth) became dozens of actionable insights:
"Complaints are up 11%"
"Overall growth masks a 25% drop in housing issues, seasonal water system stress, and infrastructure failures in specific years"
Revenue grew 12%. Traditional analysis allocated it all at the regional level. FactorPrism® found the right hierarchy: half was a company-wide lift, while Northeast was hiding a -3% drag behind that baseline.
Read case studyTraditional BI showed "Jewelry +22%" and "South +24%" as separate dimensions. FactorPrism® found the right level: South + Jewelry intersection drove 9% of the 18% growth—the real driver was one specific combination.
Read case studyFirst-pass denial rate jumped 8% to 12%. By payer, by code, and by service line, three dashboards told three different stories. FactorPrism® found that one Payer × CO-197 × Orthopedics intersection drove half the spike—built for payers and RCM vendors.
Read case studyHand-built bridges double-count, smear broad-based forces across every region, and only "tie" because of a plug. FactorPrism® builds the bridge automatically — every factor located where it acts, summing exactly to the change.
See the recipeMargin data crosses zero — segments flip from profit to loss, and standard attribution math quietly falls apart. FactorPrism® handles sign-crossing data natively, separating gains from declines so offsetting stories stay visible.
See the recipeNet flows hide offsetting stories: strong acquisition in one region can mask a churn problem in another for months. FactorPrism® locates each component — which segments, which direction, starting when — reconciled exactly.
See the recipeA miss to plan is rarely uniform—a beat in one place can hide a collapse in another. FactorPrism® locates each piece of the variance across region, product, and channel, and the bridge ties out to the total gap exactly.
See the recipeAlready testing Snowflake's built-ins? They flag where a metric changed. FactorPrism® goes further—the right level including intersections, gains and declines separated, and a breakdown that ties out to the total—all as a native app, your data never leaving the account.
See the comparisonHow FactorPrism's reconciled-to-the-total, intersection-level attribution stacks up against the tools you already run in your warehouse, and the thinking behind it.
Let AI decompose your performance into its underlying causes. FactorPrism® runs natively in Snowflake—your data stays secure while you finally see what's really driving your performance. Prefer to work outside Snowflake? Download the FactorPrism® app and run the same analysis locally on Windows, Mac, or Linux.
Free during early access · 30-day Marketplace trial · No card required · Built-in demo — see a full analysis before connecting any data.
Install FactorPrism® from the Snowflake Marketplace and grant it read-only access to the table that holds the metric. The moment the table is connected, FactorPrism® leads with an instant one-sentence verdict of what moved and why, with the top factors ranked. From there you can drill into a full attribution that reconciles exactly to the total change.
No. FactorPrism® is a Snowflake Native App: the analysis runs entirely inside your Snowflake account, and the app only needs read-only SELECT access to the tables you choose. Nothing is extracted to any outside service.
FactorPrism® is free during early access, with no card required. Post-GA (target Q4 2026), pricing follows Snowflake's consumption model: you pay only for the warehouse compute you use, billed through your existing Snowflake account, with no seat fees and no annual contracts. See pricing for details.
Yes. The app ships with a built-in demo that shows a complete analysis on sample data—no data connection required. There is also a 73-second video demo if you want to see it before installing anything.
Yes. FactorPrism® exposes a headless stored procedure, API.RUN_DECOMPOSITION, that runs the same analysis as the app from a single SQL CALL—from a Snowflake task, an alerting pipeline, or an AI agent such as Snowflake Cortex Code. Results come back as a queryable table and every run is saved for later fetching. See the Headless API reference.
TOP_INSIGHTS is a SQL function that returns a flat list of segments whose metric changed; the segments can overlap, and they are not organized into a hierarchy or reconciled to the total change. FactorPrism® computes a complete answer—each factor located at the level where it acts, gains and declines separated, everything summing exactly to the change—and presents it in a full interface with an AI-written explanation. See the detailed comparison.
