FactorPrism®
Our Technology

How FactorPrism® Works

The breakthrough AI that finds the simplest truth hidden in complex data.

Why Traditional Analytics Fails

When your metrics change, there are always multiple possible explanations. The hard part isn't finding correlations—it's figuring out which factors are actually responsible.

Overlapping Effects

Every data point belongs to multiple groups simultaneously. A sale in "East Region" is also part of "All Regions," making it impossible to cleanly separate causes.

Double-Counting

Your product is part of the market. If you drove growth, traditional methods count it twice—once as "market" and once as "your product."

Example: South Jewelry Sales Doubled

When sales of Jewelry in the South region double, it could be explained by any of these factors:

Overall market growth? Would affect ALL products in ALL regions
South region surge? Would affect Electronics in South too
Jewelry category boom? Would affect Jewelry in West too
Specific South-Jewelry factor? Unique to this segment only

The challenge: South-Jewelry belongs to all of these groups simultaneously. Traditional analytics can't disentangle which factor is actually responsible.

AI-Driven Factor Analysis

FactorPrism® Separates the Signal from the Noise

FactorPrism® uses a proprietary analytics engine to examine relevant business dimensions together, locate where each movement acted, and quantify each factor's contribution at the right level of the hierarchy.

The result is a complete Finding: measured factors, supporting evidence, and focused next investigations. Contributions reconcile to the reported change, so the explanation is clear, reviewable, and ready for action.

Built for Defensible Answers

Measured contributions come first. FactorPrism calculates where the movement occurred and how much each factor contributed before AI helps explain the result.

Relevant intersections stay visible. A factor can live at the level of a region, product, customer segment, or a meaningful combination, without counting the same movement multiple times.

Every Finding is reviewable. Users can move from the headline to factor evidence, timelines, comparisons, prior runs, and an Investigation Plan tied to the measured result.

What Makes a Finding Useful

Measured Factors Each contribution is quantified
Linked Evidence Inspect where and when the movement acted
Decision-Ready Findings Move from the answer to focused next checks

Why This Matters

This approach solves problems that traditional analytics fundamentally cannot.

Correct Attribution

Credit goes to the right level of the hierarchy

It All Ties Out

Every effect attributed once—the factors reconcile to your total

Automatic Simplicity

Prefers fewer, more powerful explanations

Scales to Complexity

Works with thousands of factors

See It In Action

Worked Example

The Region Hiding Behind the Baseline

Consider a retailer whose revenue grew 12% year-over-year. Was the lift company-wide, or did specific regions account for it? Here is how traditional analysis and FactorPrism® diverge.

Traditional Analysis (Wrong Level)
  • South (35% of revenue): +4.2%
  • West (30% of revenue): +3.6%
  • Northeast (35% of revenue): +4.2%
  • Problem: Allocates all 12% at the regional level
FactorPrism® (Right Levels)
  • Segment-wide baseline: +6.0%
  • South outperformance: +5.0%
  • West outperformance: +4.0%
  • Northeast underperformance: -3.0%
The Right Hierarchy Half the growth was company-wide (not regional). Northeast was hiding a -3% drag behind that baseline lift.

The Bottom Line

FactorPrism® goes beyond correlation. Its proprietary analytics engine locates the factors behind a metric movement, quantifies their contributions, and connects the result to evidence and practical next investigations.

Ready to See the Science in Action?

Try FactorPrism® free on Snowflake Marketplace and see what's really driving your business metrics.

Get it on Snowflake Marketplace