Our Technology

The Nucleus Platform

Our team of MBAs, PhDs, computational scientists, and applied mathematicians realized that we needed a different approach to modeling in order to get the results needed to unlock true causality of demand and make better business decisions.

We found that:
1. The 4 “persistent problems” can all be framed as part of a planning cycle
2. Those problems persist because each part of the cycle faces a technical hurdle
3. Our technology addresses each of those hurdles

The Planning Cycle

Most businesses experience some version of a planning cycle: plan, forecast, optimize, replan, forecast again, try to further optimize, and so on. While the goals are clear: how do I know what will happen?, and how do I know what I should do?, each is challenging for a specific reason. These challenges explain why the four problems persist through time across so many consumer durables brands.

WP's unique technology allows us to know what will happen, and what you should do, even under evolving consumer behavior and new business strategies.

Accounting for Causality

In order to know what will happen, you need to account for each of the causal mechanisms driving consumer demand, as well as those driving retailer purchasing. Each planning decision must be linked through to a driver to a forecast response. If moving half of your Facebook budget to radio doesn't change your forecast, how can your forecast help you plan?

We have assembled a "catalog of causes" by mining the academic business literature on consumer behavior. This allows the WP forecast to support your planning cycle by responding to every ad spend line item, promo period, store opening, product launch, and more.

Separating Concurrent Causes

A sufficient "catalog of causes" is still insufficient to solve the planning cycle problem. To learn to quantify the impact of each cataloged sales driver on your specific business, you have to assign the right sales responses to the right of many concurrent causes.

Our approach separates each cause and quantifies how much it contributes to the sales response, enabling optimizations that don't double count, discount or miscount the true causes. Our technology uses state-of-the-art Bayesian A.I. combined with our catalog of causes to solve this concurrent causes attribution problem.