M-Factor offers services for the development of sophisticated forward-looking demand models that allow consumer goods companies to continuously analyze, forecast, and optimize marketing investments and trade spend.  Due to our advanced software-based "production modeling" approach to the development process, we are able to deliver high quality models in a fraction of the time of traditional modeling techniques.

Faster Development
Production modeling begins with the categorization of market/product pairs into groups that can be effectively modeled using similar functional model forms (meaning that they share the same variables and transforms).  A common example is to use one form for grocery, a different form for liquor stores, a third for on-premise, a fourth for mass merchants, etc.  Models are then estimated across these large groups that share similar functional forms using Bayesian shrinkage techniques.  Unique coefficients are created for each model to precisely describe consumer demand for a given market and product.  Careful review of these models then provides the final confidence that they will deliver as intended.  This approach saves literally months of effort that would be required to manually estimate every individual model coefficient.

Better Predictions
Most demand models perform poorly out of sample because they are designed for retrospective analysis. This is the reason why the "other" or "unexplained" bars (euphemisms for model error) in Due-To reports grow over time. M-Factor's Modeling Workbench produces models that consistently generate more predictive models (note: In a recent benchmark example, M-Factor models empirically reduced out of sample error by 68%, with .99999 confidence).

M-Factor models are developed using our patented modeling software, which utilizes standard statistical libraries, but with a better user-interface for the modeler to produce superior models. Rather than they typical "iterative" approach, our software produces models in parallel, permitting the modeler to evaluate more models more quickly, and resulting in a superior end product.

M-Factor models answer all the traditional marketing mix questions without relying on "industry standard" approaches which don't reflect the nuances of particular businesses. This improves model accuracy and generates models that can predict well out-of-sample. All modeling transformations and data sets are completely visible within the M-Factor modeling software.

Up-to-date Models
M-Factor's state-of-the-art model building and maintenance platform allows modeling staff to evaluate changes in consumer response to new campaigns as well as market events and competitive actions/launches as they happen. These updated models are then immediately available to evaluate the impact of changed response on performance versus plan. These updates are typically conducted quarterly so that model findings and forecasts reflect the current marketing environment instead of last year’s.