Go to main content

Subscribe to Signals quarterly newsletter



More Information»

Download PDF for Printing [132k]

More Information
+1 505-995-4720
Email

Tomorrow Will be Different: Quantifying the Reality of Consumer Dynamics
Signals, Spring 2001

Nature often takes the lead. When it comes to providing a laboratory for the development of new analytic technologies, Weather patterns, speech patterns, and sunspots have been examined by academics for their underlying structure and potential for spawning advances. At Strategic Analytics, we've chosen instead to focus on the complexities of consumer behavior for our guideposts. Our team has decades of experience studying complex and chaotic systems. The dynamics of consumer behavior, critically important to the businesses of our clients, demand a fundamentally new approach to data mining.

Odds of model drift occurring by this amount or more in one year
1 in 2 26.4%
1 in 5 51.5%
1 in 10 66.0%
1 in 20 81.6%
1 in 50 97.8%
Model Drift

Consumer behavior is nonlinear and highly dynamic—consumers adapt their relationship with a company's product or service, but the environment in which they exist is unstable. Over the course of a year, models can drift dramatically. The table at right shows the drift observed on real consumer data in a calm economic period. Product or policy changes, competitive offerings, seasonal effects, and macroeconomic conditions can all drive dramatically different behaviors. Most technologies designed to project consumer behavior gloss over these factors, preferring the simpler approach of assuming that tomorrow will be like yesterday. As a consequence, tomorrow's business decisions are driven by models that meticulously encapsulate yesterday's reality.

In contrast, Strategic Analytics' technology extracts power from the dynamic nature of consumer behavior. We examine that behavior to determine what aspects are related to events and actions of a particular time frame. We quantify such shocks in the form of an exogenous curve. This curve represents the consumer's response to product and policy changes, as well as seasonal effects and macroeconomic conditions. The curve provides invaluable feedback for setting acquisition strategy and product features—and for projecting customer lifetime value and portfolio performance.

Traditional approaches lack such quantification of environmental impacts. Consequently, a model built using historical data is implicitly projecting that the past environment will recur. Practitioners refer to the resulting error in projected portfolio metrics as model drift. More than indicating that the models will degrade over time, model drift guarantees that the models will fail in times of rapid change--precisely when they are most needed. The table at right illustrates the magnitude of drift expected in key portfolio metrics, relative to projections provided by standard modeling techniques.

Through quantification of exogenous impacts, and the ability to assert new environmental scenarios for the future, Strategic Analytics' technology is revolutionary. Static models of behavior are replaced by customer lifetime value metrics computed across a range of future environments. Consumer segments are defined based upon intrinsic consumer behavior, not the environment present when they became a customer. Scenario-based portfolio forecasts replace muddied extrapolations of past results. Portfolio risk management is based upon a broad range of simulated futures, all reflecting the reality of environmental change.

In today's competitive environment, razor sharp analytics are a key to survival. The current generation of models has been tuned and retuned to its limits, and the demand for more accurate tools requires more than incremental change. We believe businesses seeking new means of competitive advantage will adopt approaches recognizing that tomorrow will indeed be different.

At Strategic Analytics, we are committed to helping them succeed.
Top of page