Marketers have access to vast sources of transactional, causal, attitudinal, behavioral, financial, economical and digital information. SMS mines large quantities of data efficiently by leveraging proprietary statistical models, providing actionable commercial insights to our clients.
SMS models incorporate breakthrough approaches in statistical modeling and advanced analytics:
- Bayesian Hierarchical (BH) modeling enables SMS to generate stable, multi-level nested estimates with limited data. Hierarchical models provide a mechanism for pooling information for disparate groups without population inclusion assumptions. The Bayesian Hierarchical approach is employed in SMS regression, logit and conjoint models.
- Multiple Comparison Procedures (MCP) are used in experimental designs where multiple test conditions must be simultaneously analyzed to understand the relative difference between each test treatment. This procedure is preferred when one-way analysis of variance (ANOVA) is not feasible for producing acceptable mean differentials due to inherent limitations around multi causal factor analysis.
- Logit & Discrete Choice Models (L/DCM) estimate consumer choice between two or more mutually exclusive alternatives based on behavioral probabilities.
- Seemingly Unrelated Regression Equations (SURE) consists of multiple regression equations, each having its own dependent variable and different exogenous explanatory variables. Each equation is an independent linear regression and can be uniquely expressed as a coefficient.
Bayesian Hierarchical Models
- Price modeling
- Promotion modeling
- Advertising, print & consumer promotion modeling
- Online advertising modeling
- Forecasting
Logit, Conjoint & Discrete Choice Models:
- Survey analytical modeling
- Consumer choice modeling
- Reference price/value modeling
ANCOVA, ANOVA &
MCP Models
- Control store testing
- Matched market testing
- Store group testing
- Post promotion testing
SURE / Rotterdam Models
- Dynamic competitive modeling
- Assortment modeling
- Market structure modeling
Imbedded in proprietary SMS analytics, four significant econometric and utility-based principles differentiate our models from other vendors:
- Asymmetric Price Elasticity: refers to the asymmetric relationship between price and purchase behavior. Driven by retail discounting practices, magnitude of consumption increase due to a price decrease is greater than the magnitude of decrease in consumption due to a price increase. Standard logarithmic models do not necessarily capture this dynamic.
- Advertising Ad-Stock & Half-Life: Ad-stock captures the long-term or prolonged effect of advertising on consumer purchase behavior by modeling how advertising builds and decays over time. Advertising half-life is the duration it takes for the advertising awareness to drop by half; it is estimated by mathematically applying the ad-stock model. SMS uses proprietary ad-stock and half-life estimation techniques to capture advertising impact on consumption.
- Diminishing Marginal Utility Overlay: Marginal utility is the incremental utility consumer derives from the purchase of each incremental unit. Law of diminishing marginal utility means that as the consumption of a product increases the marginal utility will decline. SMS model applies this economic principle in models for marketing mix and assortment analysis.
- Dynamic Consumer Choice: The link between personal preferences, consumption, and demand is one of the most complex relations in economics. SMS Dynamic Competitive Models [TM] integrate consumer preference, consumption and demand to understand substitutability and product utility.
SMS does not release proprietary algorithms and statistical architecture developed by our modeling team; however, a number of leading academic white papers that form the core architecture of our models are available for peer review and comment in Academic & Commercial Research Documents section of this site.
