An Introduction to Marketing Mix Modeling

When one considers the vast number of advertising channels available and the often short shrift given to marketing budgets, one wonders how exactly large firms make the decisions to purchase advertising in particular venues or determine which promotional activities to conduct. Every first-year business student should know the four P’s, but that yields very little insight into how to make strategic decisions about purchasing advertising that will be the best choice for the firm, whether in the short run or in the long run. Here is our begnner’s guide to marketing mix modeling.

Modeling A Firm’s Marketing Mix Refers To The Combination Of: Analysis and Simulation

There are a number of different simulation models used for assessing creative work, both proprietary software applications and models used by a firm’s internal marketing department. The most common fall under the rubric of “media-mix” or “marketing mix” modeling (MMM), which generally uses time series analysis to estimate the impact of planned promotions on gross sales revenue and net income.

Understanding Time Series Analysis And Its Uses

Time series analysis involves building models of activity over a particular period of time (over even intervals). The models are used to predict future behavior and are usually over relatively short periods, in line with marketing campaigns. The use of longitudinal data is complicated by the number of variables that enter the model the farther one goes back. For example, a ten-year analysis of online marketing for a national consumer packaged goods firm would have to account for changes that occurred within the ten years: changes in the marketing mix (development and usage of new channels like blogs), changes in the market space of the industry (recent improvements in supply chain management), and changes in the macro-environment (changes in consumer purchase behavior given the recent economic crisis).

That kind of deep forecasting is of less prevalence among marketing and sales managers who are trying to prove their worth to the firm. Advertising and its effectiveness are in some ways less tangible and harder to quantify than areas, such as manufacturing and accounting, particularly when one discusses long-term marketing objectives, such as building brand equity as we’ve seen from the recent financial meltdown, time series forecasting that focuses on short-term predictions discounts longer term variables which can be game-changers.

Simulation Modeling and Market Mix Modeling

Serious attention has been paid to the academic study of marketing mix modeling since the fifties. Most of the more cutting-edge simulation models are developed by marketing research firms or internal marketing departments and kept a carefully guarded secret. But academics have long tinkered with formulae that might yield some insight into advertising and its effect on consumer behavior.

In their excellent survey paper, Leeflang and Wittnick offer a survey of the use of simulation in marketing mix modeling in both practice and in academia. In brief, in both arenas, during the first era, from roughly 1950-1965, basic operations research algorithms were applied to basic sales problems. During the second era-from, from 1965-1970-researchers started to apply time series analysis and other econometric tools to marketing problems; however, these models usually considered too few variables. The third period, from 1970 to 1985, was marked by the use of simulation and models to describe consumer purchase behavior-but inputs were still often too limited. This is followed by the present era, which involves serious academic consideration of standardization of models for various marketing problems, a greater variety of complex marketing problems, and meta-analyses using existing studies.

Today’s marketing simulation modeling, whether of the marketing mix or of consumer purchase behavior, generally involves Monte Carlo simulation, but since the variation involved in Monte Carlo simulation usually are estimates based on historical data, and since channels are constantly evolving, simulation models have to take into account carefully the impact of new channels on consumers.

Leave a Comment