This paper will describe the collaboration between Hershey's and In4mation Insights to revolutionize how marketing ROI is modeled and how its results are spread throughout an organization. The development of new-to-the-world Bayesian statistical methods, coupled with scalable and speedy software run using parallel processing on the 'cloud' will be addressed. Results have then been placed in a marketing enterprise-wide simulation model, where the findings have been pressure-tested by senior executives. The presentation will document how a close relationship between a savvy client and advanced modelers led to groundbreaking results.
The question of the sales performance achieved through radio campaigns - not so much in tactical as in strategic terms - was foremost in the authors mind in developing an econometric model of turnover activities. The model developed was designed not only to take into account media expenditure but also to include other important parameters relevant to turnover such as pricing and distribution. Data concerning specific brands of yoghurt are used as an example to test this model and its data is studied in detail.
The writers have developed a new approach that draws together these two sets of dimensions. Based on advanced econometric methods, it presents a simultaneous analysis of both economic and image measures. This methodology permits managers to draw conclusions that are based not only on hard economics, but are also directly linked to a brand's 'meaning'. This method is a breakthrough in brand management. For the first time, a brand's performance can be understood as an integrated whole. Managers can develop specific and actionable marketing plans whose economic impact can be accurately predicted.
There are still controversial opinions on how to quantify the contribution of classical advertising to market success. The development of advertising-effect-formulas, considering the calculable relationship between advertising expenditures and market success, began in the mid- 1950s in Germany. These econometrical formulas measure changes of market shares in dependence of advertising expenditures and other relevant marketing parameters during consecutive periods. The first "WerbeWert" study by VDZ (magazine publishers association in Germany) as well as its continuation named "WerbeWert 97" will be described. This new study considers print media and TV as well as radio and poster advertising, based on a five year period. As macro-formulas with generalizable knowledge, econometrical studies do not replace creative strategies or media planning, but are very suitable for better comprehension of advertising effect correlations and could also help assess advertising strategies. Finally they prove the advantages of media mix advertising can be calculated on a secure empirical basis.
Satisfied customers and customer retention should be one of the highest priorities of any business enterprise. Many companies are using some form of customer satisfaction measurement, but most companies do not have methods to put the customer satisfaction program in relation to profitability. Only very few studies have focused on how to link customer satisfaction to customer loyalty and performance. This paper will present a microeconomic model for the relation of profit to customer loyalty, customer satisfaction and cost. It is shown that the degree to which the company lives up to customer expectations should be a linear function of the contribution to loyalty. The framework enables the managers to determine which customer satisfaction elements should be improved first with respect to bottom- line profitability. For practical use the only information needed is market information about customer satisfaction for the company and the main competitor, the importance of a given satisfaction element and the customers buying intentions.
This paper examines Retail Audit data captured in Egypt using econometric modelling. The approach currently is not purely directed at forecasts or predictions but an understanding of trend movements. ARIMA and Transfer Function Modelling has been carried out on toilet soap data captured over 1993 to 95.
Three Dutch orchestras approached Intomart and SEO for a demand analysis regarding concerts of classical orchestral music in the Amsterdam area. The main object was to determine the attributes that constitute concerts of classical music, as well as the relative weights of these concert attributes. The problem has been approached by a hybrid mix of conjoint measurement and a strongly generalized own brand of logit analysis (ROA). Some 1100 respondents have been offered varying subsets of 32 vignettes of concert offers, most of them fictitious but not unrealistic. From the analysis it is possible to determine the change in demand for classical concerts as a consequence of a change in an attribute value, as well as the price compensation necessary in order keep the level of demand unchanged. According to our knowledge this is the first study on the demand for classical concerts making use of vignettes. Furthermore, it is one of the first applications where the collection of data by supplying vignettes is blended with fairly sophisticated econometric techniques for discrete choice analysis.
The downward pressure on pharmaceutical product prices and the level of profitability is already forcing companies to consider how to cut costs and to gain the most out of the money that is spent. An improvement in the way resources are allocated will be the focus of all companies in the latter half of the 1990's and beyond. Although a great deal of information via audits and primary research is available, it is still surprising that relatively few companies are interrogating the data via econometric analysis to help understand the marketing and advertising forces that affect a specific brand's behaviour and the magnitude of these effects - a procedure used in the consumer area since the 1960's. It is vital to know what the likely effects will be on brand performance when considering a change in the promotional spend and the allocation across different media. It is also essential to understand the pricing dynamics in the market to assess the likely response to price changes. Short to medium term forecasting can be made much sharper and resources used more wisely if these issues are better understood. Furthermore, the need to review the return on the promotional investment from different brands within the company becomes a logical step for the marketing director. In future, despite the changing customer base, the attitudes of managed care groups towards different scenarios can be monitored via primary research using "choice-based conjoint" investigations to provide additional input for short to medium forecasting accomplished by econometric analysis. Creating great efficiency and helping to save valuable resources can also help to save jobs.
In this paper, a Decison Support System is presented that enables the decision-maker to make optimal use of existing marketing research data in the area of fast-moving consumer goods. Many valuable market data are not analyzed in depth by the decision-maker, who, therefore, do not get all the possible information out of the research budget. "SalesPlan" is a model that combines information that is already available by most brand producers. The main data of the analysis are time-series data of Category Sales, Brand Sales and Market Share, and corresponding values of the brand price, average competitor price, company and competitors distribution, amount of company and competitors in-store promotion, and data of company and competitors retail inventory. All these data are available from Nielsen-Data, which most brand producers buy regularly, typically on a bi-monthly basis. Also, the analysis is based on GRP- and Adstock data of company and competitors' advertising efforts. These data are also available on commercial basis in most product areas. Finally, the decision-support part of the model uses company-internal data of brand production costs. Since data are time series, "SalesPlan" models trend and seasonal patterns if they exist. "SalesPlan" consists of two separate modules: A forecasting module and a decision support module. In the Forecasting Module econometric models of Category Sales, Brand Sales and Market share are estimated as functions of the set of explanatory variables. The ability of the models to describe the historical data provide valuable insight into the relative importance of the marketing decision variables and are basis for the decision support model. Based on forecasts of the important competitors' decision variables and the company's own marketing plans, short term forecasts of Company Sales, Brand Sales, and Market Share are developed.
Heavy up TV advertising for packaged goods has been the traditional way of estimating its effect on sales in the United States. With recent advances in information technology, testing can now be replaced by econometric modelling techniques that removes the influences of other marketing factors such as price changes, promotions, distribution and competition. The result is estimates of different sales responses over a range of television GRP delivery levels. This allows the TV budgeting and the local market allocation process to be planned on the basis of its contribution to sales. This paper begins with a discussion of evaluating advertising's sales response in the U.S. in an historical context, beginning as a design-driven process and recently moving to a model-driven process. The subject of econometric analysis for packaged goods product's advertising is introduced, and the technique for handling the residual effects of advertising is discussed. The non- linear nature of the advertising-to-sales relationship is displayed in a unique format for direct, end-user decision- making. Live examples demonstrate the power of the approach on both a national and local basis. There is a discussion of continuous vs. flighting schedules, day parts and limitations of the approach. A special-case application of econometrics is discussed for those who are wedded to test vs. control designs.
The purpose of this account is to set out the results of 3 years of concrete experience gained on this model and its simulation software. It breaks down into three main sections: I. The principle of the model insofar as concerns data collection and the econometric assessment of the latter. II. The various types of simulation which are possible with the software; various examples are given in this connection. III. The setting up a new mode of cooperation between the advertiser's marketing departments and the company making the studies.