Marketing Mix Analysis
A medium-sized manufacturer had doubts about the effectiveness of its advertising spend in recent years. When approached with this issue, our team set out to quantify the impact that the various types of advertisement spend had over the previous five years. After sitting with the executive steering team and collaboratively defining a path forward, we applied a generalized estimating equation (GEE) method to the longitudinal dataset provided.
Ultimately, the client was able to look at all ten of their brands and rank-order the revenue impact that various promotions had (e.g., radio, TV, print, etc.) over the five-year period investigated. In addition to this marketing intelligence, we also built a prediction simulator that still allows the company to input an amount of money into one of several types of spend and see the predicted revenue that would result from that spend.
Developed an Interactive Presentation for Sales Team
A large financial services company sought a customized solution for their sales representatives to sell life insurance to wealthy individuals. We built an Excel engine where they could work with different assumptions (e.g., growth rates, interest rates, market volatility). The engine produced VaR calculations, for example, which ranked maximum loss not to exceed some given probability over a given period of time. The engine was an interactive model with interactive graphs, where customized assess risk scenarios could be generated on the fly. We included training to their team to use these models. The end result was an effective, flexible, scalable, presentation for their customers and sales team.
Segmented Retail Operations
A large eye wear manufacturer came to us with a large data-set on the preferences and sensitivity of individuals to price of glasses, style of glasses, children and adult glasses, size and location of the stores, internal and external labs, gender and age of the client, and several other client and retail factors. We conducted a cluster analysis to help the company group individuals and retail factors so they could design and fill the stores with the most marketable products. In cluster analysis, the goal is to use the data to group individuals into clusters or segments and also to help determine markets. We used K-means clustering which found a clusters’ center, then grouped customers around that center, and hierarchical clustering which treated each customer as a separate cluster then grouped them into larger and larger clusters. Additionally, we created an excel engine so the client could continue to add observations and determine their particular cluster.
What Product Should I Sell?
A South American Ice cream manufacturer conducted a survey in several countries on perceptions of flavors, texture, taste, appearance, and overall satisfaction of various flavors. Hundreds of inferential statistics and follow-up statistics were conducted to help the manufacturer decide what ice cream to sell next.
We created a ratings profile for different types of ice cream. The goal of their business strategy was to use customer satisfaction as a way to compete in the competitive ice cream business while keeping existing consumers loyal and targeting new customers. Hooray for researchers of ice cream producers!
Maintaining the Right to Sell Your Product: Assisted a Prosthetic Manufacturer in Determining Failure Rate
A mid-size prosthetic manufacturer approached us with a unique dilemma: It had been issued a notice of non-compliance and informed that its products would be prohibited from being sold in an entire country. Undeterred, the company brought us on to work alongside their team to conduct survival analyses that would determine whether the amount of defects was consistent with the industry. (Survival analysis in this context involved modeling the time to replacement of the prosthetic and how long it lasts.) The accuracy of the analyses was critical, both for the company’s own survival and for the patients who potentially had to undergo another surgery.
Upon analyzing the company's datasets, and then comparing our analysis to the analysis from the industry’s National Joint Replacement Registry, we found many inconsistencies that required recalculations. Of particular interest to the client were our recalculations of the revision rates and hazard ratios (i.e., the number of repeat surgeries that needed to be performed) because these were the reasons cited regarding why the company would no longer be free to sell in the country in question. And while the company was under an extremely tight timeline, our team provided them with statistically sound findings that refuted the previous analyses. As such, the company still sells its products in that country today.
Survey Instrument Validation
A survey instrument developer needed to assess the reliability of their instrument, and to certify a validation report. The data was statistically evaluated to determine the reliability. The reliability was measured using the Cronbach Alpha, which basically measured how well a set of items (or variables) measured a single unidimensional latent construct. In addition, the Standard Error of Measurement (SEM) was calculated which refers to the standard deviation of test scores that would have been obtained from a single participant had that participant been tested repeatedly.
In any type of market research, the analyses should take into account different types of error, such as reliability and validity. Reliability basically refers to the consistency in which participants respond to a set of questions (i.e., the Cronbach alpha cited above, test-retest, Spearman-Brown, or split-half). Validity refers to measuring what you say you are measuring such as content, criterion, construct, convergent, divergent, and nomological validity. When developing an instrument or analyzing yours, we take into account both reliability and validity.
Retaining State Funding: Determine ROI for a Nonprofit Organization
A state-wide not-for-profit organization sought to retain their State funding and needed to justify their efforts by conducting a ROI on their different departments. The research determined the ROI to the state for each dollar innovated.
We in-person interviewed and phone surveyed department managers, created surveys, and used economic data to perform our ROI analyses. We presented the organization with an Executive Summary, analysis of employer and employee non-responses, satisfaction, service, and summary and recommendations.
Credibility in Marketing Material: A Juice Company Investigates the Health Correlates of Their Product
A juice company was seeking marketing material, purporting that their juice was related to the reduction in stress and tension, an increase in sexual desire, mood, general health, and energy level. (Makes you want to buy a case!)
We conducted an in-depth examination of the placebo controlled double-blind dosage study research design, analyzed the correlates between the amount of juice consumed with age and gender, and used bonferroni correction on the alpha levels to account for the numerous correlations. In fact, there were numerous statistically significant relationships, and the company used this information in their marketing material with integrity and credibility.