Crowning a Champion
In this blog I discuss two of the most important differences between statistical marketing mix modeling and that of agent-based modeling. Building a better mousetrap is a key element to winning in the marketplace. All marketers continuously strive to find that edge to grow their market shares. With ProRelevant’s MarketSim, we also strive to bring you the best capabilities to make certain your investments are delivering the best value for your business. Here are two more critical differences between statistical marketing mix modeling and the new agent-based modeling using ProRelevant’s MarketSim that can help marketers improve their ability to bring that better mousetrap to their consumers.
Media centric vs. consumer centric
Statistical Marketing Mix Modeling is primarily designed to model the impact of media. It looks partially at pricing, but typically does so in a linear fashion. That is the typical S-curve of price elasticity is replaced by a straight line. Additionally, because it doesn’t model at the SKU level, it glosses over many details at the SKU level that can deliver highly valuable cross-media and cross-marketing insights.
MarketSim, because it models at the consumer level, through agents, it looks at the price paid in each of the distribution channels, all the products (SKUs) and brands offered as well as all actions in the category, delivers a very robust model of the actions in the category.
Uni-dimensional vs. multi-dimensional optimization
MarketSim was designed and developed based on many years of work with Proctor and Gamble. It incorporates many of the critical thinking of how consumers make purchase choices. In particular, it includes the only incorporation of Kahneman’s critical consumer behavior theories into the heart of the analysis.
In the past the marketing analytics was run primarily using statistical regression analysis. There were a few forms of this analysis, but generally it was a uni-dimensional solution. Each of the improvements in one media channel or one distribution channel are simply added together as if they operate independently. Only if the synergy or halo elements were explicitly built into the model could the results reflect a multi-dimensional answer. This is typically not done because a statistical model can only go so far. It has limitations in the number of data feeds it can handle.
Unfortunately, we’ve found this type of answer to be inadequate at best and inaccurate at worst. We’ve found that if we analyze any and all combinations across the 4Ps through a full multi-dimensional analysis we can easily improve results over and above a uni-dimensional statistical solution by almost 100%.
There are many differences between statistical marketing mix modeling and agent-based modeling that can help marketers to deliver competitive advantage in the marketplace. ProRelevant MarketSim offers best in class capabilities to deliver this value to your brand.
Please join us for our webinar on this topic on July 30th.