As digitalization is picking up pace, companies are emphasizing on trying, validating, and adapting to processes and technologies for data-driven business decisions. Within businesses, be it any department from sales and marketing to IT and HR operations, the focus on data has evolved in recent years. Since COVID-19, the demand for data-driven insights for optimizing the performance of decision-making processes and the utilization of budgets across teams has been a key focus and challenge.
This has also impacted the traditional allocation of budgets for the marketing domain. Strategies for crafting the optimal marketing budget allocation across the mix of marketing activities have pushed marketers to adopt the usage of Marketing Mix Models (which evaluate spending at three pillars: Pricing, Promotion, and Media decisions). These models help identify which marketing channels will be effective and efficient, may drive the highest returns, and optimize spend allocations to maximize business outcomes.
A bigger debate has emerged on the usage and consumption of Media Mix Models, which are essentially a subset of traditional marketing mix models. Brand managers and category managers are struggling to optimize their budgets, to have the right allocation for media activities. Allocating the media budget based on data-backed insights can help them both in the short term by driving sales, and in the long term by strengthening brand equity, brand awareness, and brand loyalty.
More often than not, marketers/brand managers/category managers have to deal with scenarios where they have multiple methodologies and multiple vendors running these programs (market mix, media mix, 360 mix models). And they struggle to choose the right model, right methodology, etc. The question that emerges here, is if all of these methods are conducted for the same reason, why are the results different, which one is to be believed, and how do we compile/compare the results of one with the other?
Well, I am going to present my view on how we can answer a few of these concerns with a certain level of confidence, to move towards a convergence of the results irrespective of the models or vendors involved. Especially in the context of running media mix models, and comparing the results to the marketing mix outputs.
We model all the variables which impact sales at the most granular level. Starting with UPC/SKU level promotions and pricing changes which are run weekly at the store level, to the media activities which are done at a category or manufacturer or brand or sub-brand, or SKU level.
Media Mix Models
Usually in media mix models, the emphasis is on modeling the media variables. And more often than not, the impact of the price and promotional variables are neither measured nor considered. Even when they are considered, it is only at a level that may not capture all the attributions as they should have; as these individual promotions on individual SKUs behave very differently when they are on promotion. The behavior is dependent on multiple other factors such as category, brand, region, and the store in which the product is on promotion, etc.
Hence, what you have is an over-estimated media variable in terms of attribution which is not desirable to any marketer, or category/brand manager. Thus, we face the issue of results being different despite using similar models and variables.
Now, the media mix model has multilevel reporting (depending on the business question which has to be answered):
|Total Non- Digital (print)||
|Example of bringing the consumer lens (by defining and adding variables like below in models)|
|Customer segment (Using POS estimates OR Analyzing Shopper cart/basket)||
|Result: Media deep-dive to understand attribution, and optimize the budget allocation of media activities based on Category > Brand > Individual media type/vehicle/channel etc.|
Even though you are just running a media mix model you will have to understand the impact of the promotion and pricing variables at the store level to ensure that the promotion/pricing impacts are not inflating (overestimating) the media contributions to the media models.
Since you are not running a complete marketing mix model, you will choose not to investigate and report promotion and pricing impacts at an individual SKU or promotion type level; but give a total contribution impact of promotional activities due to price changes over time.
Often you may have struggled to consolidate the findings from two models (run by two different vendors or maybe the same vendor but using a different definition of modeling). If you do not model the promotions, your media contributions will be overestimated in the weeks when promotions are happening, and will disrupt the total media contribution or ROI, depending on the variability in base promotion will impact the media contributions (i.e., base/dips & falls).
If we take the above approach, we will be able to not only able to integrate or disintegrate results from various models, but still be able to compare the models which use multiple methodologies or are built by multiple vendors and will be able to generate more or less similar results in any independent effort for model/vendor selection.
Using these models for marketing mix optimization is slowly but surely becoming a vital aspect of every business. They can be adopted by almost every business function, and not be limited to marketing managers and initiatives. Business personas from across teams can contribute to the organization as a whole, by customizing their models and catering to specific consumer requirements.
Every business function wants to ensure that they are delivering premium customer experiences, and this is driving the need for customized marketing mix models. From R&D and manufacturing units to sales professionals and logistics managers, MMM helps solve unique problems and deliver value to customers. This in turn is propelling analytics vendors such as Course5 Intelligence, to develop and deliver personalized solutions for each business function.
Join us in the following blog, for a guided tour of how Course5 drives the adoption of MMM, and customizes the solution for key account managers and stakeholders; empowering them to answer customer pain points and deliver exceptional customer experiences.
Sign up to get the latest perspectives on analytics, insights, and AI.