Market Simulation or Marketing Mix Modeling:

I hear this question a lot. Usually it comes from people in the marketing space, who need to answer questions about the effectiveness of their marketing campaigns and/or use analytics to build their go-to-market plan. Whether its appropriate to use simulation or a marketing mix modeling (or both) depends on a few qualifying questions:

What is the objective?
If you need to evaluate the past, marketing mix modeling is the way to go. The reason is that MMM is a statistical technique that can only derive insights from past data.  But if you need to forecast the future, simulation is the way to go.  A simulation approach allows you to test and explore the impact of different strategies on the future. 

What level of granularity do you need?
If you are interested in high-level channels, MMM is the approach that will give you the biggest bang for the buck. If you need to go to a more granular level (e.g., instead of TV at a high level, you can see results by station, time-of-day, and campaign level). And if you need to include anything outside of paid media and promotions – for example, word-of-mouth, product design, distribution, creative execution, and messaging – simulation is the only way forward.

How far do you want to forecast?
MMM is fine if you are interested in making changes to your next-quarter strategy. The methodology teases out the short-term effects, so as long as nothing major changes, it is the most efficient way to get an accurate read on what to do next. If you are interested in long term effects – 5-10 year horizons – because your category has a long decision cycle (like automotive or insurance) or because brand equity is an important piece of the decision-making (like luxury goods and telecom), simulation will help forecast the likely evolution of the market.

What metrics are you interested in?
If you need to evaluate your ROI in terms of sales, MMM may help, as that is the metric it reports. If you need to think about consumer perceptions, their likelihood to recommend your product, and their consideration in addition to sales, simulation is the way to go. Similarly, if you want to track results for the market as a total aggregate, MMM will help. If you need to track results for different consumer groups and segments, you will need simulation.

What data do you have available?
To do MMM you'll need marketing spend by channel and sales over time to determine  each channel's contributions to sales. The simulation approach uses these but it incorporates that data at a more granular level: by consumer segment, by competitor, and by product attribute. It also incorporates information that is not used in MMM such as: reach and ratings, consumer surveys, creative emphasis and quality, investment in the product, and social tracking.

What is your business question?
Whereas MMM is a historical evaluation of effectiveness of marketing, simulations explore what-if scenarios. MMM primarily addresses a single question: how much did each marketing channel contribute to sales? A simulation, on the other hand can answer a host of questions:

• What if consumer preferences change in the market?
• What if there is a viral event or a public relations disruption in the market?
• What if we launch a new product?
• What if a new competing product comes on the market?
• What if we change our messaging?
• What if we switch our focus to a different consumer segment?
  
  

Summary
MMM is a technique to help decision-makers evaluate the past: what worked and what did not. Simulation helps to test ideas about the future. Both approaches have their pros and cons and sometimes you may need to combine both to decide your go-to-market strategy.