Dear Dave –
I’m a marketing manager in a high-tech software company. Lately I’ve been thinking that my company might benefit from a marketing mix model to help us understand the relative value of our advertising vs. our field marketing programs vs. our channel programs. I think we could scrounge up enough data to make the analysis reasonably valid, but I have two concerns. First, I’m not certain how stable I can anticipate our industry being over the next few years, so I wonder how “predictive” my results might be from a mix model. Second, I’m not exactly working in the hotbed of analytical models. Most of the decisions around here are still very judgmental, and I’m afraid that introducing a sophisticated tool might actually work against me in some ways.
What is the collective experience with these types of models? What would you suggest given my situation?
Sarah in Texas
I’m a marketing manager in a high-tech software company. Lately I’ve been thinking that my company might benefit from a marketing mix model to help us understand the relative value of our advertising vs. our field marketing programs vs. our channel programs. I think we could scrounge up enough data to make the analysis reasonably valid, but I have two concerns. First, I’m not certain how stable I can anticipate our industry being over the next few years, so I wonder how “predictive” my results might be from a mix model. Second, I’m not exactly working in the hotbed of analytical models. Most of the decisions around here are still very judgmental, and I’m afraid that introducing a sophisticated tool might actually work against me in some ways.
What is the collective experience with these types of models? What would you suggest given my situation?
Sarah in Texas
Dear Sarah:
Marketing mix models and other regression based models may be good for telling you what has had an impact in the past on your sales, but are clearly based only on the available data. That said, it is a good starting point to know what has worked to what degree in the past, and then you can adjust from there. The fact that you say your industry is not stable may be an indication that relying solely on historical data would be highly dangerous. I must also add I have other concerns with the use of marketing mix models. My two biggest concerns are:
Marketing mix models and other regression based models may be good for telling you what has had an impact in the past on your sales, but are clearly based only on the available data. That said, it is a good starting point to know what has worked to what degree in the past, and then you can adjust from there. The fact that you say your industry is not stable may be an indication that relying solely on historical data would be highly dangerous. I must also add I have other concerns with the use of marketing mix models. My two biggest concerns are:
-
They generally only capture short-term effects and lump everything else into the baseline, where the longer term effects reside. This results in spending being channeled to short-term instruments, such as promotions, rather than investments that would be better for the long run.
-
Most marketing mix models have sales as the dependent variable and marketing spending as the independent variables. There are many other measures we customarily look at in marketing that are totally left out of the equations, such as, customer satisfaction, acquisition, number of customers, or share of wallet—just to name a few. The result is we do not get the diagnostic information that could be really useful for managing the business.
The second issue you mention is potential sophistication level. You should capitalize on the judgmental approach and wisdom that exists in the company. There is an approach which I would strongly recommend known as Decision Calculus which gives you a structured way to capture these judgments, share them across the organization, and see what the resource allocation implications are from this judgment. Taking the judgments and making them explicit in and of itself is a very useful process. If nothing else, this is where you should start.
Good luck.
Dave
The Question of the Week is your opportunity to ask Dave your most pressing questions relating to marketing measurement and effectiveness. Be sure to submit your question today.
