How to Fix Your Marketing Mix Model

NEW! Listen to article

A research paper I recently came across, " The Evaluation of Marketing Mix Elements: A Case Study*," highlighted some potential gaps that influence the effectiveness of marketing strategies—and, I thought, apply particularly well to our post-COVID world.

When I looked at the authors' list of what is typically not considered in most marketing mix models, these were the big misses that stood out:

The model does not consider customer behavior but is internally oriented.
The model regards customers as passive; it does not allow interaction and cannot capture relationships.

Most of us who have had the experience of reviewing marketing mix models, and consequently their predictions and forecasts, have always known there were gaps.

For example, for years I reviewed models at our company knowing that more than 50% of our sales and marketing efforts (those happening via various channels) were not included in the data that was compiled in the model.

That always bothered me, and I am still convinced it led to inaccuracies related to attribution, ROAS, and predictive business ROI on incremental marketing spend.

That feeling of unease is not an anomaly. My conversations with many marketing leaders have validated that the concern is more common than I imagined.

What has changed with customers?

Even in the B2B space, changes in consumer sentiment can have a huge impact; after all, we are fundamentally marketing to and selling to people—and businesspeople are also consumers.

Furthermore, companies as a whole need to respond to changing consumer sentiment. How we market to them can either assist them to do that or force them to adjust everything we do to make it a fit.

We are now living in a world that has been turned upside down by COVID, and we cannot imagine relying on a model that does not consider consumer behavior. I cannot comprehend considering customers as passive at a time when "engagement" and "community" have been identified as vital by most organizations.

We need not only more clarity about what consumers feel but also a true grasp of what has changed because of a worldwide pandemic that lasted two years. We cannot assume that nothing has changed. Understanding customer behaviors is now a vital component of post-COVID marketing analysis. The factors related to consumers are more important—they have become more influential—post-pandemic.

Those same issues and questions are valid for companies to ask about their customers, channels, resellers, and partners in the B2B space: What has changed? What new challenges are affecting our network that may not have existed before?

Why are there greater inaccuracies now?

Consumers learned a lot during the crisis. What they once merely preferred from the companies they do business with was transformed into what they now expect and require from those companies.

They learned that retailers, restaurants, and streaming services could all do better than they had been doing. Being isolated at home with an entire family online meant that Internet service providers had to improve speed and quality, and guess what? They did. Grocery stores needed to revamp their process and offer delivery and curbside pickup, and they did.

Through all those experiences, people learned that they could demand more and companies will respond and do more. Considering that new sentiment, how could any marketing model not find a way to consider customer behavior and relationships with those customers?

What about B2B? What did companies experience during the pandemic that may have changed their view and approach to business? Some received government assistance, and some are still dealing with labor shortages and supply chain issues. How do those experiences change partner expectations?

The answer may be different for each business, but the question at least needs to be explored.

Why do the wheels fall off?

Especially in large companies, internal political agendas amplify inherent inaccuracies in models. That's because those creating the models could have an ulterior motive that influences reports and dashboards. They want to provide important data, but at the same time they might want that data to reflect well on the Web team, the digital media team, the e-comm team, or others...

It's just human nature to interpret data in a favorable light.

Moreover, a model that truly reflects all channels can get difficult when you attempt to mix in data that measures sentiment and attitudes around the brand or the economy. Some data points are hard and defined, but often the additional points of measurement feel too squishy for the data analysts involved in the modeling process.

But for a marketing mix model to be accurate, it must be truly omnichannel—by definition.

In the B2B space, it is vital to understand how resellers or corporate clients feel about your brand: What is their level of loyalty, and do they see you as a vendor or a trusted partner?

Clearly, the answer to those questions requires more than merely looking at numbers on an Excel sheet; there must be a relationship with key people in those organizations. And, yes, that means some of your most important data will not come from a dashboard but from conversations.

How do we fix the models?

Regardless of how people feel about some of the data sources, whether they are skeptical or not, most companies want to receive "truth" from their marketing mix models.

To react and adjust marketing spend based on their data is only useful and effective, however, if the data is inclusive of every variable. Biases and feelings need to be put aside in order to get that "truth," and we need to uncover creative ways to add the "squishy" data back into the models and the algorithms.

Doing so is not easy, but if you are not willing to include them, then it shouldn't be called "omnichannel." "Some-channel" or "digital-channel" would be a more appropriate designation. Conversations at your call center, conversations that occur in brick-and-mortar locations, and feedback from third parties all need to find a way into the algorithm. To that end, speech analytics can be a game-changer.

It is not unusual to see conflicts between Web traffic data and call center data, for example. Is one right and the other wrong? Probably not, but understanding that conflicting information exists is helpful as we attempt to uncover the truth.

Most of us love clarity, and certainly the data analysts of the world almost worship it; however, many times clarity is not readily available. The stories we seek to understand about the customer journey are rarely black and white.

We encountered a similar challenge when attempting to create a valuation calculator for third-party sales and marketing channels, and our analysts struggled with what we were attempting to do. But we collaborated and aligned on the vision, and we eventually arrived at a working solution. Those types of internal partnerships and collaborations are needed to solve this challenge in any organization. To succeed, those managing the model should be channel- and department-agnostic; there should be nothing to defend, and the information shared should be as objective as possible

The solution involves meaningful collaboration between groups that might see the world, and interpret the data, very differently.

*Authored by Thabit Hassan Thabit (Ninevah University) and Manaf Raewf (Cihan University); published in the International Journal of Social Sciences & Educational Studies, March 2018. The list of shortcomings referenced is a modified version taken from " Revision: Reviewing the Marketing Mix," a post by Lee Kennedy.