Q&A with Dr. Dominique Hanssens on Marketing Mix Modeling
The marketing landscape continues to evolve, making it increasingly difficult for brands to accurately measure which aspects of their marketing investments across various tactics and channels are generating positive returns (ROI). To be successful in today’s market, brands need a holistic strategy to effectively predict and measure all the factors that influence marketing optimization.
In the Q&A below, GBK Advisor Dr. Dominique (Mike) Hanssens, Distinguished Research Professor of Marketing at the UCLA Anderson School of Management, shares his thoughts on how Marketing Mix Modeling (MMM) is evolving, common challenges companies face with marketing attribution and measurement, as well as emerging trends and opportunities in the space.
Q: What is marketing mix modeling and what are the benefits for brands?
We’re all familiar with the concept of the marketing mix. The concept was actually first introduced in the 1950s when Professor Jerome McCarthy wrote about the four Ps of marketing – price, promotion, product, and place (distribution) that drive consumer demand. If you want to explain variations in the performance of your brand, you not only have to look at product and price but also your distribution and promotional efforts.
The goal of marketing mix modeling is to determine how much impact was generated by each factor across the four Ps and to forecast what future impact can be created by altering or optimizing your marketing mix. Marketing mix models start by estimating top-line effects of marketing (e.g., impact on sales revenue), using econometric methods, and then derive profit and ROI effects by incorporating marketing costs and profit margins in the equation.
Many companies today solely used marketing mix modeling to measure the ROI of their advertising or media investments, when in fact, MMM can be used to forecast and improve ROI across functional areas. Ideally, companies should be using marketing mix modeling to provide insights to inform decision-making and effective budget allocation across marketing, sales, pricing, operations, and product development.
Q: How has marketing mix modeling evolved in the past 10 years? And what emerging trends do brands need to be aware of?
The most obvious areas of evolution in Marketing Mix Modeling in the past 10 years has been broad acceptance across business sectors and nonprofit sectors that traditionally were unaware of it. Marketing mix modeling was first applied by the CPG industry with companies such as Procter & Gamble. Today we see MMM used across nearly every industry, including both B2C and B2B companies. With increasing pressure to drive top-line growth, brands need to constantly seek out better data to optimize their marketing spend and generate positive ROI.
Advances in marketing data, computing, and modeling technology have influenced how mix modeling has evolved, especially with the rise of digital marketing. Consumer privacy is another area that directly impacts how brands track engagement, leading many companies to rethink their marketing and communications mix.
Q: What are the most common challenges or gaps brands face when it comes to effective marketing measurement and attribution?
One of the biggest challenges for brands is focusing on the right data. We often assume that data are plentiful and available at your fingertips, but in many cases, it's not. The starting point for any marketing mix project is to understand what data you need to collect or can be made available to inform your model. It’s also important to use your existing data to get started. Companies can get stuck in analysis paralysis or become overly focused on waiting for complete or perfect data, rather than moving forward based on the first- and third-party data they have.
“another challenge is that brands are hyper-focused on clicks or digital ad performance to measure success, when there are many other areas that impact business performance or why customers choose a brand. ”
In recent years, another challenge that has developed is that brands are hyper-focused on clicks or digital ad performance to measure success, when there are many other areas that impact business performance or why customers choose a brand. Yet, many marketing mix models are based almost entirely on media investments, so they really should be called “media mix models.” Marketing mix models should account for the overall go-to-market strategy and investment, including sales, price, promotions, etc.
Another challenge is having the right team and resources in place to experiment and learn through modeling. To fully take advantage of mix modeling, we not only need marketing data and statistical software, we also need to continue to educate practitioners that marketing mix modeling needs to be more than comparing, say, the productivity of Google vs. Meta ads.
Q: Are there any other factors that brands need to think about in terms of just customer attention? In addition to measuring the effectiveness of different media by channel, there are other factors that can impact consumer attention or that brands need to consider to accurately measure ROI.
Absolutely, and that’s why a well-defined brand and marketing strategy is so important. There are a wide range of key business questions across the three Cs (Company, Customers, and Competition) that you need to answer to inform your strategy. Who are your target customers and what are their needs? How do you optimize your product portfolio and operations by channel to best reach those customers? And who is your competition, what are their offerings, and how effective is their marketing relative to yours?
Before you launch a product or a new marketing campaign, you need to answer these questions and many others based on the needs of your target customers to effectively define your brand strategy. For example, if your CEO comes to you with an ask to increase sales, there are several avenues available to achieve this. Are you going to lower your price or keep pricing the same? Sales to whom – anyone (which is all too often the reaction), or sales to target customers with higher CLV? Are there changes you need to make to your distribution approach, or will the desired sales be entirely driven by your promotional mix? To answer these questions, you need to conduct upfront segmentation research to define your brand and product positioning, which provides the foundation for your allocations across the marketing mix.
Q: Traditionally, MMM has focused on providing marketers with a top-down view of their marketing investments by channel, but there are a number of other factors that can influence outcomes with customers (from consideration and purchase to loyalty). What steps do brands need to take to develop a holistic strategy to measure marketing ROI and map them to specific outcomes?
It starts with interviewing the key decision-makers and asking them, what are the drivers of your business? What factors cause your business performance to go up or down from month to month, or across markets? Once you have a good overview of the business across functional areas, the next step is to talk to customers.
The voice of the customer and the voice of the executives should match each other in terms of what's important. But often that’s not the case. For example, decision-makers at a company may say ‘we're at the mercy of the economy. When the economy gets better, we do better, and vice versa. But more is needed to understand both the macroeconomic environment and the key value drivers for customers.
“Once you have a good model of how the business works, you need to continuously update that model with data and adjust as new information becomes available.”
What exactly is it about the economy that affects the company? The number of potential customers in the market? The purchasing power of people? The current interest rates or unemployment rate? To effectively predict customer behavior, you need to take all of these factors into account to have a holistic interpretation of where to invest your marketing resources.
We often see a marketing mix model being presented to executives as a set of PowerPoint slides, without subsequent follow-up. But these are not static models. The best companies combine mix modeling insights to actually inform decision-making on a monthly or weekly basis across marketing, operations, and other areas and evolve the models with new data over time
Once you have a good model of how the business works, you need to continuously update that model with data and adjust as new information becomes available. As such, there is a close link between the firm’s database capabilities and the effective implementation of a marketing mix model. Depending on the industry and the scope of the marketing mix, this monitoring and updating can be done monthly, weekly, or even daily in some sectors. For example, if your customers are becoming more price-sensitive, a good mix model will quantify how consumers’ reactions to prices are evolving and what that implies for the execution of your marketing plans.
Q: With the shift to digital, brands have access to higher quality data to optimize spending. That said, not all data is created equal. How do brands separate the signal from the noise to identify the right data to apply to their model?
This is where the econometrics and statistics come in, as enabled in various computer software packages when used by knowledgeable practitioners. While executives can be expected to identify the key value drivers of their business, they need help to actually quantify the effects of these drivers. For example, if doubling the advertising budget is expected to increase sales, the key question is: by how much? A good marketing mix model will quantify and compare the relative power of each of the four P’s in generating consumer demand and brand revenue.
Fortunately, we're guided by a lot of good economic and psychological theory. We have a good idea of what many of the most critical variables are, but prioritization of those drivers is key. For example, the strength of the economy may be a driver. But how will you measure that? Based on interest rates or unemployment rates? To improve the accuracy of your model, some experiments on the data may be needed to determine which one of these two metrics is a stronger driver. That is an exercise you need to do up front before you finalize the marketing mix model. Again, the vast knowledge base in econometrics and statistics is utilized here, along with powerful software packages that implement these methods.
Q: As discussed earlier, many marketers place a lot of emphasis on the media components within a marketing mix model. And, in addition to measurement by channel, brands need to look at the effectiveness of their ad creative and message quality. How do marketers accurately assess both the impact of their media plan and creative quality when assessing the overall effectiveness of their campaign?
Understanding the effectiveness of your media spend by channel is relatively easy to do, so long as you have good data on spending and results. However, understanding the effectiveness of your ad creative and messaging is more complex.
Several years ago, I worked on a project where the CEO told the marketing team, we need to sell the product by pointing out what a good deal it is relative to the competition. He was adamant that we focus on a value versus quality message. So, we conducted research to test and compare two different sets of ad creative – with one creative campaign focusing on quality as the primary message and the other focusing heavily on value or price.
“By running a well-timed experiment, you are able to show the quantitative and qualitative differences between your creative and messaging, alongside your media plan. ”
What we found is that the value or price-themed campaign required twice the media spend of the quality-focused campaign in order to achieve the same impact. Thus, we were able to demonstrate to the CEO that we could deliver the same bang for the buck for half the cost by focusing on quality in the ad creative.
By running a well-timed experiment, you are able to show the quantitative and qualitative differences between your creative and messaging, alongside your media plan. More recently, computing advances in natural language processing allow us to diagnose differences in advertising content and to understand what is it about the creative that is drawing an appeal? Is it the humor? Is it the factoids? Is it the use of imagery versus words and so forth? In other words, we can now figure out what is it about creative that matters.
Q: That's a great example. It also speaks to another point which is companies can deteriorate and devalue their brand when they focus too heavily on price or cost savings as a differentiator.
Absolutely. In fact, research has shown that, among all marketing mix variables, one needs to be managed very carefully lest it results in negative long-term effects. I am referring to temporary price cuts, i.e. price promotions.
While offering sales promotion discounts tend to be effective in the short term, there is a balance when it comes to maintaining the value of your brand over time. In other words, frequent sales promotions will train your customers to only buy your product when it is on sale. This can become a trap of sorts for a company, especially there is pressure to meet certain near-term sales goals.
A good marketing mix model will quickly reveal such a scenario and show the trade-offs between marketing that promotes the brand versus marketing that only provides a short-term sales lift. With inflation increasing, there are also some questions brands need to answer related to pricing power. Do I have a strong enough brand to raise prices, given my supply chain costs are higher? How often can I run sales promotions without cannibalizing sales from my most valuable customers?
Q: What other types of attribution modeling or predictive analytics do brands need to ensure an accurate and consistent view of marketing ROI?
Attribution modeling is great because it gives you precise metrics on the impact of various digital marketing activities along the consumer journey that we didn’t have before. The challenge is that it is almost completely restricted to digital media, while the impact of other channels and marketing mix actions is harder to track beyond the point of sale. This of course is a bit different for B2B brands who typically have the capability of tracking activities outside of digital channels. Another challenge impacting attribution modeling is consumer privacy concerns that have motivated some providers to restrict the use of third-party “cookies”.
“When you have high levels of awareness, the last thing you want to do is lower prices. The better use of dollars is to focus on playing to your brand’s strengths.”
In terms of predictive analytics, there is a lot you can do to optimize your brand positioning and messaging based on consumer attitudinal studies. While a marketing mix model will help to inform your marketing plan across the four P’s, it doesn't tell you why people buy your brands. Through a combination of survey analysis and various metrics (brand perception, virality of messaging, consumer sentiment metrics) you can uncover the specific drivers or reasons people choose your product and how that compares across competitors.
I remember attending a presentation by a Coca-Cola executive. With a strong brand and high market share worldwide, a big focus for Coke is crafting messaging that helps elevate its brand with consumers. One avenue for achieving this is to make sure the brand is a part of conversations that are happening in local markets. So, marketing teams are monitoring various international markets for interesting conversations that are taking place about issues that the brand could become a part of. In other words, they try to ride the waves of consumer interest that are different regionally, even if these have little to do with soft drinks.
Coke is an interesting example of what highly established brands need to consider improving marketing effectiveness. When you have high levels of awareness, the last thing you want to do is lower prices. The better use of dollars is to focus on playing to your brand’s strengths, for example by engaging in conversations that increase the virality of your brand.