Skip to: Main Content / Navigation

  • Facebook
  • Twitter
  • LinkedIn
  • Add This

Why marketing mix modeling is more than just a ‘black box’



Article ID:
20140110
Published:
January 2014, page 48

Article Abstract

The author tackles and debunks seven myths about marketing mix modeling and explores the factors that have generated so much conversation about the methodology.

Mixed signals

Editor's note: Nancy Smith is CEO of Analytic Partners, New York.

Right now, marketers are experiencing a perfect storm of new, digital media channels, the buildup of large volumes of transactional and sales data (aka big data) and increasing competitive pressures in the marketplace. In this ever-changing business environment, marketers are increasingly being challenged to apply deep insights and analysis to maximize the effectiveness of each marketing dollar spent. Through this trend toward increased accountability, marketing mix modeling has risen to the top as a methodology to allocate marketing dollars to achieve maximum marketing ROI.

The rise of big data has created additional pressure on marketers. They are being called on by their management teams to generate actionable insights from the vast libraries of transactional data that are being generated and collected daily. They are seeking tangible solutions to their big data challenges, and for many, marketing mix modeling offers a tried-and-tested solution.

So what is marketing mix modeling? Marketing mix modeling leverages advanced econometric modeling to help organizations understand the drivers of their business performance. These types of analyses take into account a wide variety of detailed and granular data to estimate the value that marketing and other business drivers deliver for a brand or organization. Marketing mix modeling is used to help companies improve their sales performance and marketing ROI and can be leveraged to run simulations, scenarios and forecasts for future business performance.

As one might expect, the newly intensified interest in marketing mix has motivated skeptics and the curious alike to express reservations and to ask hard questions. Outlined below are some of the misconceptions about marketing mix modeling we’ve heard over the last 13 years, as well as the facts to prove them wrong.

Myth 1: Marketing mix is a black box

Because marketing mix modeling involves advanced statistical analysis of vast volumes of data, it can appear to be obscure process to non-practitioners. Some have accused marketing mix of being a black box, with inputs going in and results coming out with no transparency to the process. Without a clear understanding, how can one know if the model results are accurate?

Though it is certainly a complex methodology, marketing mix is not a black box. It is a proven analytic approach that incorporates science with business acumen to drive actionable insights. Marketing mix has been leveraged successfully for decades now in the hypercompetitive markets for consumer packaged goods. The core statistics and methodology behind marketing mix are the same tools used by the medical community to track disease rates and to test the efficacy of new medications.

The practitioners of marketing mix understand that the complexity of the models makes it all the more critical to drive transparency and understanding through training and knowledge-sharing with their clients. Organizations with the strongest and best approaches to analytics will provide clear outlines, milestones and model performance benchmarks to their key stakeholders.

Additionally, marketing mix is not a one-size-fits-all approach. Every business is unique and the data available to each business is different and of variable quality. Therefore, each engagement requires a customized approach to analyzing and modeling the data. The approach chosen is based on many factors, including the data available, type of industry, how many business lines are involved, competitive landscape, etc.

Myth 2: Marketing mix treats all impacts on business performance the same

Some observers view marketing mix as a cookie-cutter approach, treating all sales effects the same without factoring in the intricacies of channels, campaigns, promotions, pricing and other unique factors.

In fact, best-in-class marketing mix models are customized to consider the unique factors and consumer response to each activity within the business. Consumers are constantly connected, and being exposed to different messages and offers, so a cookie-cutter approach does not work. Therefore, models are developed to account for each different marketing stimuli. The model can measure promotions, loss, media, search and display, etc., and take into account their contributed influence on sales.

For example, while promotions typically impact sales immediately, there is a potentially negative longer-term effect on sales as consumers tend to stock up or purchase ahead of when they would have ordinarily. On the other hand, while media may not generate immediate sales boosts, it tends to impact sales positively in the longer term. A quality marketing mix methodology will isolate these effects and will assign quantifiable measures to these factors so that marketers can make informed strategic and tactical decisions about their marketing investments.

Myth 3: Marketing mix does not include complex digital channels, such as search

While it is true that marketing mix was originally pioneered by CPG companies in a non-digital setting, the methodology has not remained static. Models have evolved and become more sophisticated in measuring interactions across all channels – paid, earned and owned. As the consumer becomes increasingly connected, marketing mix has expanded in scope to help understand all interactions, across all customer stages and objectives.

A recent evaluation of marketing mix practitioners has found that 87 percent of respondents include digital channels in their marketing mix models. Of course, the quality of digital analysis will vary by practitioner. However, this affirms that advanced analytic approaches are also evolving to include digital data (social, e-mail, search, display) and marketing mix is an appropriate tool to help marketers determine the right investments in new digital channels.

Myth 4: Marketing mix is biased to promotions and does not capture the true effects of advertising

Some critics contend that marketing mix tends to overvalue the contribution and investments in promotions over other advertising and marketing efforts. The truth is that a quality marketing mix analysis will not have bias toward any particular activity. Instead, it holistically measures all impacts, accurately estimating each one’s effect on performance.

While there is no marketing mix bias toward promotions, some businesses, especially CPG, tend to be drawn into a promotion-heavy strategy. When promotions are applied over the long term, it is known as deep discounting. While these discounts may help the business and retailer in the short term, neither party flourishes in the end. A deep-discounting strategy conditions consumers to buy on deal, leading to a devaluation of the brand and the category and eroding margins for both the manufacturer and retailer.

Marketing mix is certainly not the cause of this situation but it can be a part of the solution. The models identify opportunities for the manufacturer and retailer. High-return promotional events can be prioritized in a way that does not erode margins in the longer term.

Myth 5: Marketing mix cannot attribute sales to a specific customer segment

Some observers question whether marketing mix treats all consumers the same and therefore does not have the sophistication to inform marketing investments by consumer segment. While marketing mix has been a guide to help marketers plan future budgets, it is being leveraged more and more to monitor and predict behavior of customers and identify the best way firms can interact with specific customer segment groups.

Segmenting a marketing mix analysis to understand a specific customer group’s behaviors allows for firms to customize experiences, make relevant offers and customize communication strategies, based on what the model uncovers about certain behaviors across different touchpoints.

Segmenting model results by customer segment can really help shape a marketing strategy. For example, within the hospitality sector, business and leisure travelers have a meaningfully different response to marketing, particularly for e-mail and TV. This learning was leveraged to shape strategies and budgets, particularly within key seasons of the year. This is a great example of where going deeper on the analysis provided actionable insights that supported budget planning and allocation across key customer segments.

Myth 6: Marketing mix lacks real-time value

During this time of increased reliance and availability of real-time information, marketers are expected to be able to instantly identify and respond to events and the demands of their customers. Marketing mix modeling can deliver just that: It provides real-time insights and a robust tool, allowing businesses to leverage the data to make decisions.

Business performance assessments in real-time can be leveraged to evaluate new campaigns, understand the impact of new competitors and assess pricing actions or changes in promotional strategy. Additionally, real-time insights allow for optimization of spending, such as determining the minimum budget to meet the business goals (e.g., specific share or sales objectives), increase profit given the same spending or reallocate for optimal spend given real world business constraints.

Real-time insights provided through marketing mix modeling enable marketers to review plans and forecasts with the latest results, correct changes in the course of business and integrate other research and learnings to fine-tune their plans.

Leading marketing mix practitioners will provide software tools that provide up-to-date information to allow marketing decisions to be made in real time. Furthermore, some practitioners provide simulation and forecasting tools such that marketers can simulate potential marketing actions in a virtual environment to leverage the model results in a forward-looking manner.

Myth 7: Marketing mix cannot tell you what you haven’t done before

Some observers express concern that marketing mix can only show you what has happened in the past and therefore cannot inform you about new activities that you may be contemplating. But marketing mix demonstrates its true value and best use in forecasting and forward planning by using information from the past to help predict the future and align business practices to accommodate and adapt to the evolving marketplace.

From my experience, when working with clients to apply modeling insights to their business, we actually spend significantly more time looking forward than backward. The models represent a great deal of knowledge about the business and, when combined with other research and business intelligence, we can support business planning and deliver robust forecasts.

With the growth of data available and ability to measure it, organizations can have more insight into the performance of their business lines and behavior of customers, allowing them to leverage the rich data available to accurately predict future investments.

Holistic picture

With the myths dispelled, what are we to conclude? Marketing mix modeling can paint a holistic picture of an organization’s business efforts. Organizations seeking to improve the ROI of their marketing efforts are increasingly leveraging marketing mix to improve their results. The practice is evolving to meet the changing media channels, data sources and competitive challenges of our hyperconnected world.

Again, the increasing use of marketing mix has motivated critics to question the practice, which is healthy for the marketplace. No one should accept any marketing analytics on blind faith. The best practice in marketing mix involves a transparent approach and a willingness to invest the time and effort to educate clients on how the models function.

Comment on this article

comments powered by Disqus

Related Glossary Terms

Search for more...

Related Events

INTERNATIONAL SHOPPER INSIGHTS IN ACTION
November 3-5, 2014
IIR will hold its international shopper insights in action event on November 3-5 at the Sheraton Grand in Edinburgh Scotland.
RIVA COURSE 241: QUALITATIVE ANALYSIS AND REPORTING
November 3-5, 2014
RIVA Training Institute will hold a course, themed 'Qualitative Analysis and Reporting,' on November 3-5 in Rockville, Md.

View more Related Events...

Related Articles

There are 2344 articles in our archive related to this topic. Below are 5 selected at random and available to all users of the site.

Adding ‘prosumers’ to your groups can provide a creative boost
Under the right circumstances, including - rather than excluding - marketing professionals normally screened out of focus groups can lead to breakthroughs. Examples from British Airways and WD-40 are cited to illustrate how this approach has helped develop new products and services.
Data use: Appropriate use of regression in customer satisfaction analyses
In October, 1992 Quirk’s Marketing Research Review, Dr. William McLauchlan critiqued the use of multiple regression analysis to model customer satisfaction and asserted that self-stated importance is a superior approach. The author of this article disagrees with McLauchlan, and uses this article to respond to these critiques and explain why multiple regression analysis would be a more powerful choice of analysis technique.
Data Use: Evaluating paired comparisons, maximum difference and traditional ranking
The author examines a host of data interpretation issues related to the use of paired comparisons and maximum difference surveys.
Data Use: Thoughts on our overreliance on
Many researchers don’t fully understanding how to properly interpret the results of statistical testing, which makes them incorrectly interpret non-significant findings as not meaningful and place too much emphasis on findings that are statistically significant.
Is a longitudinal approach the secret to measuring an ad's ROI?
The author argues that interviewing the same consumers before a new ad campaign is launched and again after the campaign has been in-market for a period of time is an effective way to isolate the in-market impact of advertising.

See more articles on this topic

Related Suppliers: Research Companies from the SourceBook

Click on a category below to see firms that specialize in the following areas of research and/or industries

Specialties

Industries

Conduct a detailed search of the entire Researcher SourceBook directory

Related Discussion Topics

Hi Giovanni
10/17/2014 by Dohyun Kim
request
06/06/2014 by Monika Kunkowska
TURF excel-based simulator
04/17/2014 by Giovanni Olivieri
XLSTAT Turf
04/10/2014 by Felix Schaefer
TURF excel-based simulator
03/25/2014 by Werner Mueller

View More