According to a 2016 report, while 84% of marketers opine that “identifying users, personalizing messaging, and measuring impact are ‘very important to growth,’ only 10%-14% are able to deliver in these areas. [1]” A time-tested analytical solution that many marketers turn to when gauging marketing effectiveness and planning budgets, is Market Mix Modelling (MMM). Over the last few years, MMM techniques have evolved with the growth of digital marketing. A recent Statista study on digital shoppers revealed that, “In 2016, 209.6 million U.S. people were online shoppers and had browsed products, compared prices, or bought merchandise online at least once. These figures are projected to reach 230.5 million in 2021. [2]” In lieu of the same, traditional MMM methodologies are being revisited to make sure that the right data and the needed levels of data granularity are factored in.
Through this article, we will delve into the process of setting up a customized MMM solution and the vital importance it holds for companies looking to quantify the direct impact of marketing activities and the indirect impact of factors such as the halo effect, inter-channel interactions, long-term/brand equity, market movements, macroeconomic drivers, etc.
The fundamentals of MMM
The concept of Marketing Mix is said to have first been proposed in 1960 by marketing professors. Edmund Jerome McCarthy, a marketing professor, as the 4 Ps of Marketing concept – price, place, product, and promotion. This was considered revolutionary, as McCarthy chose not to define marketing, but rather focussed on improving it by deriving insights on human behavior from other fields like sociology and psychology. Decades later, this concept remains fundamental for more marketing processes to be built on. Let us take into consideration, the beverage firm, Coca-Cola, and understand how this concept is in play even today.
Place: The beverage has a global market presence spanning hundred years, and their product is such that anyone who wishes to quench their thirst with this beverage only has head to their nearest local store. The team has meticulously ensured that the product is available at all possible customer access points.
Today, technologies like locating tracking can help to leverage WiFi signals, mobile networks, etc., and curate the what, why, and when aspects of customer shopping behavior, and determine stores in close proximity to fulfil customers purchase needs, at the right time.
Price: Identifying the most marketable price is of vital importance – consider how 99 cents feels like a much more affordable rate than a dollar – even though there is only a cent difference between the two prices. That is the length to which marketing can alter consumer behavior. Keeping in mind the market conditions and consumer behavior, Coca-Cola has ensured that it always stays affordable.
With the help of AI, accurate price forecasting is more possible than ever. Instead of relying solely on expert-insight/intuition-driven pricing, companies can utilize historical and competitive data to identify the most marketable price.
Product: Without a well-thought-out product or service on offer, the strategies employed will yield very little results. Hence, Coca-Cola has established its market presence with a variety of products that can appeal to different kinds of consumers and keeps updating or revising its products based on market trends. As of July 2019, the company’s leadership stated that close to 25% of the company’s revenue comes from new or reformulated beverages – as compared to 15% in 2017 [3].
By using predictive analytics, companies can cater their product or service, in accordance with market trends and customer preferences, and keep revisiting these techniques to remain pertinent in an ever-evolving market.
Promotion: Creating buzz about the product via unique messaging that influences the customer to purchase the product, is an important aspect of marketing. And Coca-Cola has cemented its presence in the mind of consumers through multiple promotional tactics. From traditional ads, billboards, to also partnering up with TV shows and movies, they have done it all. Case in point: Coca-Cola brought back its 1985 product, New Coke, (which was unpopular during its release as it deviated from the traditional Coke recipe), as a limited edition. This initiative was in collaboration with the popular TV show ‘Stranger Things,’ given that the show was set in the same period when the product was originally released. [4]
Today, the marketplace is such that consumers are constantly bombarded with offers/perks/seasonal sales on different platforms (social media, mail, text messages). Instead of having a one-size-fits-all approach to promotions, companies can customize promotions to build long-standing customer relations; for instance, loyalty programs tailored to customers’ historical purchasing behavior.
As elucidated by these examples, the fundamental 4Ps have remained pertinent, as it identifies consumer behavior as the foundation for building marketing techniques. It has proved to be dynamic enough to build the technique of market mix modelling such that even four decades since its inception, the product remains conducive to evolution and changing in accordance with the evolution of the market landscape.
Formulating the right Market Mix Model
The right Market Mix Model can help to efficiently measure the success of marketing initiatives and the ROI they generate, provided that it is backed by quality data, expert analytics, and enables informed, data-driven marketer decision making.
Fast food giant McDonalds, for instance, considers marketing mix modelling as their ‘secret marketing sauce.’ Chris Graham, the head of global media accountability and sourcing, shared in a 2019 interview that the firm assessed their ROI from media investments to reach the most beneficial marketing mix, and are looking to continue employing their market mix modelling technique in the future too, and cited how in Australia, the ROI had almost doubled in approximately half a decade. [5]
However, identifying and deploying the right Market Mix Model is easier said than done. Here are some fundamental factors essential for a well-established Market Mix model:
Quality data: The efficient functioning of a market mix model heavily relies on the quality of data collected and fed into the model for eventual analysis. Without accurate data, marketing activities cannot be linked with surety to sales generation. Ensuring that historical company data is reliable and vetted, and correlating that with relevant market data and even third-party resources, ensures that the predictions made by the model are reliable. The data collated, would provide an in-depth analysis of the companies’ marketing activities on various platforms including TV, radio, point-of-sale data and much more.
Expert analytics: Once the data is sourced, expert analysts make sense of the raw data to derive insights that can empower necessary decision-making. Some pivotal marketing data types include customer-centric data, marketing metrics (cost of customer acquisition, churn rates, etc.), market research, sales, competitive intelligence, purchase transactions, customer feedback, and customer feedback. [6]
Ample Resources: The software, storage requirements and other resources needed to run the market mix model need to be set-up.
Internal Evangelization: When Volvo first built a market mix modelling with its media and advertising partner, in the initial days, other teams with the organisation reportedly likened what the analytics teams did to “smoke and mirrors or voodoo.” [7] Therefore, a scorecard was set up in place to track ongoing predictions and compare these with predictions that were measured based on results, to build confidence among employees.
Updating the model: As the market factors keep changing, the model too needs to be constantly updated. Sellers need to align themselves with contemporary market drivers, including social drivers such as population density and lifestyle, economic drivers like economic disparities and resource availability, and technology drivers like payment channels and new tech tools. [8] In all, the effectiveness of the channel is determined by how relevant it is to the ever-evolving market landscape.
Marketers + Model = Winning formula: While many marketers rely on insights to identify the most beneficial marketing strategies, just intuition is not sufficient anymore. A recent study observed marketers who used their insight to analyze last-click results when formulating their marketing budget [9]. The study showed that while “they were correct in allocating more to top-of-funnel actions, such as ads on affiliate websites,” their intuition “did not go far enough and left money on the table,” and was wrong on aspects such as retargeting. It showed that “MMM is not a complete substitute for expert judgment but it is becoming an indispensable supplement to it.” Therefore, keeping abreast with the times, it is important to drive marketer-intuition with relevant data backing.
How MathCo deployed an MMM solution
To understand how MMM can be efficiently employed, here is an insight into a solution devised by MathCo for a CPG giant.
One of the largest CPG companies across the globe was interested in quantifying the impact of their marketing efforts on sales in the food product category. They wanted to in turn leverage this knowledge to optimize their marketing budget and maximize the ROI.
Solution summary:
MathCo partnered with the CPG giant to perform an exhaustive & robust Market Mix Modelling (MMM) exercise to quantify the direct and indirect impact of marketing activities. By leveraging the data they had available on the impact of all marketing channels and campaigns (and their interactions and indirect effects), an optimizer was developed to get the best marketing mix/spends allocation to drive the best ROI across the product portfolio.
A simulator was then built and delivered to the business stakeholders, enabling them to perform dynamic what-if analysis with insights and deep-dives. This helped them plan their next year’s marketing budget & execution.
Creating the analytical tool:
A total of 23 different marketing channels, including,
- ATL – TV, Radio, Billboards, Newspapers, Roadshows, etc.
- BTL – Sampling, Sales Promo, Activations, Point-of-sale marketing, etc.
- Digital – Paid Social, Paid Display and Paid Search
- Trade Promotions – Price Support, Premiums, etc.
along with competitor price & spends, and macroeconomic factors, were considered for the analysis. The analysis considered marketing spends and sales for the last ~3 years.
An extensive and thorough data processing and wrangling exercise was done to create the Analytical Data Sets (ADS) at a weekly level from all the relevant data sources.
Time-series interpolation techniques were applied for a few data sources where the required time granularity (weekly) was not available.
The Impact of marketing, promotions, brand equity, and market changes on sales was measured. The interplay of channels, halo effect, long & short-term impact of marketing on sales were analyzed in detail. Finally, the opportunities to tune marketing budget allocation to maximize return on investments were identified and recommended.
An exhaustive front-end tool was designed encompassing all the outputs from the MMM models, along with a what-if analyser which lets the end user play out various spends scenarios and dynamically visualize the impact.
Impact:
The resulting tool helped the business
- Understand their historical effectiveness of spends across various marketing channels.
- Plan and allocate budgets for different marketing channels for the upcoming year.
- Provide users an option to change spends of different marketing channels and know their impact on sales of all the products in the category using the simulator tool.
Market Mix Modelling vs A/B Tests vs Multi-Touch Attribution
In some scenarios, deploying just an MMM tool might not prove sufficient. The last few years, quite a few different techniques of marketing have also come to the fore, and two methods have particularly stood out – A/B testing and Multi-Touch Attribution (MTA). Here’s a look at what these different methods indicate, and an analysis of their advantages and limitations as compared to MMM:
Given that all these techniques provide value, most organizations employ them in a collaborative manner – especially MMM and MTA – to provide optimal, accurate recommendations for marketing activities to generate optimum ROI.
MMM: A summary of the good, the challenging and the upcoming
Global ad spend is touted to reach $605 billion by the end of the year – a 4.2% increase from 2019, and businesses in the US are touted to spend close to $110 billion on digital advertising, i.e, a sum more than TV and print ads combined. [10] And as the world changes, the marketing mix also changes.
Market Mix Modelling’s presence has been constant in the last thirty years, but in no way has it been stagnant. A well-executed Market Mix Model, therefore, can accurately link marketing activities to sales and provide real-time estimates of the impact of past and future marketing activities, and if not implemented well it can prove time-consuming and costly for the client. But with more and more innovations and contemporary approaches building on the traditional style, the future promises easily adoptable MMM solutions that evolve with the latest trends in marketing.
Bibliography
[1] Gilliland, Nikki. “10 Important Stats from Econsultancy’s 2016 Research.” Econsultancy, January 3, 2017. https://econsultancy.com/10-important-stats-from-econsultancy-s-2016-research/.
[2] Coppola, Daniela. “Number of Digital Shoppers in the United States from 2016 to 2021.” Statista. Accessed November 2, 2020. http://www.statista.com/statistics/183755/number-of-us-internet-shoppers-since-2009/.
[3] Lucas, Amelia. “Coca-Cola Raises Revenue Forecast after Earnings Beat, Sending Shares Higher.” CNBC, July 23, 2019. http://www.cnbc.com/2019/07/23/coca-cola-earnings-q2-2019.html.
[4] Winchel, Beki. “Coca-Cola Embraces Past Failure with New Coke ‘Relaunch.’” PR Daily, May 23, 2019. http://www.prdaily.com/coca-cola-embraces-past-failure-with-new-coke-relaunch/.
[5] Marketing Mag. “How McDonald’s Does Mix Market Modelling – the Secret Marketing Sauce.” Marketing Mag, November 13, 2019. https://www.marketingmag.com.au/tech-data/how-mcdonalds-does-mix-market-modelling-the-secret-marketing-sauce/.
[6] Spacey, John. “9 Types of Marketing Data.” Simplicable, August 25, 2017. https://simplicable.com/new/marketing-data.
[7] Sullivan, Laurie. “How Volvo Uses Marketing Mix Modeling to Build Brand Equity with Search.” How Volvo Uses Marketing Mix Modeling To Build Brand Equity With Search, December 13, 2019. http://www.mediapost.com/publications/article/344502/how-volvo-uses-marketing-mix-modeling-to-build-bra.html.
[8] Owyang, Jeremiah. “The Three Market Drivers: Causes for the Collaborative Economy.” Jeremiah Owyang, May 9, 2013. https://web-strategist.com/blog/2013/05/09/the-three-market-drivers-causes-for-the-collaborative-economy/.
[9] Gray, Kevin, and Koen Pauwels. “Using Marketing Mix Modeling to Improve Marketing Decisions and Romi: Article.” Articles | Quirks.com, June 5, 2015. http://www.quirks.com/articles/using-marketing-mix-modeling-to-improve-marketing-decisions-and-romi.
[10] Edwards, Mandy. “30 Marketing Statistics to Consider for 2020 – Business 2 Community.” Business 2 Community, April 28, 2020. https://www.business2community.com/marketing/30-marketing-statistics-to-consider-for-2020-02305426.