Why you may always be coming second

18 June 2020

Marketing Strategy

Real-time marketing mix modeling: the way forward

Marketing Mix Modeling was first adopted by large consumer packaged goods companies in the late eighties and early nineties. This statistical technique shone a light on which parts of their marketing program were delivering the best results and which were the least effective in driving volume and revenue growth, in order to adjust their marketing programs accordingly. Today, the technique is widely used by companies across industries such as consumer packaged goods, retail, telecoms, financial services, travel and hospitality, automotive and many others - industries which traditionally spend heavily on marketing - in a bid to optimize their marketing spends and maximize Return on Investment (ROI).

Times Have Changed

Since the time Marketing Mix Modeling was first adopted, the business and marketing landscape has changed significantly and become more complex, more dynamic and more competitive. A multitude of  cutting-edge marketing and promotional channels and formats, frequently changing consumer preferences, the impact of external factors and increasing competitive activity means speedy and data-driven decision making is more critical than ever before to drive growth and success.

Marketing Mix Modeling, as it is delivered presently, has tremendous opportunity to evolve and improve to better address today’s marketing challenges.

But Marketing Mix Modeling Hasn’t (Much)

There are several limitations in the way Marketing Mix Modeling is done today. One is the sheer amount of time it takes from start to finish. Gathering, reviewing and ingesting data from numerous sources to building the right attribution models to running the models to generate insights and reports takes months. Typically the insights generated are based on data that is 8 to 12 months old. In today’s dynamic environment, it is likely that during this time many of the marketplace and external factors change significantly so that the insights end up being merely “rear-view looking” i.e. they may help in diagnosing what worked and what didn’t last year or in the past, but are not particularly useful for meaningfully making future marketing spend decisions and taking corrective actions.

A second limitation is that businesses are completely dependent on specialists such as large consulting firms or analytics solution providers for every step of the process - collecting and inputting data, creating models, running scenarios - making it difficult to generate insights quickly and on demand.

Getting to Real-Time

What marketers need today are Actionable and “forward-looking” insights which can be used to make quick decisions on how to tweak their marketing tactics and campaigns for maximum impact. For insights to be truly actionable, they need to be Real-Time (or near real-time, to be more precise) i.e. based on the latest data that is no more than a few weeks or at best a month old.

To achieve this, analytics providers must adopt what I call an Always-On approach to Marketing Mix Modeling. An always-on approach enables Continuous Marketing Effectiveness Measurement by building all or most of the following best practices and features into their solution.

  • Collecting the latest data continuously and automating data ingestion
  • Leveraging Artificial intelligence (AI) and Machine Learning (ML) for quality control to identify incorrect data and capture outliers
  • Updating models quickly and regularly based on the latest data
  • Automating reporting and insight generation using technologies such as Natural Language Processing (NLP) and Natural Language Generation (NLG)
  • And providing businesses with in-house, transparent, “do-it-yourself” access to Marketing Mix Modeling through intuitive interfaces and easy-to-use tools

With such an always-on solution, marketers would be able to literally log in to the system, evaluate recent media or promotional campaigns by rapidly updating models based on the most current data, and generate real-time insights on demand.

And How It Actually Helps 

Let’s look at a common business situation to understand how an always-on Marketing Mix Modeling solution delivering real-time insights puts more power at a business’s fingertips. It’s the start of April. Green Foods’ Q1 2019 numbers are in and sales are below forecast. The performance team logs in to the system, confirms that the latest Q1 data till 31st March is already updated in the system and runs a set of analyses to help determine what they need to do to achieve their targets for the year.  

  • Variance Analysis: They generate Due-To reports to diagnose reasons for the variance in sales. This helps them accurately determine how much various marketing spends contributed (or did not contribute) to Q1 sales.
  • Campaign and Promotions Analysis: They evaluate their Q1 marketing and promotional campaigns to determine which ones worked best and which ones didn’t deliver returns.
  • Simulation Analysis: They run several simulations on the marketing budget, varying different spends to evaluate the impact on sales and revenues.
  • Optimization Analysis: They run an optimization analysis that recommends the optimal marketing allocations to maximize sales and revenues within their budget constraints.
  • Marketing Budget Reallocations: Using the real-time insights gained, they re-assign their remaining marketing budget for Q2 through Q4 across products, markets and marketing programs.
  • Real-Time Performance Tracking & Gap Analysis: And they repeat the process again in a month to track actual performance versus forecast in real-time and take ongoing corrective actions to bridge any gaps and achieve their targets. 

Analytics providers must move to such an approach to deliver true value. Real-time insights from Marketing Mix Modeling give companies the tools and enablers to transform how they manage their business and marketing planning, and helps put them on the path to growth and success.                                 

Tags: Marketing Mix, Marketing Effectiveness, Marketing Mix Modeling, Marketing ROI, Digital Marketing, Artificial Intelligence