Finding an alternative to Marketing Mix Modeling and Marketing Attribution
Many times, we have heard the story. Revenue targets are missed. Analytics show marketing is “working” but it’s not translating to sales. The CFO asks to bring more rigor to the process. Another analytics tool is needed. There is a slight organizational groan. In a few short minutes, your Google search leads you to investigating marketing mix modeling (MMM). Marketing mix modeling is the de facto statistical model that attributes sales results to media spend. It usually requires an outside consultant who takes months to cobble together data and deliver a report. The best ones have an interface you can use to test new ideas, but the reality is it is based on months of old information. The problem is not really solved. The marketing mix model tells you what happened but struggles to tell you why and cannot tell you what to do in the future.
If you have started to think about replacing your marketing mix model or starting on a new one, alternatives now exist that make this once stalwart method no longer the go-to solution for marketers. Let’s take a look at the history of how marketing and modeling have changed to give you some insight on things to consider when picking a new alternative.
The context mattered.
Have you ever asked yourself what drives the color of professional football teams’ jerseys? One team wears dark colors and one always wears white. They had to do this because they broadcast in black and white. The jerseys were a requirement to make the technology work. Simple capabilities required simple inputs. In this era of simplicity, marketing mix modeling was born. Here is why it was great tool then:
- Marketing mix modeling was created when the number of media alternatives were few. Three TV stations, two local papers, some radio stations and maybe billboards if there were enough cars going by.
- It came into being before the CMO role was created. It was designed to support advertising spend at agencies rather than advise on strategy. DDB had one of the largest teams in Chicago in 1972.
- Tracking impressions was a simple exercise of understanding viewership of each media channel. There was no such thing as pixels, cookies or customer journeys.
- Statistics used to be all we had. It was done manually or in Excel, so obviously it took months to get answers and those answers were limited in nature. Now we have more advanced sciences like simulation, AI, machine learning, agent-based modeling and behavioral economics to build better models.
This simple world allowed a method like marketing mix modeling to give answers of value, but the world has changed.
The context changed.
Complaints about marketing mix modeling have arisen in the past view years. Many have said the models are wrong. I think a better way to think about it is that the context has changed. Marketing mix modeling is not wrong it is now incomplete and insufficiently designed to the task.
- Media and content are proliferating at a blistering pace. Picking the most effective tools is relative to the complex world that each brand exists in.
- The CMO role has gone well beyond advertising – creative ideas, pricing decisions, competitor actions, spend optimization, and ad budgets all must be managed simultaneously.
- Data and insights have exploded yet the analytics field and CMO are still confused. Not with any one insight but how all things work together. It is estimated that on average a CMO has 17 tools to look at each day.
- Marketing attribution has emerged and now marketing mix modeling and marketing attribution are being force fit together in an even more complicated modeling process.
- Data science teams have been wrongly asked to manage more than data science. Consultants deliver results to teams and the team must then reinterpret results for business users. These demands overwhelm both sides of the equation.
Complex environments, with proliferating insights, tools, and options can no longer be managed with the tools that link a network of agencies, consultants, and internal stakeholders together. The need for predictive analytics, speed, flexibility and real-time answers is where new solutions must emerge.
Build a better analytics process.
As you look for a better alternative to solve this problem, we would recommend thinking of the problem from another context. Instead of thinking of this as a media measurement or ROI problem, think of the issue as a matter of building an analytics process designed to enable learning and smarter strategic decision making.
- Find a more meaningful business question. Media does not by itself overcome churn, bad products, and tough competitors. Instead of asking, “What is a better way to allocate media?”, identify the business question that if answered will reveal the strategy to meet the CEO’s growth objective.
- Αlign key stakeholders. When looking for an alternative to marketing mix modeling, you need to align several stakeholders within your organization who each have different requirements. It’s best to do this early. In order to ensure usage and value on the backend, you’ll want to build a collaborative team across analytics, marketing, finance, procurement and IT to evaluate any potential solution to determine the selection criteria upfront.
- Identify your reporting needs. Analytics are only valuable if they can be delivered to business decision-makers when they need them in the format they are needed. Determining whether you need weekly, monthly or annual reports will help you better evaluate which solution is right for you.
- Decide what to do in-house. Evaluate your team’s skills and decide what level of support you need
- Do you have an analytics team to build models and forecast what-if scenarios?
- Does your marketing team need answers daily, weekly or monthly?
- Is your organization concerned about sharing data externally?
If you answered yes to any of the above questions, you are poised to bring this capability in-house with a software solution your team can use to accurately forecast the impact of hundreds of what-if scenarios in minutes. If not, you can still start small and build up your analytics capability over time.
If you are interested in learning more about an alternative approach to marketing mix modeling, read our blog post that compares marketing mix modeling with a new approach used to evaluate and forecast marketing effectiveness called agent-based simulation.