Businesses are perpetually on the lookout to optimize every element. Including the marketing department. It means there is a lot of data generated and collected, reports, PPTs, graphs, and the best part: analysis. Good news? Many models can help you with the study. Bad news? It involves math. Math that’s hard to do. If you came into marketing to escape math, tough luck. The marketing department is responsible for making the most of each dollar. For that, you’ll need to understand where you are spending and how it’s affecting your metrics. An excellent method to do so is to use market mix modeling.
What is market mix modeling?
Market mix modeling is a type of statistical analysis that uses regression techniques. Remember regression from school? If you don’t, this is going to be arduous. Market mix modeling defines a relationship between variables and KPIs. Business and marketing departments use it as a decision-making tool. It is an excellent method to determine the ROI of a campaign and marketing channels.
Many companies use multiple marketing channels. Such as TV marketing (OTT), billboards, social media advertising, emails, events, etc. Each marketing channel will use the marketing budget in different ways and convert different people. Moreover, each channel may broadcast individual campaigns at different sales funnel levels. Most customers will go through multiple touch points before they perform an action.
Market mix modeling allows you to analyze past data to understand how each channel is performing and what they are contributing to the campaign results. Therefore, market mix modeling is a cross-platform measurement tool that helps you understand the performance of your marketing media using historical data.
Read More: SaaS Marketing Strategies: 10 Amazing Tips
How was Market Mix Modeling (MMM) formulated?
Let’s briefly cover MMM History 101. It is just to give you an understanding of how MMM was formulated, with no test at the end.
Neil Borden coined the market mix in 1949. According to him, “the marketing manager has to weigh the behavioral forces and then juggle marketing elements in his mix with a keen eye on the resources with which he has to work.”
Jerome McCarthy stated they were four variables that affected the model. They were price, promotion, product, place, and distribution. And the 4 Ps of marketing were born. Albert Frey had his own set of variables: offering and process variables. The offering included product/service, packaging, brand, and price. While, the process variables included promotions, advertisements, publicity, distribution, etc.
In recent news, Mary Bitner and Bernard Booms created the seven P’s method. They added people, processes, and physical evidence to Jerome McCarthy’s list.
Over the past few years, many companies have adopted MMM. And it is now considered a trustworthy method of analysis. It is now a widespread practice due to the availability of data, improved data collection methods, and computing power.
Sales components of a market mix model
It is driven by “organic” demand for a product. It includes factors like pricing, market trends, seasonality, and brand awareness/loyalty.
You can think of this as a “sponsored” demand. Such sales are due to promotional activities such as discounts, etc. It includes aloes bought in by tv advertisements, billboards, magazines/newspaper ads, coupons, etc. Moreover, within this type of sales, there are two categories of factors. They are short-term factors (coupons) and long-term factors (TV ads).
Base and increment volume
Base and incremental volume refer to the sales made by the base and incremental factors, respectively. Now, what is this important? All data is data, right? Not really. The point of using market mix modeling is to understand sales. And the channels that contribute to that sales. Therefore, it is crucial to sort and categorize sales.
Additionally, base and incremental volumes will help you understand brand awareness and strength. Incremental sales can contribute to base sales in the long run. Simply put, a new customer who uses a discount can become a regular brand customer. The base volume is an excellent indicator of brand performance and loyalty.
How does one prepare for market mix modeling?
Like most studies and analyses, your efforts should start before the analysis begins. Why? Imagine writing an exam without preparing for it. For success, you need to prepare for it. There are a lot of things that go into conducting an analysis. And for the best results, start prep work before you cook. Because the recipe is not the only success factor.
Step 1: Detail objectives
Market mix modeling is a methodology. Therefore, the first step is to formulate an objective for the study. Setting the purpose helps you decide the framework and data collection/sorting methods. Additionally, it will aid in the selection of KPIs. You should use the objectives to determine the study questions. Your questions should cover budget allocation, media channels, and revenue generated. You can also include your competition in the study. These questions will dictate what data is relevant.
Step 2: Sort and tag data
Different people in an organization will have different levels of access to data. Therefore, reach out to several people to gather data. It is always a good idea to have open communication lines throughout the organization across different levels of management. It is easier to collect data and sort through it if everyone knows what their responsibilities are.
Moreover, not all data you collect for the study will be relevant and impactful. Most organizations have data repositories, and you will have to sort through the data. The formulated questions and objectives will help you identify applicable data. The quality of data is crucial to a successful model. Therefore, be wary of false or duplicated data. Your data should be clean, consistent, and categorized.
Plus, you want to establish deadlines for data submission and processing. It will help in keeping your team on track.
Step 3: Understand biases and limitations
Personal bias is not the only type of bias in analytical studies. Market mix modeling has a heavy bias for time-specific media compared to less time-specific media. Time-specific media refers to the ability of the same to publish a campaign at a specific time. An example of time-specific media is TV commercials and social media posts. Fewer times-specific media refers to media that can not control the publishing time or date. Such as monthly magazines and journals. Biases also crop in when comparing national or non-demographic-specific media to regional and demographic-specific media.
Plus, you also have to factor in the limitations of the model. Since market mix modeling uses historical data, it can not forecast a new product. Plus, a short history of data results in inaccurate or unstable results. The model-generated values will fluctuate in the launch period and stabilize the longer the product is in the market.
How does one build a market mix model?
MMM involves using market channel data and internal consumer data to quantify sales and other KPIs to analyze the effectiveness of channels and campaigns.
Step 1: Data collection and categorization
You need to collect sales data along with marketing campaign data. Divide the sales data based on their drivers, based on sales, and incremental sales. If you forgot what they were, scroll up. After that, depending on the objective of the analysis, you divide marketing campaign data based on campaigns or marketing channels.
Step 2: Modeling a model
It is probably the most taxing part. Or the second hardest, depending on how well you work with data collection and sorting. Now market mix modeling is based on regression. Regression helps determine how independent variables affect dependent variables. While this might seem simple enough, there are many types of regressions.
There are two main types of regression models used to create market mix modeling:
- Linear regression
- Multiplicative regression (semi-log models and log models)
There are three types of interaction effects between the variables:
- Categorical variables
- Continuous variables
- One categorical and one continuous variable
Going any deeper into the topic would take up another fifteen pages. Plus, it’ll be hard to understand unless you have a math background. Or a major in the subject.
The model you choose will depend on a few factors. Such as objectives, data collection, etc.
Step 3: Analysis
Once your model gives you value, you start your analysis. For any objective, you measure the effectiveness of the channel/campaign, the efficiency, and the ROI/MROI. Over time, you can also compare channel metrics over the years/months/seasons to identify trends.
Step 4: Start optimization of processes
The point of market mix modeling is to provide data to optimize your processes. Easier said than done. How does someone actually go about doing so? Well, at the end of the process, get a few equations from the regression. These equations show how one variable can impact another. Therefore, you can use these equations to predict. Hence, the term used to describe the market mix model is predictive.
For example, your objective is to see pricing compared to sales. With the equations, you can play a good game of “what if?”
What if I raised the prices of the product by 10%? Or reduce it by 8%? The equations will let you have fluctuations in variables that will theoretically affect sales.
Which market mix modeling solution is best for you?
Here are four solutions for you, depending on your needs and budget. Keep in mind the workforce and effort such models takes. It is no small feat to build such models.
No one is going to understand your company data better than your employers. However, building a team for market mix modeling is like building a team for a race car pit stop. You will be able to customize the model and data security id better due to restricted access. But it is not a cost-effective solution for smaller companies. Moreover, building such models can take up a lot of resources and capital.
The next best thing for an in-house team is to hire an agency. This way, you do not have to worry about internal resource allocation, labor, and model building. Plus, it is an affordable option for businesses with limited capital. It is also the most popular option among many small and medium-sized companies.
It is the “bang for your buck” option. Expensive but high in quality. You also get a comprehensive analysis and a recommendations list after. It is the perfect option if you are a big company with capital. Especially if your data is complex and huge.
Think consultancy level work at agency level cost. It frees up your team’s time to perform other tasks while providing you with the modeling for each channel and campaign. Plus, you get actionable insight and recommendations along with continuous updates. This solution is for companies that have complex data sets, several marketing channels, and campaigns.
Benefits of Market Mix modeling
ROAS and ROI
Once you know the performance of each marketing channel, you can assess ROAS (return on ad spend). ROAS is the impact your marketing budget has on a specific channel. It is the ratio of revenue generated to the capital spent. It gives you insight into the revenue generated from a specific ad campaign. The ROAS only tracks ad spend and revenue generated from that.
On the other hand, ROI (return on investment) gives you an overview of the entire marketing campaign. It is the ratio of profit to the total amount spent on the campaign. It includes all costs. Such as campaign creation, running the campaign, etc.
Both ROI and RAOS are crucial metrics when analyzing the performance and contributions of different marketing channels.
Budget allocation and optimization
Once you know your ROAS and ROI, you can decide on budgeting. Most companies experiment with budget allocation to see its utilization and impact across platforms. With the ROAS and ROI data, you can identify channels that are underperforming and channels that are exceeding expectations. You should redirect capital from low contribution channels to high-performance ones. Or you can stop investing in low contribution channels. However, it is imperative to remember that low contribution does not mean no contribution. Plus, you can optimize for low performance in your campaigns.
The sales and KPIs do not remain constant throughout the campaign. They vary depending on creatives, copy, events, products, discounts, sales reps, etc. Therefore, you need to track several variables you use market mix modeling. Once you have all the data, you can predict what future trends are going to look like in the future. Hence, it will equip you better to allocate your budget optimally by predicting future performance.
The Biggest Limitation of Market Mix Modeling
– Lack of real-time data and complex data collection methods
All predictive models use past data during analysis. And this past data is then used to predict future trends based on past trends and actions. MMM is an excellent predictive model. However, like most predictive models, it is not 100% accurate. Therefore, factor in an error margin. Since all information taken is historical, it does not account for present trends. It is a lot like the weather forecast. You’ll be happy when it works. And stressed out when it doesn’t. Plus, data collection and classification are challenging since the model is multi-channel. You need data from all your channels on revenue and contributions. It means there will be tons of data to sieve through and sort.