Marketing mix modeling artfully presents statistical analysis from sales and marketing data. It can be used for determining the impact that various marketing efforts have on sales. Marketing mix modeling also renders data to help you choose the optimal marketing channels to utilize for a maximum ROI. A marketing mix will vary from organization to organization and takes into account many factors such as available resources, industry trends, competition, historical data from previous marketing campaigns, and the future strategic direction of the company. The marketing mix must be aligned with specific company goals . Sometimes this can be a fairly predefined recipe but most of the time, the ongoing marketing efforts of a company involve constant testing and adaptation.
The Pros and Cons of Traditional Marketing Mix Modeling
The benefits of using marketing mix modeling are seemingly endless and support sound decision making and marketing investment decisions. There are however a few limitations to note.
“If too much emphasis is placed on the short term sales it will devalue the longer term benefits of certain marketing channels such as search engine optimization and social media,” says SEO Specialist Kate Riley.
Another con is that many marketing mix models can have a strong bias towards time specific media placements such as television commercials versus less time specific media such as a trade magazine appearance. Also, the impact of brand equity is difficult to capture in marketing mix modeling.
Marketing-mix models use historical performance data to evaluate marketing performance and are therefore not an effective tool to manage marketing investments for new products. This is because the short period during a new product launch does not provide the accurate data needed for solid marketing decisions. Therefore, the more budget and time devoted to data collection the better For example, the initial advertising performance of Coke Zero was really poor and had low elasticity. In spite of this Coke then increased its media spend and focused on a better strategy, thus radically improving its advertising performance. Their changes were much more effective than the launch period. A typical marketing-mix model would have recommended cutting media spend and instead resorting to heavy price discounting.
For smaller companies with smaller marketing budgets, it can often take more time to collect online marketing data simply due to low traffic volumes being generated. Let’s say for example you want to test a product sales page using multivariate or A/B testing. Using a small marketing budget which drives low volumes of traffic you would need 30 to 90 days to collect significant data, sometime longer. Amazon.com for example receives so much traffic that they can test a page in 30 seconds. The bottom line is that you need good amounts of the right kind of data while also understanding how to gain insights from the information.
So What is Multi-Channel Attribution and How Can it Be Used in Marketing Mix Analysis?
Multi-channel attribution or multi-attribution modeling is the process of collecting and analyzing data from all of your marketing channels and using that data to determine the value of each channel. Using that data one can gain valuable insights and understand the success of the individual channels, how they affect each other, and their overall success as part of an integrated strategy. Multi-attribution collects data across all channels and aids in defining which marketing elements are the introducers, influencers, and closers. New software tools now exist that can follow a customer’s online purchasing path from their first search all the way to the purchase or conversion. It is important to track all of the various touch points in order to truly understand the value of each of your marketing channels.
Cookies Versus Fingerprinting
There are two tracking methods that can be used and are based on either Cookies or Fingerprinting. Cookies are little pieces of code placed into the web browser of a potential customer when they visit your site. These are typically deleted by the user after a period of time or flushed out by the computer itself. While the cookie remains on the browser, you can track the user’s search behavior, impressions from ads they are being served, and the click throughs on those ads.
Google Analytics, for example, has cookie tracking in its reporting. You can still gain solid insights using Google Analytics and many companies don’t need data more granular than this. That said, many large companies with significant online marketing budgets want and need better data and the ability to track consumer behavior across all channels for longer than 30 days. Google Analytics also uses last click data which means the analytics data is attributing the value to the last place the user came from. So for instance you might see conversions happening from your branded PPC terms but not be able to understand the entire click path that a user took or how they were originally introduced to your brand.
This is where fingerprint-based (fingerprinting) technology comes into play. This topic has become somewhat controversial as the technology uses information such as computer settings to identify a user across multiple computers. However, it does not collect any personally identifiable information.
The benefits of fingerprint based multi-channel attribution:
- A user cannot delete a fingerprint profile like cookies so it can essentially be used for an endless amount of time
- Ability to track across multiple devices
- Multi-channel attribution using fingerprinting can track an endless amount of touch points and the whole click path
- Ability to attribute value to all marketing channels and identify the introducer, influencers, and closer
- Ability to make truly informed marketing decisions based on understanding the true value that each marketing channel contributes to the overall marketing plan[KS2]
No matter what a company’s overall online marketing strategy is, multi-attribution analytics can be a very effective tool for marketing mix modeling.