Co-Founder/CEO of Bhuja D Digital with a mission to discover a model of conscious capitalism and media organization .
Over the past two years, while marketers have been trying to solve the problem of losing third-party cookies, new, smarter ways to invest in media have emerged. As cookies become obsolete, tracking the triggers of desired marketing behavior will become more important than ever.
How do you know if you're reaching the right consumers or if your media and marketing mix is working? Waiting for the cash register to ring is no longer enough, and media mix modeling (MMM) or multisensory attribution (MTA) are being replaced by dinosaurs. What replaces it? I believe this model is a large mix of media data.
The ultimate solution for smart media buying and performance analysis can be found through big data media mix modeling. It becomes the most sophisticated way to track and predict future performance of marketing and direct media investments to get the best ROI and ROAS.
The democratization of machine learning, the ability to process big data in virtual machines, and access to big data analytics from Google and Amazon have changed the rules of strategic media investment and performance prediction.
Where does big data media mix modeling come from? Direct Response Television (DRTV) was the prototype of BDMM. This was the beginning of the integration of linear media investments with digital campaigns.
So what makes big data media mix modeling so powerful compared to previous campaign analytics and media planning options? Media mix modeling using big data provides a holistic view of marketing. It combines all your marketing, economic, political, seasonal and even retail data into one view to show the performance of your media investments and campaigns.
The beauty of combining them is that you get a real performance view of what's happening over time with every possible variable. Big data media mix modeling allows marketers to see the correlation of direct sales by adjusting channel investment based on media type, season and level of investment. By integrating machine learning, data gets smarter over time.
Artificial intelligence compares spending levels to show when you're overinvesting, so you can invest or reinvest in better channels for better ROI. For example, instead of using information and multiple data points to increase your spend on Instagram, you can say that TikTok has a direct relationship with other media elements.
Big data media mix modeling can inform you about your proposed media mix and where to increase or decrease spend based on actual performance and the interaction of all variables related to revenue and sales. Its goal is to provide less guesswork and more informed real-world decisions based on statistics rather than guesswork.
What's wrong with Multi-Touch Attribute (MTA)?
Labeling problems led to the downfall of the MTA. This trick always leads to complex tagging problems for different types of media. You need tags for all types of media, including branded and non-branded, as well as the walled gardens of Google and Facebook: you'll never get a complete 360-degree view of your campaign's performance across all media types and investment levels. Marketers are forced to make decisions with partial or inaccurate information, making continuous improvement a dice roll.
How about media mix modeling (MMM)?
Although not as complex and expensive as media mix modeling, big data does not lend itself to deep analysis. Big Data Media Mix simulation provides information at low computer speed and with a very high level of confidence. You can model it according to your KPI. Neither MMM nor MTA can boast of such an indicator.
So one might wonder why marketers are slow to take such a smart approach to media planning and buying. While this is one of the most sophisticated marketing analytics and media planning techniques and tools available to marketers today, there are some things that can make it difficult to implement.
Cost, media investment required for model calibration, and expertise required for implementation are the three most common barriers to implementing and performing big data mixture modeling. You need a data scientist or data engineer, several hundred thousand dollars to access cloud-based data management and analytics services from Google or Amazon, and a substantial advertising budget for at least three months.
If you're a CPG brand, the only potential problem is accessing retail sales data. This can be avoided by paying for access to the data to get a comprehensive view of performance. The reality is that BDMM will not be as effective unless you are missing a significant percentage of your retail sales data.
The key to successful big data media mix modeling is access to the most accurate and complete data. Partial data or data from different locations that are difficult to collect over time can lead to an inaccurate model.
The power and accuracy of modeling a robust media mix based on big data will pay off many times over. As the saying goes, no pain, no gain.
Forbes Agency Council is a closed community of leaders of successful PR agencies, media strategies, creative and advertising agencies. What am I entitled to?
