I can’t think of a better year to have kicked-off our inaugural Marketing Sciences summit, given the data-driven marketing cross-roads we’re at. On the one hand, many speakers at the summit recognize that advances in ML and AI allow us to find patterns in huge volumes of data, far surpassing our capabilities just a few years ago (Jaime Punishill’s presentation on ML techniques like word2vec is a great example). On the other hand, many also recognized we have a lot less of the “right” data to analyze than we like to admit. While many companies have amassed huge volumes of marketing content and related targeting information (e.g., data lakes), this is rarely organized in a manner that lends itself to fast and meaningful analysis. Increased regulatory pressure and public mistrust has also led many advertising platforms, particularly Google and Facebook, to further restrict the data they share with us in the first place, leaving us to wonder what the data landscape will actually look like in the coming years.
In other words, it’s best of analytics times and it’s the worst of analytics times.
The speakers at the Marketing Science Summit addressed these difficult questions head-on. Three key ideas kept emerging, if we all followed, would make us both better data stewards and better marketers:
1. Be Intentional About Use-Cases for Data Collection and Analysis
Before building (or re-building) massive data warehouses, focus on the use-cases first. Currently, many companies are collecting as much marketing data they can and then develop hypotheses to test. This means a lot of time and resources have gone into creating massive, agile databases that aren’t close to being fully leveraged. The risk in this latter approach goes beyond inefficiency– you also risk not being able to make a clear connection between the information/privacy that a customer has entrusted you with and any value that the customer gets from sharing that information. Joerg Corsten and Jeet Uppal’s conversation on Data Lakes touched upon this topic nicely.
2. Focus on Data that will Actually Make a Difference
While the digital data explosion continues, many marketers feel that there is a dearth of valuable data, especially data that helps them understand their audiences or predict some sort of real-world behavior. A lot of the passive data that’s being produced by consumers is intriguing, but it’s most actionable when it’s being modeled against something of value. Sandra Matz and Greg Durkin spoke about this in some detail, specifically on how behavioral social media data can tell us a lot about how audiences wants to be communicated to, but that data has to be modeled against self-reported, psychographic surveys to make sense. David Hardtke also reflected on some of the missed opportunities he’s seen when clients don’t connect their sales-type data to the rich behavioral segments they see in media platforms such as Pandora. Ultimately, data needs a compass to make a difference at scale.
3. Always Ask, “Does the Data and Analytics Provide Value to the Customer too?”
This was one of the most consistent themes throughout the day, and it really should be the first question anyone asks before acquiring and using customer data. For one, it’s wasteful to do otherwise. Customers are quick to shut-out any content that’s not relevant to them. As an example, David Hardtke said it takes Pandora customers at most 5 seconds to turn off an ad they don’t like (even when in the car). Kevin Johnson’s panel with Mary Michael, Julissa Viana and Kieran Fagan also had several case studies demonstrating how analytics reveals what’s relevant to health care audiences including one that helped educate caregivers on detecting early signs of dementia.
I hope everyone left the summit feeling optimistic as I am about the future of marketing and how thoughtful use of customer data create value for everyone, from manufacturer to customer. I look forward to a great discussion on how the space has evolved at our 2nd Marketing Science Summit in 2019!