Data isn’t just bigger these days; it is also fundamentally different than it was 10 years ago. The nature of this change is driving several innovations in the way marketing is done, particularly around targeting and measurement.
From a predictive targeting standpoint, ad tech firms are realizing that knowing a user regularly visits an investing blog and regularly searches for stock tickers is more valuable than knowing the age, gender and income of that user when targeting for a financial services brand. Traditionally, demographic and lifestyle data has served as a proxy for a good audience. With modern server logs holding behavioral data that tracks every last click, marketing firms can do away with the proxies and build audience segments with a high likelihood to take some sort of specific action — like converting. Ad tech startups such as Dstillery (full disclosure: the author works for Dstillery) and Rocket Fuel have based their respective approaches around this concept. Big data technology coupled with machine learning best practices has enabled the use of event-stream behavioral data to accelerate in the last five years. The market is starting to notice the value this approach is bringing, with Rocket Fuel being a recent IPO success story.
Better user-level targeting isn’t the only innovation brought by log file data. The digital promise of having better insight is slowly being realized by firms offering third-party ad effectiveness measurement. Companies such as Adometry and Visual IQ are pioneering the use of machine learning to model the causal effectiveness of ad exposures on user conversions. Using these models, brands can better evaluate which digital strategies are the most effective at driving up their ROIs.
Although the innovations discussed above have been developed mostly around web usage, the ideas can be directly applied to other digital channels, such as mobile and TV. In mobile, the fine-grained behavioral data is being measured as app usage and the geographic movements of the user. Companies like Flurry, Foursquare and the aforementioned Dstillery are making use of this type of data. TV advertising is probably the most embedded in traditional advertising practices, but even it is poised for disruption as household-level TV viewing logs become available. DirecTV, which controls access to both the content delivery and the data, is starting to target commercials to the household level based on its content viewing profile.
While these innovations are driving better targeting efficacy and more accurate measurement, they don’t come for free; the bigger challenges in particular are social in nature. Tracking user behavior at the event or click level requires a constant focus on protecting consumer privacy. Firms with access to this data need to bake privacy into their data storage and usage policies, and start adopting machine learning methods that are privacy preserving by design. Additionally, communication is often a constraint for firms innovating the use of data in marketing. The ultimate decision makers in marketing are business people, who, while smart and creative, are not often trained in the state of the art in big data technology and machine learning. The “Black Box” nature of the new suite of advertising technology creates an impediment for adoption. To increase the rate of adoption, ad tech firms need to adopt more transparency and be able to present their technology at an intuitive and approachable level. Demographics are easy for a marketer to understand and digest — logistic regression on sparse, hashed features is not. When this gap is bridged, we should expect data-driven marketing to evolve at an even faster rate.