As one of the leading technology brands in the U.S. and with each of its customers literally carrying around a beacon that can be used to interact with the company at a moment’s notice, Verizon has enormously rich data resources at its disposal. Formed in 2015, Verizon's Consumer & Marketplace Insights (CMI) group is charged with making sense of it all. To do so, the team deploys best-in-class consumer insights techniques with advanced data analytics capabilities. They have developed innovative solutions with scope to integrate with many different platforms, giving the flexibility to handle a variety of use cases, and answer a wealth of different questions.
As a direct B2C brand, Verizon works with third party research panels, and also will launch customer surveys using lists sourced from internal databases. This allows for integration and cross analysis between survey data and internal behavioral data. Similarly, Verizon CMI has been on the leading edge in leveraging third party data made available by partner firms. These work streams require connective tissue, good data hygiene, and proper planning - work that can prove as challenging in practice as performing advanced statistical techniques, and gleaning the final end insights. There are many road blocks that came up, some easier to see coming than others.
Verizon is a huge company, formed as a legacy of many mergers, and we have many different platforms and solutions in place. We do not use the same vendors for customer relationship management that we use for digital ad targeting. We use yet another vendor for syndicated market research, another for media research, a fifth for market sizing, and a sixth for outbound acquisition campaigns. We have countless research projects where we can’t deploy our typing tool – but these are largely syndicated sources where the vendor may be collecting PII. Unless you are able to field your typing tool to every consumer in the market, activating your segmentation across different kinds of vendor platforms benefits from a common key to bridge the gap between disparate data sets. That also opens up a lot of interesting doors as well, along with a lot of challengers in keeping all of your consumer data safe, and maintaining compliance with research ethics rules and regulations.
Having this linkage allows for using techniques like classification or lookalike modeling, but a variety of options are possible, such as building a crosswalk between your segmentation and a more popular third party one that’s in wider usage. If a linkage is not possible, you could try to recreate your segments in other platforms, by using business rules that produce consistent results to your segments. That level of complexity is just what exists internally. Reconciling the broader external landscape introduces even more challenges. All of our efforts in this area are an attempt to make sense of this cacophony to the best of our abilities.