Time after time, Stephen Jepson’s team at market research company DISQO has seen that people react positively when they are empowered to make informed choices about how they interact with brands.
The EVP of advertising effectiveness leads media measurement across the U.S., working with market research agencies, analytics companies and brands to improve their advertising efforts. The power of personalization has driven Jepson and his teammates to identify more ways to help users personalize their experiences on the company’s customer data platform. For example, they pinpoint how people receive communications and which research opportunities they see first.
They’re not the only marketers leaning into the potential of CDPs.
Seventy-eight percent of organizations either have or were developing a customer data platform in 2018. That’s according to Forbes Insights and Treasure Data research, which surveyed 400 marketing professionals.
LA adtech firm Viant uses third-party software to “avoid the complex and error-prone process of building in-house, deep-learning systems from scratch.” By leveraging deep-learning models to connect customer interactions across platforms, the Viant team is able to advise clients on marketing spend.
Tech professionals from both companies shared insights gleaned from their CDPs and the impact personalized customer experiences has had on their businesses.
According to Chen, those features wouldn’t be possible without artificial intelligence and machine learning techniques he and his colleagues are able to apply to the data sets.
Which customer data platform is your team using and why did you decide on this one versus other products on the market?
We use open-source, deep-learning libraries TensorFlow and Keras. TensorFlow is a framework developed and maintained by Google. Keras is a Python library that provides the TensorFlow project with an easy-to-use interface. Using Keras allows our team to design, fit, evaluate and deploy deep-learning models in just a few lines of code. We are able to avoid the complex and error-prone process of building in-house, deep-learning systems from scratch.
“Using Keras allows our team to design, fit, evaluate and deploy deep-learning models in just a few lines of code.’’
Tell us about surprising or valuable insights your CDP has unlocked. How are you leveraging those insights to improve personalization?
Our team is tasked with a wide range of problems that require using artificial intelligence and machine learning techniques to unlock new platform features and enhance existing ones. The processes of developing new AI and ML algorithms are often challenging and require many iterations of explorations and improvements. For data scientists, Keras’ simple interface allows them to stand on the shoulders of experienced code developers, quickly prove their ideas and iterate based on previous results.
The TensorFlow and Keras libraries also reduce project turnaround time by alleviating the burden of writing production codes to run ML algorithms.
What results have your company and your customers seen by creating a more personalized customer experience?
One of the features our team has developed is called machine learning viewability. By applying deep-learning algorithms to each incoming bid request, machine learning viewability is able to discover viewable impressions and increase viewable scale by four times that of an internal accounting standard segment alone. With this technology, advertisers are able to increase the overall visible impression volume, thereby hopefully reaching more consumers.