TechBytes with Brian Atkiss, Director of Analytics at Anexinet

Tell us about your role and the team/technology you handle at Anexinet.

I am the Director of Advanced Analytics at Anexinet. I help clients devise strategies to ensure the success of their machine learning and analytics initiatives. Our team takes advantage of many technologies to enable clients to best solve complex and varied analytics problems. I am also product owner of our ListenLogic Platform, which uses natural language processing and machine learning to analyze customer experiences.

Why are B2B marketing teams steadily moving toward Applied Data Science for Sales and Marketing initiatives?

Data science is revolutionizing analytics initiatives across all departments. Sales and marketing teams use machine learning and data science to create and adapt campaigns faster than ever before, and apply predictive analytics to answer questions previously based on a hunch. One specific example is the use of supervised and continuous learning for lead scoring. Previous lead-scoring techniques were either manual or based on limited fields/datasets marketers believed had a large impact on conversion. Machine learning allows marketers to utilize much broader datasets and automatically identifies critical buying signals, based on common attributes of historical leads that converted. Further, these models continuously learn and improve, resulting in much greater score accuracy overall.

 Tell us about your customer experience products. How can MarTech customers benefit from your AI/ML products?

Anexinet offers a wide range of AI/ML services and products. Our ListenLogic platform ingests customer-interaction data (calls, emails, chats, survey, social media) and uses advanced natural language processing, sentiment detection, and machine learning classifiers to turn unstructured data into valuable and insightful analytics used to prevent customer churn, improve customer satisfaction and contact center efficiency, and increase the effectiveness of sales.

We also provide consulting services that start with strategy to identify the best use cases, datasets/systems, ML algorithms, etc. and build a roadmap tied to business objectives. Our deep expertise in data-science also enables us to help clients implement custom AI/ML projects.

What is the current state of machine learning and AI for digital transformation?

Machine learning and AI are critical components of a full digital transformation. Every company focuses on providing the best possible customer experience, which requires utilizing advanced analytics to build seamless, personalized experiences. A great example of this is the use of recommendation engines to identify and present the ideal product/service to a customer and allow them to buy it without friction. These engines require vast amounts of consumer data and robust analysis to identify the exact product a customer needs and reveal the best time and method to send the recommendation to the customer.

Other examples include the use of chatbots and automated assistants to provide customers with quick, adaptive answers without requiring any human intervention. Predictive analytics is also being used for anomaly detection to identify fraud or even to predict and prevent customer churn. These are all examples of machine learning or AI being used right now in the digital transformation process. The use of ML and AI is only going to expand further as more and more companies realize the amount of data that can be analyzed and adopt applications to analyze this data and provide a better customer experience.

How much of this state is influenced by the maturity of Data Science and Machine Learning algorithms?

The maturity of machine-learning algorithms has certainly had a large impact. I would argue, however, that an even larger impact has resulted from advancements in cloud computing that let companies of any size implement deep-learning algorithms which require GPU processors. The theory behind data science and machine learning has existed for decades in the academic community, but very few businesses had the amount of data and processing required to realize all the benefits. Now that there’s more data available than ever (and it’s increasing exponentially), and companies subsequently are able to scale their processing, the practical use of machine learning is only just emerging.

While the future is very bright, we’re also at a point where machine learning is being touted as a panacea. But not every dataset may be utilized for a data science application, and not every use case can be solved by machine learning. It’s important for businesses to take a practical approach to machine learning and identify the best use cases—supported by the available data—that will lead to actual return on investment.

What are the major pain points for Product Management and Innovation teams in building/ scaling Analytics for customer experience?

Data quality is currently the largest pain point, as many of the datasets used for measuring customer experience were not initially designed to be used for analytics. Customer interaction data (call recordings, emails, surveys, etc.) doesn’t always include identifiers that can map across other internal systems (e.g. CRM and transactional databases). Additionally, call recordings are often low-quality, and heavily compressed, reducing the accuracy of speech-to-text conversion.

An additional major pain point is around legacy vendors who “own” the interaction data clients need to access for customer experience analytics. These vendors tend to charge large fees to access their data, whereas modern cloud-based vendors provide a much more flexible model for clients. The challenge for companies is to get out of contracts with legacy providers in order to make their data accessible for analytics.

How do you work with AI/ Machine Learning at Anexinet? (edited)

Our ListenLogic platform uses machine learning to classify unstructured text data from customer interactions. We created sentiment and emotion classifiers, along with custom business classifiers for specific industries. For the pharmaceutical industry, we use supervised learning to create ML models which allow our clients to identify and report adverse events in social media data without the need for human compliance analysts.

At Anexinet, we also provide custom AI/machine learning services, which—most importantly—involve helping clients devise their ideal strategic approach to machine learning use cases, and building a roadmap for rolling out AI and machine learning initiatives prior to implementing any large-scale projects.