How Comcast’s Xfinity Home is using Analytics and more to Drive Business Decisions

By: Comcast’s Shuvankar Roy, vice president, Xfinity Home and Neeraj Grover, director of business analytics and reporting, Xfinity Home

Q: Why are analytics right to use to drive business decisions?

A: Analytics can help to identify the actual pain points in many parts of the business, including the customer journey. By using analytics, it’s easy to prioritize initiatives and avoid making decisions based on hypothesis, with no evidence to back up your work. Data offers significant insights into your business and can help you to course correct, as needed. Additionally, data can help you to define goals, make forecasts based on trends, patterns or the season.

To give an example, we here at Comcast are hyper-focused on customer satisfaction. For us, it is imperative to gain insights into our customers’ interactions with our front-line employees, such as technicians and customer service representatives, to help serve them better. Measuring the net promoter score or NPS at all levels including transactional, products, and employees help us to identify pain points that customers may experience from our service and products. Without these insights, we would not be able to improve or change processes and better serve our customers at every level.

Q: What type of analytics are essential?

A: To a large extent it depends on the maturity of the organization. To start with, you need Business Intelligence tools to identify and measure leading and lagging indicators. With this level of analysis, you will be able to understand what is driving an uptick and what may be causing a downward spiral.With Business Intelligence, your organization will be able to start predicting what is happening and provide corrections as needed.

As the organization matures and starts collecting more and better quality data, consider digging deeper beneath the numbers by using machine learning algorithms. These can provide further insights into top factors impacting your KPIs and help come up with a few “if-then” solutions that may help improve your metrics. We recommend testing each solution separately to see if the outcome actually enhances the indicators you want to influence.  The tests help foray from predictive into prescriptive analytics.

A particular type of Analytics can also be used for specific applications. For instance, Artificial Intelligence when paired with chatbots can offer your customers immediate access to assistance, and the text analytics can also help provide insights into areas that may need improvement.

Adoption of technology that leverages analytics to solve issues could become the critical differentiating factor to improve service delivery.

Q: With so much data/analytics that exist within a company, what are some best practices to narrow down the data that matter the most?

A: We see three factors of success for any projects: rely on extensive domain knowledge, be very skeptical of visible past trends and don’t  overanalyze,

Providing enough domain knowledge at each step of the way is key to identifying the most relevant data for any analytics study. The structure that works the best for us is having the analytics and business intelligence teams work carefully (or at times embedded) within the business units. This ensures that there is participation from subject matter experts to provide real-time feedback as insights are provided. The input and validation from SMEs ensure that the right data elements are being considered.

The other factor is to be very skeptical of apparent past trends as market conditions evolve. Lastly, avoid over analysis. In some cases getting 80 percent, accurate data is sufficient to understand directional and correct patterns which can help you make timely decisions. Avoid the pitfall of ensuring 100 percent accuracy as you may miss the opportunity to course correct in time.

Q: What type of business decisions can be made based on analytics?

A: Many decisions can be made based on analytics, but it’s important to look at the whole picture such as market conditions and what’s going on in the world. As you begin to make decisions based on analytics keep in mind these key points:

Analytics may be wrong sometimes as correlation is different from causation so take immediate corrective actions when you realize the change that was implemented is not working. Don’t be afraid to pivot and move on to another solution.

Prioritize analytics initiatives based on business goals. You can get a lot of data, and there may be many areas that need to be fixed, but you can’t do it all so narrow in on the few that will make the most significant impact to your business and go from there.

At times, be sure to complement data with other approaches. Sometimes it’s essential to conduct a few focus groups or review processes to find the triggers leading to the lagging data.

Q: Can you share an example where you made some critical business decisions based on the analytics?

A: Losing customers or churn is a measure that is key to most businesses. A while back our team leveraged decision trees and other machine learning algorithms to predict the type of customers that may have a high propensity to churn, and we identified key factors that led to it. The outcome of the machine learning algorithms identified customer engagement – the lack of activity and usage with the service – as the most impactful predictor of customer churn.

The importance of this factor led our teams to dig deeper into customers’ engagement with their services, their tenure, the services they have subscribed, and their preferred channels of engagement with us. This led to further insights into how our customers engage with each product and what service delivery steps could help drive customers to have a better experience. Ultimately, we found that customers who participate with or used the product(s) regularly led to more satisfaction with their service, which lowered churn.

Xfinity Home touch-screen

As the IoT space expands with more and more devices in a secured and connected home, the value for AI to help further improve customer service will be imperative.  Machine learning supported chatbots will become more sophisticated as they can scan for any system issues or other similar customer issues and quickly help to resolve and respond to customers. This level of customer care and service can be provided at an increased scale and response time will be quicker without adding to the cost of operations –the cost to provide the best customer service may even decrease.

Facebook Comments