Enterprises across the globe have come to a startling realization, albeit arguably a little late, that customer experiences go way ahead in business growth.
As Duthaluru takes a moment to explain the nitty gritties of the same at Uber, she says, “We build a marketplace system so that we can take our different customers request and match it in the best manner possible. However, because we operate in a digital world where things don’t go as per plan sometime, particularly in India. Our goal as part of this consumer help support or customer obsession is for us to be able to understand ride signals, pick up insights from the customer and give back to other teams so that it can automate correctly in the future.”
Machine learning for better experiences
Uber has relied on machine learning long enough to deploy it efficiently in its vision for better customer experience. The company’s home grown, ML-as-a-service platform Michaelangelo is now a de-facto system for its data scientists and engineers. Duthaluru, in the same vein, employed ML–coupled with interaction signals on the ride–to primarily predict any customer and/or rider-based issues.
The signals that double up as data sets can be wide-ranging: Completion of a trip, estimated time of arrival longer than suggested, change in destination, waiting for an order, etc.
“These signals allow us to indicate that your request might be about some of these things. We use machine learning to understand what that request might be about. We give the right option for that but the propensity that the issue is related to trip and not related to your account and not related to something in my account, we use machine learning to figure out what that might be,” says Duthaluru.
Post more information on the ticket, Uber uses natural/multi language processing technologies to understand what that ticket is to send a request to the right agent. “And because we’ve looked at your texts and figured what your issues might be, we send the query to the right agent,” says Duthaluru.
Additionally as the agent reads the message gets recommendations based on machine learning and natural language processing to get this information. “There are these phases of lifecycle and we are definitely looking at investing more in this sort of predicting, personalization and proactive behaviour,” she says.
Data and insights for better operations
Like many organizations, Uber too is gearing up to be data-driven. “We use data and insights to look at different experiences throughout from the standpoint of an user app. For example, if you go from one city to the next city and if you use Uber app, it will give you options relevant to that particular city and usually those are the options where you have been there before in that particular city,” she says.
Duthaluru mentions using data at scale and all of these insights to do a bunch of prediction throughout the lifecycle: From the time you book the ride to the trip, all the way to the point where you are likely to call help and support.
“In addition to that there are other use cases where we see the drivers fighting for certain types of issues, fares or a particular location, we go in and fix that particular experience, see how the fares can be calculated correctly. So, these are all instances where we are trying to drive more data driven insights to use allocated teams to define better experience,”she concludes.
Credit: Google News
