How data science, ML, analytics, are key to customer journey lifecycle at Uber
In an exclusive interaction with Express Computer, Vidhya Duthaluru, Lead - Customer Obsession team, Uber India highlights the continuous innovations in next-gen and proactive support systems implemented by Uber for an exceptional customer experience
How is the data science shaping-up the business and deepening customer relationship at Uber?
Data, data science, machine learning, analytics, are key to customer journey lifecycle. So, when a user opens the app, in order to book a ride, we need to know where the user is going, who the user is, as we also ensure to track and block the potential frauds. And also identify the user’s preferences.
Meanwhile, we are also accessing our data and spotting the available drivers, calculate the fair from point A to point B. Who are the drivers that can potentially match for the user based on certain characteristics? How do we ensure the best customer experience? We also ensure to deliver the best customer experience, as we capitalise the driver’s livelihood.
Based on these data points, we calculate the whole dispatch mechanism, which is heavily dependent on data. In addition, we apply a lot of analytics. It is actually the combination of the historical data and a lot of real-time marketplace data, like, how much is the supply in a particular area? Is it raining because of which our supplies are constrained? Is the traffic is crazy, because of which the drivers are not around? Thus, we analyse the layers of real-time dynamics on top of any historical forecasts.
Moreover, we need to keep providing these real-time insights to our local operations teams, that are in the city. For example, is that really at optimal levels based on that they might decide to do certain things, they might run certain promotions, they might take away certain promotions, they might try to balance that whole supply-demand marketplace. Therefore, we must provide them a lot of local real-time analytics.
We definitely analyse trip information to understand any indicators during that trip that could potentially be areas where we could improve our product.The riders’ support tickets are also analysed from an analytics standpoint in order to identify how to best handle these queries and issues, from an approach of feeding it back potentially into the product. Can we make maps, marketplace or fair prediction better, comprises on data science and machine learning and feedback loop to constantly make the product better.
How is data and technology helping you to ensure frictionless support experience?
Firstly, we are being proactive in trying to create these frictionless experiences. If the user opens up the help section and actually files a support ticket, at present there is some friction. Because the user already has an issue, and then they go on to search for a solution.
Here, we are trying to understand what the issues could be, be predictive. The other is, can I understand the lifecycle of events that might lead me to believe that a certain case could become a support ticket in the future, therefore, I can proactively act upon it?
How are you leveraging emerging technologies for hyper-personalised experiences ?
We are heavily data-driven and leveraging ML. We have algorithms that listen to a stream of events, learn patterns, and provide us that proactive reach and further layer it with the user’s information to interpret a certain action.
We are also using sensory data, which is the data we collect from the driver’s behaviour. For example, how long does a courier wait to pick up an order from a restaurant, how long does the courier takes to go from the parking area to the restaurant, etc. It could be a bunch of activities which help in making the ecosystem efficient.
We use data to feed information into every part of the ecosystem.
How are you keeping pace with constantly changing technology and customer behaviour?
Customer inputs are incredibly important to us, hence, we analyse the user activity on the application.
We’re also looking at driver behaviours, like, how often do drivers cancel their rides? When do they cancel the rides? Is there a pattern? So customer behaviour and driver behaviour allow us to get more data into our ecosystem so that we can layer multiple sets of models and create hierarchical sets, and ensure the best experience.
We can’t necessarily solve it for every single user, but definitely solve it in segments of users, depending on city, location and the user’s demographics. We use the data in a very rapid fashion and always have to iterate.
It’s actually important to leverage data at scale. But it’s equally important to match up with constant changes, and have this data pipeline setup, where we constantly feed the data, improve the models, build capabilities to configure and tweak this very efficiently.
At the end of it, Uber is a whole tech component. But our operations team require tools to enable them to override certain leavers if needed.
Where has Uber reached in leveraging speech recognition?
In speech recognition, Uber is pretty a fledgling. We are still evaluating whether to invest in voice technology or work with someone who has already invested in the technology, and build the potential solutions.
There is an element of regionalisation or localisation in the driver’s Uber application today. If the drivers need the support, the driver has a number to call for human support and get connected to someone who can speak their language, and assist them in various issues.
But we’re also exploring the visualisation efforts within the app, like drivers can do a bunch of things within their app in local languages. And we are gradually exploring the capabilities of voice bots, as part of our roadmap.
Uber is also in the food tech industry, how is technology innovating Uber Eats?
We call the customers as “hangry customers”, because hungry customers are really angry.
We use a lot of data to quickly respond and resolve the queries. We are still in a very evolutionary process, as in the Uber Eats marketplace, we have three elements, the eater, the restaurant, and the delivery partner.
Now any one of these could potentially behave in a different way than expected and the situation may fall apart. Hence, we continuously improve the app, experience and be more proactive being relevant to the user.
What are the next-gen technologies Uber is excited about?
From a customer support standpoint, Uber is focusing on building multiple support channels. We are trying to establish more real-time channels for support, where we enable the chat channel. We’re enabling augmented phone support for our drivers and driver partners.
We have enabled phone support for everybody who has a live Uber Eats order, including courier partners, and restaurants. Being able to give the user support in whichever channel they want, when they want, is another big innovation. Thereupon, provide that automated experience wherever it makes sense, where the user doesn’t really have to interact with human support.
According to me, the industry at large is on their journey towards AI. Because to me, artificial intelligence is when machines are able to learn and piece together different snippets of information and come to a logical conclusion. AI technology has the ability to maneuver the way around the conversation.
In the space of voice, NLP and conversational commerce or conversational interactions, we need to reach to an ability to stitch together the thread of the conversation. It’s an exciting time to be in the space because there is a lot more data to help us, to be more informed and eventually get there.
If you have an interesting article / experience / case study to share, please get in touch with us at [email protected]