In this week's 'Voice Search in Mobile Apps', we break down the voice search feature in 'Paytm Travel' section of the Paytm app, owned by One97 Communications.
Welcome to a new blog series by Slang Labs, called "Voice Search in Mobile Apps". Here we tear down the voice assistant and search functionalities that have been added by businesses to their mobile applications and discuss them in detail. In this edition, we breakdown the voice search feature in travel domain of 'PayTm' Android app owned by One97 Communications.
We believe that it's essential to recognise the trendsetters and show how they are adding voice assistants inside their applications and its result. This analysis will also help developers understand voice search's usage and the importance of voice assistants in applications.
We have done a similar analysis on Amazon's Shopping app which has Alexa integrated for all customer's needs.
Did you notice the blue sticker on the walls of shops, behind the seat of the autowaala or at the tapri near your office? Yep! You got it right. We are talking about Paytm, owned by One97 Communications. Paytm saw a growth of 150% in transactions between April 2017 and March 2018. Paytm once limited to Tier 1 towns has gone deep and spread its roots further into Bharat. Paytm has added more and more features to its app moving towards 'One app does all' or super-app model. This model has been tremendously successful with apps like Alipay, WeChat in China and Grab and Gojek in south-east Asia, doing it all from food deliveries to payments.
There is a fundamental problem with super apps, the more services that exist, the more complex the app gets. The video below shows how many services the Paytm app has right now, and how cumbersome it is to find them.
According to Paytm's Vice President and Head of Paytm Travel, Abhishek Rajan, "The maximum growth will come from regional languages hopefully going forward with Paytm's increasing penetration in tier-II and tier-III cities. The next set of users coming online are from regions where they are comfortable in their native languages. So the use of regional language will become higher than what it is today."
Tier 3 cities and Tier 2 cities have seen a growth of nearly 100% and 75% in ticket booking by using regional languages. Paytm recently launched voice search for train booking witnessing this trend. They also noticed that people tried to type what they were thinking in mind. One such example Rajan gave was, people were typing Delhi 'se' Kanpur while searching.
Let's look at how Paytm has enabled voice search in their apps to reach the Next Billion Users from Bharat.
Paytm has been informing the users about Voice Search feature through periodic notification alert from the perspective of marketing. Along with this, they also have a running banner on the homepage of the app. On the UI front, when a person comes on the train booking page, there are coach marks that explain the voice search capabilities. In addition to this, whenever the user clicks on the source and destination city, they see a mic button right next to the text field nudging user towards trying it out.
Mic button is placed next to the search field. On clicking the mic button, you hear a familiar sound, and it opens up a waveform UI which covers nearly half of the screen. It shows "Please say your source city/station name or code". Even if you change the language of the app, the prompt remains in English.
A user can say origin station and destination station in one sentence with conjunctions in English and Hind, e.g. 'se' and 'to'. A user could also choose to give just the origin or the destination city one at a time, and the voice assistant would fill in the details for you.
The ability to say sentences with multiple entities (entities are variables in a sentence like a source and destination city) makes it more natural. It allows the user to express the intent in the same way he/she is thinking.
Unlike Gaana or YouTube, which take some action like playing a song and seeing watch history on the voice search, Paytm doesn't take a transactional action. Instead, it shows a drop-down list with the possible station matches where the user can select the closest match. This approach is unconventional to what other companies like Gaana and YouTube have been taking.
They believe in a voice to command model, at which point if the user has a clear intent, the action is taken, prioritising action over accuracy.
Ability to do voice searches and navigation in different Indian languages is missing. With 400% YoY increase in Hindi searches, it is necessary to make the search at least bilingual.
The NLP capabilities of the voice search can be improved significantly. It will allow the users to speak free form sentences and ask to book tickets without being constraint by a boundary.
Slang allows you to add voice search in multiple Indian languages in a fraction of the time with inbuilt analytics to gauge how users are using it. It also gives businesses the ability to add more voice-enabled functions quickly to the app.