This January, MediaNama held an open house discussion on the future of Indic languages online. This is Part 2 of our coverage of the discussion. Read Part 1 here.
Indic languages can have a tough time on the Internet, and even Indian companies are facing unexpected issues when trying to replicate their user interfaces and app flows into local languages. As apps get more and more complex, and as people who don’t speak English as a first language get used to the language anyway in those apps, Indic language adoption faces new challenges.
Users may themselves not be very enthusiastic, even if their first language is not English, to switch over to Indic languages in apps they use. MakeMyTrip’s Director of Products Apar Surekha said: “If I have to give you a perspective on the role of Indic languages in e-commerce, I’ll have to divide it into two parts: from a user-need perspective, there are two needs. One is what to buy, what to book, in the travel world. For that you need a lot of information, for example if you’re trying to book a flight, you’d need information on the departure time or cancellation rules. If you’re booking a hotel room, you need reviews and other things. The other big need is obviously to complete a transaction and if I have to compare these two needs and put it in comparison to the fulfill need for Indic languages– the second need is to complete the transaction.
“What we’ve seen is that people are more comfortable in English, especially because the user flows have been made quite easy to traverse, to make a purchase and a booking. However there’s still scope and a need for content on helping users buy the right product or book the right hotel. It’s early days right now because whatever experience we’ve had, we’ve launched a couple of applications, a mobile site in different languages, for flights and rail, and a standalone app for rail in different languages. But whatever we’ve seen the transactions mainly happen through the English language. There is a small transaction base, very small user base, that prefers to transact in regional or Indic languages.”
Ola’s Supply Product Head, Sumit Kumar said: “When our business team went to on-board auto drivers, they were absolutely unable to understand our driver app. That’s where the business need arose, and obviously it gives us an edge over our competition, and hence we decided to venture into that. Once we did that, the response was pretty cool. We ended up developing around 9 languages.” On why the Ola consumer app isn’t being localized as much as the driver app, he said: “I’d say it was a matter of business need, as well as a prioritisation of resources. There was no real genuine need which came up in any of our consumer interactions. Consumers wanted us to solve a different set of problems as compared to displaying the booking flow in this.
“There is another interesting technical challenge we’re still trying to solve. When you talk about Indic languages or vernacular languages, there are two parts to this: there is something called a static string and something called a dynamic string. Translating the former is easy, but doing that for dynamic is harder. On the consumer side, the dynamic strings are more in number, because you are taking drop location and pick-up location, right? Translating that in the exact same manner across nine languages is quite hard. A simple example is ‘St’. ‘St’ could mean a street, or a saint, as in St Stephen College. So I think that problem is still to be solved. A lot of companies are trying to solve it but I don’t think it’s been fully solved. It doesn’t render properly across multiple OSes, multiple device sizes and so on.”
Rohan Khara, head of product management at MobiKwik said: “MobiKwik is barely over 1MB in size, it works on 2G, it has about seven Indian languages, with three more coming up. The kind of feedback we’re getting is fascinating. However, a lot of things are still left to be desired in terms of on-boarding. The way we thought we solved on-boarding was to have a language selection menu as soon as the app is fired up, and as soon as you choose that, that’s done. Then we realized, in lieu of making the app footprint smaller, we made a bad trade-off. We could have baked in the Noto font, which would have rendered Gujarati or Telugu properly on certain phones.”
He added: “A very interesting insight we got from merchants was that– we’d go to merchants and said ‘we have these ‘MobiKwik Accepted Here’ stickers in multiple languages, which language do you want?’ and some of the responses we got were hilarious. One said “paise ki bhaasha toh sab samajhte hain” [everyone understands the language of money], “if you want to give us the Hindi sticker or the English sticker, doesn’t matter, just give it to us.” People are very careful when money is leaving their wallet, but when it comes to usage, users are very happy about the fact that we have Hindi and other languages.”
Apar Surekha said: “Fundamentally it is the need of the user– I can give you another example when we acquired a Thai company, which was for inbound tourists into India. We were also trying to sell hotels in different geographies. We were operating in countries like Japan, Korea and other places. What we saw there was that there is a certain set of users who comes on a Japanese site, for example TripAdvisor’s Japanese site, and then will come to the Japanese hotel site, the company that we acquired. And then will transact in Japanese, and then– even the post-sales and payment and everything will happen in Japanese. But if we talk about that set of users that exist in Europe and Japan and Korea and so many other places, who prefer their native language as the primary language, we are not seeing that kind of traction from India right now.”
Issues to fix
Rohan Khara said: “I think the number one thing I’d like to fix, both from product and operations point of view is the customer support. Small example: the majority of the use before November 8th was to do prepaid recharges or postpaid bill payments, and invariably because of a lot of chewing, the problems go to a ‘pending’ state. When users would reach out to a team, almost 15-20% of the cases were in vernacular, and our call center team was not ready for that. And we would try and figure out, we would transliterate and provide some ‘jugaad‘ in getting to the root of the problem, because once we get your order or transaction ID, a lot of the problem can be smoothed out, but even getting to that point was hard in some cases. Right now, it’s more so. So from a product POV, if we can get some kind of really powerful customer CRM, where even though the keyboard used to input things is English, but it does a transliteration and pops the result in the particular vernacular, that would block a lot of queries. Secondly, text may not be the right way to do customer support, at least for payments. You could do it for so many customers, but when it comes to Indic languages, you may need phone-based support. So those are the two challenges we’ve been working on.”
Ola’s Sumit Kumar admitted to chinks in customer support too: “There was a CSC problem for us also. One year ago, I was managing the call center operations too in my first role at Ola, and I remember my call center foreman used to tell me that when he gets a vernacular call, the guy who has received the call, will remember from memory that this is a Kannada-speaking person on the phone, let me transfer the call to a Kannada-speaking guy, so imagine the loss of productivity and bad driver experience. Thankfully we’ve solved that with the help of our third party companies. The two main challenges we’re solving right now is: one, which is internal, is the entire mindset. Everyone, right from the design team to product team to development team to QA team needs to change their mindset. Hindi language, for the same word you need three layers (for the script). For Malayalam, words are over twice as long on average, in terms of length. And the device size is almost the same, so everyone in the entire life-cycle of product development has to keep this in mind. Otherwise, the experience can completely go for a toss. And that specifically happens a lot for new features. So we end up receiving feedback from the market and changing.”
He added, “Lots of discipline is also needed. We use a third party to translate content. So let’s say there is a video script or some content on a particular app, people literally translate it. You tend to lose the meaning there. A lot of discipline is needed internally across all teams. So if we have translators who are deeply immersed in our needs as a business, and translate accordingly, that would be great.”