Bots are the talk of the town, and Anand Chandrasekaran is helping to create the buzz. The former CPO of both SnapDeal and Bharti Airtel, the third largest mobile provider in the world, has plenty of experience with big ideas and global strategy. But he has really been stretching his wings as Global Director of Platforms and Product Partnerships for Messenger at Facebook. With the integration of open-API bots into Messenger, Facebook’s marketing strategy has changed considerably, and Chandrasekaran was kind enough to share some of those thoughts with me.
In part two of a two-part, in-person interview, I asked Chandrasekaran about how businesses can best benefit from chatbot technology, and to share some of the use cases he’s seen that have stood out as the most innovative and successful.
Newlands: How can businesses and entrepreneurs benefit from understanding how bots work with Messenger?
Chandrasekaran: The bot can be the solution, but that’s after first defining what the problem is.
If you’re a brand, entrepreneur or developer — using those words interchangeably — there are over a billion users who are on Messenger, and they are using Messenger for a greater amount of time every day.
If you’re an entrepreneur, this is a platform that you can’t afford to ignore. The way we think about it is, it’s your customers who are on that platform. Wherever you are on that spectrum, if you’re a developer, a brand, or selling a digital service, it’s your customers, fans and users who are on that platform. That is the main benefit for anyone who is looking at Messenger.
Newlands: It seems that some bots struggle to be an effective and engaging experience – why is this? And what are solutions for that?
Chandrasekaran: I wouldn’t necessarily define this notion as “struggling”, but I will say that when the purpose of the bot is not well-defined, it leaves the user with too many options. Like I said before, our best practices for developers is to always to start with a problem to solve, because when you define the problem very clearly, it becomes very obvious what part of makes sense to automate. That’s when AI functions very well.
One of my favorite bots is called Brew the Way; they’ve had iterations of the the bot where it says, “Do you want to buy a coffee, or do you want to fool around?” The joke is that almost everyone who uses a bot wants to trick the bot into doing something silly — it’s everyone’s favorite trick. But what we find is when the goal is clearer, the bot does the fantastic job. So when the bot introduces an experience like, “If you want to order a coffee, here are three steps, but if you want to just play around, we can take you there.” Then they do fantastic job of setting expectations clearly and creating an experience of play that’s fun and engaging for the user, rather than making the bot completely open-ended.
Newlands: What are some of the advantages that AI and bots provide for businesses?
Chandrasekaran: AI allows people to solve basic problems. Sometimes people ask some very basic questions, many of them repetitive, and those are problems best suited for AI to solve. Especially where there is training data available, and there is a use case that is fairly repetitive on top of that training data; even something like, “Are you open right now?” That’s a fairly easy question for an automated system to answer without a human getting involved.
One compelling success story is Globe Telecom in the Philippines. They have a hybrid bot called Gie that they’ve given a personality, a woman with a playful nature. When people ask the most obvious questions like, “Can you show me what my balance is,” or “What are your most popular plans,” those don’t necessary need to be escalated to a live person, because they will probably give the same answer. They’ve built an AI layer to respond to those questions promptly. Gie also has the ability to know when the question is a bit personalized or complicated, and can say, “Can you hold on a bit while I ask for a live person?” The question is passed thoughtfully to a human, and then the human can answer the question and pass control back to the bot. What we find is these kinds of purpose-built AI end up being super valuable for developers.
You may know that about 15 or 16 months ago we acquired a NLP engine called wit.ai which we use for basic heuristics and basic NLP. It is doing extremely well, and powering a large chunk of AI and NLP today on top of Messenger.
Newlands: What about sales?
Chandrasekaran: With sales we try to narrow the definition of the problem even more. Sales for a developer is customer acquisition; sales for a commerce provider is revenue; sales for someone who tries to generate awareness is like someone coming back and re-engaging. What we try to do is, regardless of the size of partners or the entrepreneurs that we speak with, we try to sculpt the problem down to: What is the problem statement, and what does success look like? Because the more generic the problem, the harder it is to measure if the project was successful or not.
Newlands: Where are customers engaging with Messenger and what are the business cases of engaging with Messenger? How do you see people do that?
Chandrasekaran: If you look at the funnel of interacting with customers, they are engaging throughout that funnel, depending on where you slice it. On the top end of the funnel it’s just generating awareness. At Facebook, we talk about this “generating awareness” case study of Call of Duty. They created a Messenger experience when they launched the game where you answered a bunch of questions in character and in the same language as the game, and if you answered all those questions correctly, you got an unlock code to see the trailer of the next game. If you are Call of Duty fan, it’s awesome that you got an exclusive piece of content for engaging with the Messenger bot. It was purely awareness generation, just to create buzz for the game coming out, and they were able to see about 6 million messages within 24 hours of launching the game, purely with the bot from Messenger.
The last use case I see towards the second half of the cycle is customer care. The comparison between calling a number, being on hold, and giving them a lot of context information and just pinging a business on Messenger to get a response back is incredible. The awesome thing is that messaging between two people doesn’t need to be synchronous. A customer can go and get dinner and come back to respond to the message, or the agent could look it up, take five minutes and respond. It’s not like they need to respond right away on the phone, so it plays with human nature a bit more and it’s also super convenient. From an agent or call center perspective, we have a couple case studies — like Rogers Telecom in Canada. They actually got a 65% increase in CSAT by using Messenger as a channel.
If you look at awareness generation, enabling new functionality, acquiring customers, and well as customer care, those are the four top areas where businesses can really utilize Messenger.