Learn more about data for your chatbot from Ilker Koksal
Mentioned in the podcast:
Also, please find the whole transcript below.
The Chatbot Podcast
Or subscribe with your favorite app by using the address below
– Hi guys, it’s Bálint from chatbottutorial.com. Today, I have a guest with me – Ilker Koksal. Welcome to our podcast. Would you like to introduce yourself in a couple of sentences?
– Thank yo!. I’m Ilker, CEO and co-founder of Botanalytics, which is a conversational analytics and engagement tool for chatbots, based in San Francisco. We support more than ten messaging platforms, like Facebook Messenger, Slack, Line, Telegram, Kik, WeChat, Twilio, Viber and many more. Chatbot makers analyze their chatbots and find their bottleneck points or set up bottles? To observe their conversation flow many more things they can do at Botanalytics. We are supporting it, we are a team of about ten.
– Amazing. I assume you didn’t start with a team of ten and who … this globally. (1m17s). Tell me about your start. What was your first bot experience?
– We have two co-founders in SF and on those hackathons (1m32s), we built chatbots. At some point, we realized that conversational analytics is very different than mobile and web analytics, and in those months we dived deep into conversational analytics and pain points, and then we opened Botanalytics.
– And when was that?
– About one and half years ago. Then we backed like 500 start-ups, and we got some couple angels and 500 start-ups and a couple VCs … And then we were growing fast ourselves.
– I can imagine. This was before Facebook opened its platform. Did you know that when you… the company?
– Yeah, I had a medium blog post before the Facebook announcement last year that you can take a look at. I said that Facebook would announce this kind of thing and that the bots would simplify our daily routines in this kind of way.
– What was the turning point for you when you said “Okay, this bot thing is going to be huge and I need to establish a company?”
– Actually, we had an analytics soon and opened up a tiny brand for chatbots, and we got great traction on chatbot and … to the Botanalytics, because we see huge potential here and we are bot makers, because we build a lot of chatbots and hackathons and we got prices for those chatbots, so we know what we need as a bot maker. And because of that, we really solve the problem and we …tutor the company in all conversational analytics.
– Ok. If I recall correctly, you have thousands of bots in your system. How many bot makers trust you?
– Yeah, more than thousands of bots are using Botanalytics, and we have Messenger bots, Slack, Kik, Line, Telegram, Twilio, Viber and some generate SMS bots as well. Recently, I published a chatbot market over BU, because we stored thousands of bots and their data and I published like general retention rate, functionality use, which use, and … and onboarding experience of bots, I published this kind of report. People can look at it on VentureBeat And because we stored like educational bots, banking bots, travel bots, shopping bots, any kind of bots we are storing.
– So, you have many unique cases and see the numbers. I don’t need the exact numbers, but can you please tell us roughly – like I guess Messenger is dominating – but by how much? …
– Yes, 40 or 45% of our bots are coming from Facebook Messenger, kind of dominating our system, but there is a big potential in SMS bots as well. I think customer support agent bots will dominate this year, but Messenger as a built spot in there and we have great studies from Messenger, like the Robbie Williams Bot we have. Robbie Williams is a UK pop star and he is selling audios and cloths on Messenger; we are tracking the payments and selling out funnels and they can track payments and see the bottleneck points in the conversation really easily. We are getting like a lot of messages from the Robbie Williams Bot, because he has a great fan base on Messenger and he turned offline users and offline followers to online ones, so this is a great Messenger case. And because there are minimal cases of this kind on Messenger, this one is kind of dominating.
– So you mentioned that you can track the payments as well. I assume attracts a lot of marketers that use – for example – Messenger as a marketing channel. Ratio wise, how many people are these who for example a brand, a sole entrepreneur uses Messenger, whereas someone who tried to do something like an app before but now a bot or a utility application?…
– Yeah, actually, I don’t know the exact number, but the fact is that currently there are 6 or 5 million business pages on Facebook that are now converting to a chatbot. So any business page, can convert to a chatbot. You can think about the 6 or 5 million pages will become a Facebook Messenger bot, so it’s huge if you think about it.
– Is there any industry that is dominating? You see the data, it’s probably finance or health care…
-Yeah, actually, we have in our corporate cases where we are specialising on banks, we have some banking clients, because we thought that banks are the very first example for chatbot, because they have a lot of customer support agents and they are trying to shift those agents to a chatbot. So in some of our cases, they shift some part of it, like 10%, 15% of their customer agents, it is a little bit, but they made it and we saw the bot cases on those banking accounts so we are storing a lot of data on
– How is the legislation on banking? I assume there are some privacy concerns about that topic.
– Yeah, we are offering 99% of our all services on demand, but for banks or some corporates we are doing on prime solution. So for banks, yeah, the security and all those kinds of things are really important, so at Botanalytics we are doing on primes for just banks.
– You offer many platforms, you mentioned, SMS or LINE Messenger. If someone has a bot on different platforms, do you combine data for them? How does that work?
– Yeah, of course. If they open an account at Botanalytics, they can integrate many bots into one account. So if you have a Telegram bot, a Messenger bot, a WeChat or LINE bot, you can integrate all of your bots into one account.
– And you can also see the data separated, so it’s not like it’s mixed up with Skype, etc.
– Yeah, separately, exactly.
– I know you’re a big fan on finding the bottleneck of the conversation. What do you mean by that exactly?
– Actually, currently I think there are two kinds of bots. One is rare, like a structured/structural bot and the others have some AI on their backends, and they’re kind of good on conversational outside/side and the others are kind of good on using some UI elements and these kinds of things. On bottleneck points on the conversational outside/side we see a lot of cases. Actually, if your bot is stuck at some point, you need to know that. So right now, at Botanalytics you can save your conversations… and you can dive deep into your conversation and see the bottleneck points…. We just released a Botanalytics’ assistant; it’s big and we believe analytics should automate it and should simplify bot makers’ daily routines, because bot makers are trying to simplify people’s daily routines while building chatbot. So, we should simplify their daily routines, and we are making analytics automated. So for bot … Botanalytics’ assistant can automatically understand your conversations and see the bottleneck points and alert you “Hey, today there’s a bottleneck point in this kind of conversations with this kind of sign to… users. So it’s really big, so you don’t have to dive deep into all those kinds of things, the Botanalytics’ assistant will just tell you about it. So we work hard to make it happen, and this week we’ll release it. We are improving it, because it’s really hard to find bottleneck points in this huge data, so we are currently focusing on some verticals like banks and then we will come to other verticals on Botanalytics’ assistant. On bottleneck points, this is the one exact solution to find.
– I see. So for example, the assistant tracks where the conversation gets stuck or retention rates. Is it sometimes just that the user stops for a while and just wants to reengage later on, because he has something going on???
– Yeah, we will come to this point, you know, in our mission you don’t have to take actions. As a Botanalytics we should take action for you, like we need to tell you “hey, this kind of … user right now is starting to churn” Now as a Botanalytics we are taking actions for your site and we are sending notifications while looking at previous messages of your unique users and we keep engaging your users, so you don’t have to do anything. We will report on what we do, and that’s it. It’s our mission, you will make ???
– It sounds amazing. What are some of the best cases that you saw where you could improve bot engagement or the retention in it?
– Yeah, actually, it depends on the industry, on celebrity bots, there are some KPIs on shopping side, on travel side on education side, on banking side; there are a couple of different KPIs. For example, for the banking side, you need to look at your funnels and you need to set up a funnel, because in our clients’ cases they build chatbots for their FAQ. So people are mostly coming to their website and asking a lot of questions that need to be answered correctly. So if your users completed the funnel well, you’re good to go, but if there are some things missing on the funnel, you need to dive deep into the conversation details and see what’s happening there. And then they look at it and they iterate their conversation flow experience, and then they republish it, and then they enhance their conversation flow to the best level.
– You mentioned a segmentation part. I think one of the most amazing things about chatbots is that they have to context. So they have to context in terms of segmentation as well, some people have different needs than others. So how do you help people analyse their segmented data?
– Yeah, actually at Botanalytics – without writing any kind of code in the conversation details – you can define your events. For example, your bot asks for your name and address, so you can define those kinds of things as an event and you can set up a funnel and see how many of your users reach out like what’s your address, like that.
– Ok. So for example, I could segment all the people who paid, because you can track payments as well and probably see through the data the differences between those who didn’t pay, right?
– Yes, exactly, you can also segment your users according to their retention data. For example, people at the beginning of conversation flow who follow 5-step conversations are maybe less likely to churn according to people who follow 20 step conversations at the beginning of conversation flow. So you can compare two kinds of beginning conversation flow experiences and then you can iterate your beginning conversation flow, because for retention, this is a really important starting point. So we have these kinds of segmentation options at Botanalytics as well.
– You’ve been around for at least one and a half years seriously in the chatbot space. What are the developments that you are the most excited about?
– Messenger is Facebook sided, really content chatbots and they are pushing the chatbot market, but on the other hand, customer support site companies are coming to the site, we don’t hear too much news from them, like Zendesk, LivePerson, this kind of companies, but we will see those great cases on there, because I think they are shifting their customer agent thinking to a bot in newer terms, which we will see probably at the end of this year.
– What do you think? What will happen in the next couple of years concerning the bots?
– Actually it’s really realistic to think that the bots will simplify our daily routines. Currently, I am using just one bot at Messenger, it’s the finance bot. I connected my bank account to the bot and I can unsubscribe some of my purchases like that, without logging into my bank account. So this bot really simplifies my daily routine, so if we see like 5 or 10 bots like that – simplifying the people’s daily routines – we can say “yeah, the market is really great”. and I think we will, because we are putting more UI elements into the chatbot conversation flow; it’s a good thing, because we need to simplify this experience. It’s hard to put AI at the first to grab some users. So the UI elements are really important for that.
– What do you think about voice bots? I think it’s coming in the future and I’m very fascinated about them, especially if they properly understand my English or Hungarian accent. Do you support them, or do you plan on working with voice bots?
– Yeah, we are supporting them, and I’m using Google Home and Alexa in my home; I’m really using them and it’s a really good case. It helps me at home and I think they are training the data sets , especially on bot Google Home and Alexa side. You will training the data sets and they will help more to users. Currently, it’s kind of basic, but they will improve.
– Some people say that they are becoming the God bots and that little bots will join them. How do you see that?
– Yeah, Facebook for example could be this kind of bot, because Facebook launched it two years ago and they are still working on it. If you think that all those bots currently on Facebook is a great skills set for Facebook. like a God bot. So, yeah, we will see these kinds of bots, but I think you will use other bots on those God bots, inside the God bots, but we will see the cases, it depends on the …cases, on other chatbot space and the God bots’ success rates while training the data.
– I see. I think Facebook has some kind of disadvantage with less voice options, and the less native options, as for example Siri or Google would have.
– I think they can get what you say in a rapid way because the voice is converted to a text, so it’s the same thing actually.
– Yeah, actually, I really like to use these voice mails but sometimes it’s not appropriate to send them because people can’t listen to them, so a great thing about the future is that you can just send it in voice and read the other part. But back to Botanalytics, people are sometimes very fixed on certain little hacks or tricks and timing can be one of them. Indeed, timing is very important to get the attention. What do you see in the data? There are certain times when people are more receptive to certain things or it’s not a big deal, at all? The best time to send out a message for example.
– Yeah, that’s a good question because currently, we are calculating the best reengagement times of users to send broadcast messages. It’s really important, because if you give the same service to all users at the same time, it will stick. While I’m at the gym, I may not get some notifications from the bot. For example, think about the food delivery bot, If it offers some foods to me, while at the gym, it’s the worst case, but after the gym, you could offer some deals, that would be good. So you need to know the best reengagement times of your users. It’s really crucial. We are currently calculating those times and we will launch it soon.
– Actually, while looking at bots with a marketer’s eye, I think the timing is very good with the bot knows that the user’s time is on If you play globally – like I always see my emails when they are sent out to me, but some of the companies are American and I’m in Europe, and they have different times for sending out the emails, but bots can solve this issue.
– Yes, exactly. It’s a really easy job actually, calculating the time difference and sending those exactly is really easy and yeah, it should be like that.
– It’s not about the calculating, but the fact that for example, on Messenger, people automatically receive data that enables them to use it, so you don’t have to do an email segmentation like “what is the best time to email you?”
– Yeah, exactly.
– I signed up a little bit late to Botanalytics and I cannot use the data from previous conversations. Will there be a solution for that to some bot developers who are just like “oh my God, I forgot this tracking thing”?
– Like importing to previous conversations?
– Yeah, you can import your previous conversations, just let us know at Botanalytics.com about your adjacent structure and according to it we can provide a link, and then you can post all your previous messages to us. It’s very easy.
– I see. How about people who use bot builder services, who don’t have to accustom code or other hosted variation.
– Yeah, actually we are approaching all bot builder platforms as an integration.
– I just wanted to mention today when we started like Chatfuel offers integration for you, from today on.
– Yeah, today we announced our Chatfuel integration, any Chatfuel bot maker can use Botanalytics within Chatfuel; it’s a good thing, so we are using this kind of integration with other bot builder platforms as well.
– Ok. How do you plan to monetise in the long run? I know that you have a freemium model, so people can use it for free with some limitations. What are these limitations?
– Yeah, currently we are starting to offer just one premium feature, that’s Botanalytics’ assistant, so for all other features will be free. If you want to use Botanalytics’ assistant, it’s just the premium feature, that’s all.
– Ok, so there is no limitation on the data
– No, there are unlimited messages, everything for free.
– Ok. How do you see your future in the company in one year from now?
– Actually if you look at similar cases from five or ten years ago, on web analytics companies, they evolved within 3-5 years, and they became billion dollar companies. The good thing is you need to iterate your product well. We will launch our redesigned product in one or two weeks. We always pay attention to our product and we are getting a lot of feedback from our users, so if you do it well, if you listen to your customers or users well, and we iterate the product well, and present it well, I think we will come to great point and we can analyse all those kinds of bots and help bot makers, educate them in their roles, because it may be hard for them to iterate their bots on some level. We want to help them, we are helping them, like we are giving Botanalytics 101 Webinars every week, we are giving funnels, events in like every 15 days, we are doing many things. And that’s the point.
– Yeah, and compared to websites, like you compared the company’s evolution, I think one of the advantages to bots is that there are so many data points that you don’t really have on websites. For example I always go back to Messenger, it starts off the relationship with the user with so much data. So I think it could become really, really valuable to a level that it couldn’t be compared to these days.
– Yeah, exactly.
– You are based out of San Francisco, you work with a lot of bot companies together. I always ask the guests who you would like to listen to as our next guest and why?
– Actually, there are many great guys in SF, so many great bot builders, bot enthusiasts. I can’t give you one name. There are many names. I have a lot of friends in SF, so.
– Yeah, just name one or two that you think they have an interesting bot case or you would actually spend more time with them talking about bots and you could call them to your podcast.
– Yeah, for a next podcast you could talk with Tony from Converse AI or Dmitry from Chatfuel, or Lauren from Pandorabots? They have great cases on their platform and they are helpful with educating bot makers within their cases.
– Ok. Thank you. So if someone is interested in your platform, what is the next logical step, where should they go?
-Yeah, just sign up to Botanalytics.co, and it’s really simple to integrate a bot at Botanalytics. It takes just a few minutes or seconds to integrate the bot.
– Yeah, I actually did it with another bot today as well, and it’s very easy, and I’m just excited to get ll the data. Thank you very much for being here and sharing your thoughts with us.
– Thank you so much.