I started really short, but then I cross 700 words!
Chat Bots seem to be all the rage right now and they are slowly seeping into the Enterprise landscape. Users, whether they are employees (or) customers, are looking for [Play background music], “Instant Gratification”. This means that they want to access business services like banking information and product catalogs – whenever, wherever they want. So where is this wherever – the holy channel of messaging, obviously.
It is amazing the way social media took off, but even more astonishing is the rise and dominance of messaging apps. From Telegram to Messenger, users love and use messaging in their daily lives. So why do we need to give them another mobile app (or) website when we can serve their needs through existing channels. Kind of brilliant, isn’t it?
The need for bots is evident, but will you have questions before actually starting to build one? Absolutely! So we will now go through some of those questions and what the possible answers might be.
Where should I put my bot? [Channel]
Let me counter – where are your users? The channel that you use for your bot is critical for its success. Don’t force your users to come to you, the beauty of chatbots is that you can go to the users. If they are already on Facebook Messenger, build for Facebook Messenger.
Ensure that you keep yourself open for multi-channel as well. One channel might not fit all your users, to ensure that you can provide the same services through various channels. See if you can let users continue conversations as well so that they can start on Facebook Messenger and later finish it off on your website.
What technology should I use? [Tech Stack]
It does not matter! Chatbots are supposed to be about the user and the business. If they are using the greatest technology yet don’t help the business do things better for its users, then it isn’t serving its purpose.
Bots are like any other applications, web (or) mobile. So choose technologies that fit your team and your existing applications. Running on Azure already, deploy your bot there. Comfortable with Node JS, continue with that for your bot.
What should I build a bot for? [Use Cases]
The use case for your bot is absolutely critical. Ensure that you have multiple functions within your bot – password resets, FAQs and more. Ensure that your use cases fulfill some of the below,
- They are repetitive items where empathy and human intelligence are not required to make a decision
- Ex. A Password Reset
- They happen very frequently and currently, take time to get resolved for your users
- Ex. A Password Reset
- They can really be solved by a bot -> don’t convert a 30 field form into a bot for the sake of it
I seem to love password resets!
Should I break the bank? [Budget]
You should. No, I’m kidding 🙂 Start small and go big, chatbots should leverage what is already there in your business and this means that even teams with small budgets should be able to build bots. The more the merrier!
How much is enough? [Scope]
You will never have enough being handled by your bot, there will always be outlier cases. So make a conscious decision of how much your bot needs so that it does not frustrate your users. 60% of all questions, is that good for your needs? If it is, get that rolling. But make sure you have three things in your strategy,
- A way for the bot to respond back in “unknown” cases
- A proper escalation strategy to a human (or) a ticketing system (or) a knowledge link
- An agile strategy to continuously roll out new enhancements to your bot to make it better
Do we need a Data Scientist? [The Team]
NO!!!!! You need a domain expert, somebody who understands the use case for your business and somebody who understands what your users need. Chatbots are all about the business and its needs. You can always get a data scientist but you surely don’t need somebody for now.
So who do we really need to build a bot? Make sure that you have a full stack developer handy, they will work on building the bot middleware portion which handles the actual logic behind the bot. You will another developer who understand natural language processing, training NLP Dialog Models, and building Dialog Flows in your bot. Last but not least, the domain expert to help guide the needs of the bot.
Bot building is not tough unless you make it tough - Chanakya Lokam 04.05.18 :)
Have you built a conversational bot before? Have you had questions? If yes, post them in the comments below and I will try and answer them!
This is an easy read, simplified approach with the important basics. Great motivation to get going.