Abhishek Anand

@abyshake

Most effective 3 step program to developing a chatbot

Your business needs a chatbot, and now you can have one.

Step 1: Build a chatbot using one of the many services that exist today.
Step 2: Delete everything you have just built
Step 3: YOU be the chatbot and start talking to consumers

No, this isn’t a prank. That is really what you need to do if you decided to jump onto the AI, Chatbot, NLP bandwagon and build a chatbot for your business.

Want an incredibly sophisticated, state of the art neural network to process customers’ messages coming in via Facebook chats? Want the system to have cognitive processing capabilities so as to understand, identify and analyze natural language messages? Want it to be robust enough to segregate and process the queries, compare it with historical message patterns and experiences and finally send back a friendly, personalised natural language response? Guess what, YOU already have all those capabilities!

Chatbots are nothing new. They have been in place for years. It wasn’t until Facebook opened up its bot platform that businesses all over went crazy about the need to have a chatbot. Why? Because an average Facebook user spends almost an hour on Facebook every single day. The problem? Businesses today are building chatbots without first figuring out why they need a chatbot, and what would their users ultimately be using their chatbot for, or what they would be expecting out of it. Just look at this Facebook chatbot, for example.

76,000+ people like Blink, almost 30,000 people messaged it last week alone, and yet, when it comes to me, I think their bot is trying to force me to cannibalism.

17 unread messages, and yet, every morning, they send me a “Cook human” reminder.

WHY SHOULD YOU HAVE A BOT AND MORE IMPORTANTLY, WHEN?

See, I am not going to deny the necessity of having a chatbot. There is a reason businesses are going gaga over having chatbots. One-to-one interactions are just not scalable. The only relevant word in that whole sentence is just that — ‘scalable’. So, before you decide to build a chatbot, you need to ask yourself a basic question —

Do I have the scale where I can no longer handle the questions on my own?

If the answer is a resounding YES, then sure, go ahead and build yourself a bot. But build it to do exactly that. For exactly that.

To have conversations with consumers because you can no longer handle the traffic. And have conversations the same way as you were having so far. NEVER compromise on the experience you were delivering that led your business to this point.

When your first step itself is building a chatbot, you completely missed out and skipped on the most crucial step — having the conversations your bot is expected to replicate.

The reason why most of the chatbots are unable to maintain engagement is because many marketers suffer from the problem of ‘knowing what their consumers want.’ Do not make that mistake.

Learn what your consumers want.

Listen to what your consumers are saying.

Understand what gratifies them, and more importantly, what ticks them off.

Start understanding the patterns in those conversations.

Then and only then get on to designing the conversational flow for your bot.

AN EXAMPLE

Let us assume you want to build a business around holiday destinations/packages (or you already have one).

What do people do when they come to your website? Possibly, one of these things:

  1. Look at the different holiday destinations, the price for different destinations, the kind of activities included in each package, and the duration of the holiday package.
  2. Search for a particular destination, and compare the various packages available before choosing one.

Now if you try to build a chatbot for your business, it is possible that you will follow this exact route.

Step 1: The bot will ask the user where does he/she want to go to?

Step 2: Where from?

(Step 1 and 2 are interchangeable for obvious reasons)

Step 3: Dates

Step 4: Display results along with booking buttons.

That’s what your bot will probably do, wouldn’t it?

The problem in that approach lies in the fact that you completely killed the soul of a conversation.

Chats are not supposed to follow the same route your website works. Chats are supposed to be conversational in nature. Had you been talking to your prospective consumers, it is possible that your audience would ask you questions like this:

  • Which are the holiday destinations within driving distance from <city X>?
  • Do you have hotels in <city X> that are pet friendly?
  • I am looking for a 3 day vacation, preferably in a hilly area. Do you have some options for that?
  • I am looking for a hotel in <city X> that has conference rooms as well available on <date>
  • <let your imagination run wild>

If you have had these conversations, you would have been in a better position to build the tech product that could have handled the kind of queries your consumers have. By doing so, you would have delivered a better customer experience and probably increased your conversion percentage much better. Instead, you decided to build a chatbot that would do what you wanted it to do, and not the bot that your consumers would have wanted.

Basic mistake, but quite a critical one.

Tech is meant to help you scale things past the point where it can no longer be handled by us puny humans. If you are trying to solve the problems before that point, it may very well be possible that you miss out on what the problem was to begin with.

And if you do that, the only thing you would be scaling up would be your mistakes.

That’s it for today; see you tomorrow!

I am Abhishek. I am here... there.... Everywhere...
Medium | Twitter | Facebook | Quora | LinkedIn | E-mail
Click here to join the mailing list.

More by Abhishek Anand

Topics of interest

More Related Stories