I have seen many variants of this question online.
Can a chatbot increase sales?
And then there are those who question the ROI.
Can a chatbot improve your ROI?
And naturally, some people are focused on the bottom line.
Can a chatbot improve your bottom line?
Read on for some anecdata that I generated to answer this question.
I added a test in very plain sight, and put it into my Support chatbot. The SupportBot helps people navigate my site and broadly speaking, exposes them to the different training courses and tools I have created to help Dialogflow users. It can also answer a few questions about Dialogflow (this is still a work in progress).
When people first send a message to the bot, it tries to understand their intent. (This in itself is an interesting experiment, but I can probably only publish the results after a while).
If it fails to understand the user’s intent for the first message it prompts with a second message which provides a list of choices.
At this point, the user is presented with a list of options as you can see. Of the 5 choices presented, notice that the 3rd one (middle) is “learn more about my training courses”
The training course choice
So you can see that the choice where the user wants to learn more about my training courses is placed right in the middle. In other words:
- it isn’t the first choice
- it isn’t the last
- it isn’t the longest sentence
- the user needs to type their words – it is not a clickable button, or a list control (so it means the user needs to do a little more work to select that option)
It is a reasonable assumption that people pay some attention and choose the training course choice only if they are at least nominally interested in my training course.
At this point in the conversation, when the user types a message such as “training course”, the bot shows them info about my training course.
Notice this: Right below the first course, there is an offer to get the “Intro to Dialogflow” course for a 50% discount. And in addition, there is also some scarcity: only the first 5 people who use that link will get the discount.
And also remember, all of the following needs to have happened to reach this point in the first place:
- the user could have started interacting with the bot only by clicking on the Demo menu at the top or by going to the very end of a post like this one
- the user needs to have started the chat by typing Hi
- the user needs to have typed at least two messages to see this offer
So this is an already “engaged” user.
So how many people redeemed the coupon? Till date, exactly one! (Yes, it is still available, if you want to use it)
How many people have seen this message in the last 30 days?
This is what my Chatbase analytics tells me:
So of 62 people, only 1 person has actually redeemed the coupon.
Why didn’t more people avail of the offer? I happen to know that a large chunk of visitors to this website are looking at the introductory tutorials, so it is also quite likely they would be interested in this course. In fact, I can say this even more emphatically – the main introductory guide to Dialogflow is a 19 step guide, and there are a total of 19 pages you need to navigate to read the whole stuff. From my Google Analytics, I can see that there are a lot of people who spend over 30 minutes total time to read these pages.
If attention is supposed to be a currency, I might as well be a millionaire! 🙂
Here are some plausible reasons:
- the offer doesn’t stand out (quite intentional – I don’t want to drive them away from using my bot altogether)
- they might not be interested in this specific course
- they might have thought – well, the offer is probably already gone
- and probably some other reasons that I don’t know
So can a chatbot increase conversion rates? Quite clearly, there is at least one person who got incentivized by the chatbot offer to purchase the course. I cannot conclude that they wouldn’t have bought it otherwise. But neither can all those retailers sending you coupons every week tell you for sure that the coupon is the actual reason for a purchase! And for that matter, until ToIP becomes possible – Telepathy over IP that is, no one will ever know.
An interesting thing I noticed (from my Chatbase analytics) is that a large percentage of people took the quiz inside the chatbot. With a little more work, it would not be very hard for me to use this quiz to get people to read more articles, which can then be used to get them more interested in my courses. This tells me that the chatbot can be used like an “interactive advertorial” – but its effectiveness for the purpose of conversion rates probably depends heavily on the existing authority and brand recognition of the business which creates it.