It’s been 3 months since ILP’s end-users have been using their chatbot to receive career guidance. ILP’s end users – students, parents, teachers say that they are getting a pretty good response from the chatbot. It is timely, the messages and the data being provided is good enough to start planning out their future path. Bhanu, ILP’s program manager, says the impact that has been achieved is more than 60-70%. You can read on to learn from their experiences:

Abhishek: Can you tell us about the programs you’re running on Glific?

Bhanu: Right now we’re running Career guidance and Counselling programs. These are intended for students, teachers and parents. They can get information regarding courses and careers for future development. Career guidance is one part of how ILP helps a student in their holistic development. Based on the impact of this program, we would expand the use of Glific across other wings of our organisation.

Abhishek: How did you start with the chatbot and what problems were you trying to solve?

Bhanu: The reason we started looking at a chatbot is because in 2019 we had a small proposal to expand the digital footprint in our communities. We tried some feasibility studies on the ground to explore chatbot and automation solutions. Due to not much headway, we shelved the project. However, last year in July we revisited the idea. Which was when we got introduced to Glific.

By the end of january 2020, we got to know that there are a lot of problems to build the solution ourselves. We needed a server on our side, application development, a lot of stuff that we didn’t know about since we didn’t have the required technical person or tech expertise at that time. 

What we wanted was a frictionless way for our community to get career guidance. And for us to generate data from guidance & counselling.

We hadn’t used any other tools earlier, we just researched and found it difficult to get to a solution. It was when we found Glific that it clicked and we knew this is what we were looking for.

Our automated vs. manual interactions are now 85 to 15 percent, even with 2000+ users.  

Abhishek: What were some of the prototypes and feasibility ideas you got in the beginning when you were thinking about this approach?

Bhanu: Mostly our website. We considered different apps for an automatic response system. There were some companies who had their internal systems too. 

Abhishek: Why was WhatsApp chatbot a good way? Did you try any other method?

Bhanu: I’m not sure if you’re familiar with the CSM portal. But we do have a microsite for career guidance specially. There we used a pre-structured program for guidance. The students selected their grade and the courses or jobs they wanted to pursue after 10th. They would get information based on their choices. But remembering the website link, navigating to the right place, and failing to login were a few challenges we wanted to further eliminate. 

WhatsApp chatbot is dynamic. And our users don’t need to visit our website so frequently. It already has a notification system. It makes it easy for users to answer and for us to get the data. Also, everyone has it these days.

Thinking about user engagement from the start is also a good approach because Glific can then help to achieve that engagement.

Abhishek: How did you roll out your program?

Bhanu: After creating all our conversation flows, which were 87 different flows in 7 languages, I tested it out myself first and then I asked our team from other states – about 20 people from different locations. We found that we had to cut down bigger content into smaller parts. It took us 4-5 steps to test out with only a 100 sample group. 

We rolled out to 300 students in Hyd. Within a week we released it in Karnataka and Tamil Nadu. The initial public release was in the first week of Feb 2021. By 15 Mar, we were almost into every location in India wherever ILP works. 

We had spread the information about our chabot by creating some posters asking students to send us a Hi at the chatbot number. We used our social media channels and embedded the chatbot link on the website. We also have a good reach with the students so onboarding them wasn’t a challenge.

Abhishek: ILP is one of the earliest users, what state is your chatbot in now?

Bhanu: Our automated vs. manual interactions are now 85 to 15 percent, even with 2000+ users.  

We are now learning from the manual conversations how to automate these too. Most of these manual conversations are something students would ask outside of the flow. Even when we used to go to schools and do the sessions, sometimes students would not really follow the flow of information and ask something that was going on in their minds. We see a similar simulation repeating in the chatbot too. We have responded to about 700+ students manually till now. 

We’re recording all those questions from the users that are out of context and capturing the exact sentences and structure of the messages. So that in the future versions, our chatbot is prepared to answer those questions automatically too. I’m exploring DialogFlow to use NLP for that.

Abhishek: You mentioned you’ve used WhatsApp groups before, how has it changed for you from using WhatsApp groups to 1-to-1 conversations?

Bhanu: There has been a change for sure. In whatsapp groups we used to post the details of 1 career per day. And the students would ask their questions. After receiving all the questions, we go through them, and type a common message for all those questions. This was difficult for students to understand because it was not personalized, and students were missing out on the feeling of 1 to 1 conversation. 

Teachers never asked any questions in the group. Now with the chabot they are asking more questions particularly when it comes to one section of careers and entrance examinations. So in a way the engagement has increased. We still use groups when we need, and it helps us for a particular type of message to remind and motivate students to use the bot.

Abhishek: What is your biggest learning from building and running this chatbot?

Bhanu: Focusing on public needs is a big one. I resonate with what Reap Benefit is doing. They are able to connect with a large number of users. I’m pretty convinced with their philosophy.

Thinking about user engagement from the start is also a good approach because Glific can then help to achieve that engagement.

Thinking about program rollout is also important because designing it was simple for us. 

And another important one is to keep evolving the conversation flow design based on the learnings from running it with the users. It’s a growing and evolving process.

Abhishek: And for my final question, how did Glific help you with your program?

Bhanu: The best thing for me was that whenever we had a question, or a query or some information we needed, the Glific team was available.

You’re all really approachable on the discord platform where we’re continuously engaged. Even when I don’t have a question, I can follow conversations with other NGOs and get a lot of information about the new updates to the product. 

The best thing for me was that whenever we had a question, or a query or some information we needed, the Glific team was available.