Recruitment Chatbot for the Civil Service College

Recruitment Chatbot for the Civil Service College

Description of Project: This passion project involved the creation of a recruitment chatbot called "Rena" for the Civil Service College to improve their hiring process. The bot was designed to attract and screen candidates, answer queries promptly, automate repetitive tasks, enhance the candidate's experience, and strengthen CSC’s branding.

Stakeholders: CSC HR and Candidates

Responsibilities: Although this was a prototype, I was responsible for the overall design, development, and implementation of the chatbot. This included creating the bot's conversational framework, programming its AI capabilities, and working with a team to troubleshoot and attempt to integrate it into CSC’s existing systems. I also worked on troubleshooting, refining, and updating the bot based on user feedback and performance data.

Duration: Nov 2019 to March 2020

Tools used: Chatfuel, Placeit.com, Canva, Filmora, Magisto, Google Dialogflow

PROBLEM:

I was a hiring manager from a different department and realized that we were losing many qualified candidates because the entire hiring process took extremely long and good candidates usually can’t wait. Therefore, this became a passion project which I took up to help solve two problems faced by my HR colleagues.

Problem 1: 70% of HR’s time is spent on manual completion of tasks. Typically, 50 - 70 candidates apply for one job posting. On average CSC HR spend 30 minutes per candidate to screen for experience, knowledge and skills, answer Job FAQs which included the application process. If we engage 20% of all candidates, that is seven (7) hours spent on qualifying these candidates alone. And that’s for just ONE job posting.

Problem 2: Communication is a mess in many hiring processes. Workopolis found that 43% of candidates never hear back from a company after one touchpoint. This “Applicant Black Hole” problem, over time, will weaken CSC’s brand. So how can the College improve the responsiveness of its hiring process and to hire at scale?

SOLUTION:

I came across Chatfuel while attending a talk about digitalisation in CSC and understood that it was a free chatbot builder that didn’t require too much coding experience. I wanted to learn chatbot building as I felt it could solve real business problems and this would be a great solution to help automate many aspects of the hiring process.

The Recruitment Chatbot solution should be able to do the following:


The chatbot, which I named “Rena”, was designed to engage with candidates, conduct initial screenings, answer frequently asked questions, and provide updates about the hiring process. By using AI, the bot could operate 24/7 and consistently present the CSC’s brand in a positive light.

PROCESS:

The process started with a simple analysis of the current recruitment system by interviewing HR colleagues, followed by identifying the typical questions candidates will ask as well as determine what features the Chatbot should have.

Based on the data gathered, I decided that “Rena” would be able to do the following tasks:

  1. Share about CSC, its culture, and benefits.
  2. Allow candidates to browse Job Openings
  3. Collect Candidate’s information and to pre-qualify them based on certain expectations.
  4. Answer frequently asked questions about the Job and the application process

THE "RENA" EXPERIENCE

Here are some of the screenshot recordings of the experience for Candidates.


CREATING NEW CONTENT:

We had a fair number of pictures and videos of the College, but I needed to curate these assets as well as create new content to describe our work and culture. For content creation, I used Magisto, which was an AI tool that generated polished videos, to create video content as well as Placeit.com to generate mockup images of CSC Officers wearing T-shirts with CSC’s logo. Even “Rena” was a mockup created using Placeit.com. Canva was also used to create images. 

PROGRAMMING:

Next, I designed the conversational framework and programmed the bot's AI capabilities. Thereafter I asked friends and colleagues to test out the bot and these were some of their feedback:

Up to that point, I was manually creating responses on Chatfuel based on the frequency of questions asked and I felt that “Rena” needed greater sophistication. I discovered Google’s Dialogflow, which was a natural language understanding platform, and managed to integrate Chatfuel with Dialogflow. Dialogflow was extremely clever in that it could guess the intention of the user rather than focus on semantics and I only needed to programme the responses in Dialogflow. This changed the conversations between “Rena” and users to something that was more natural.

THE PITCH:

Seeing I had a minimum viable product (MVP), I pitched the idea to my director and shared how it could help our HR colleagues. My director, having personally experienced the pain of slow hiring processes, was supportive and brokered a meeting with the HR director.

The HR director, together with her entire HR team, sat in for the pitch and were impressed by the presentation and demo. They supported the idea and requested that I presented this to CSC’s Senior Management team as well.

In Nov 2019, I presented at CSC’s Senior Management meeting, showing the problem and how “Rena” the recruitment chatbot can solve those problems. I did a demo and invited members to experience the chatbot themselves.

I showed the potential reach of the chatbot:

And the placement opportunities to attract talent to CSC via paid media:

These mockups were created by admocks.adpalor.com which unfortunately is no longer available. 

I also shared what other Spin Offs can happen with a chatbot, such as seamless onboarding for new hires, getting quick sentiments as part of employee engagement as well as provide immediate answers on HR questions for staff. The potential was limitless.

Senior Management was impressed and supported my request to work officially with HR, IT dept as well as Branding team to develop this further.

HOW IT ENDED:

Alas the time working with the team was short lived as I was to go on a planned Sabbatical in April 2020, leaving only 3-4 months to work on this.

Eventually the entire project was called off in March 2020 because the Public Service was testing out the use of chatbots for recruitment and it didn’t make sense for CSC to use a different system. 

REFLECTIONS:

This passion project to date has been one of my proudest moments in my career. I learnt a lot about chatbot development and spent countless hours working through the night. It was worth it.

Although I did not see “Rena” go “live” and to live up to her full potential, I was extremely grateful for the learning as well as support from my organisation in pursuing this endeavour.

I still believe that implementing a recruitment chatbot would transform our HR processes but I guess I’ll leave that for another time.