Just last week we attended Beyond Conventions where we took part in a pitch challenge in the competition for a HR Chatbot for ThyssenKrupp. Among other things, this event is proof that it is definitely necessary to develop new ideas and innovations in order to stand out as an employer among all the sought-after qualified employees. Therefore, in order to do a good job in employer branding, companies need to understand that you have to be present and active on all social media channels. As Generations Y and Z get older and ready to enter the business world, it becomes more natural to find out about career opportunities not only through the website but also through social media such as Facebook Messenger about the company. We will be publishing a series on demos, concepts, insights and much more on HR chatbots.

In this blog post we explain the showcase we developed for our HR Chatbot Pitch for ThyssenKrupp. You'll learn how to perform one of many application processes in Facebook Messenger. This process includes collecting information about candidates, asking screening questions and placing candidates.

Step 1: Selection of a job

 

Applying for a job with a chatbot within Facebook Messenger

 

  • First of all, you have to create an intention that starts the application process. This should also be integrated into the navigation as a quick response (button).
  • The next step is to integrate a job overview as a WebView of the respective website so that users can select their preferred job to apply for.
  • So how does the chatbot know which job has been selected by the applicant? Firstly, all the website URLs of the jobs have to be entered into the Whitelist on Facebook have to be entered. By integrating the Messenger Extension SDK on the website, we can enable communication between the bot and WebView. So when the user clicks on a button, e.g. "Apply now" within the WebView, an action can be triggered that closes the window and sends the job the user wants to apply for along to the chatbot in the parameters.

 

Step 2: Collect information

 

A chatbot collecting information from a candidate

 

  • The first step is to ask the applicant for the CV or other relevant documents. It is important to know that users are not able to upload a PDF file on Messenger via mobile phone. It can only be transferred via the desktop. However, you could offer users to upload a picture or screenshot of their CV. Of course, this must be clearly communicated to the user in advance.
  • Once the candidate has uploaded the files, we recommend integrating an API that can automatically extract all information such as contact details about the candidate. For example, we use the Vision API from Google to extract the text from the uploaded image and then we search for specific Entities via Google's NLP APIs to find the individual information. You can also implement a method to request missing information if someone forgets to include it in the documents (see example image).

 

Step 3: Questions on the screening question and ranking candidates

 

A chatbot asking screening questions and ranking the applicant

 

Ask screening questions

The 3rd step is to actually see if the candidate is the right one for the position by testing how well they match the requirements of the job advertisement. In this case, one can use already existing requirements from the job offers. This sounds like a lot of work if you have to insert all the requirements from thousands of job postings into the chatbot, but there is actually a very scalable way to automate this. Just by integrating an API into a CMS or other system where information is stored, you can use it immediately in the chatbot.

Tendrils of Candidates

To be able to rank the candidate, one can create a scale from 1 to 5, where the user has to choose a number, how well the candidate meets the requirements. Choosing a 5 means that the candidate meets the requirements by 100% and a 1 obviously means the opposite. After the candidate has answered all the screening questions, the chatbot calculates a score in the background of how well the job matches his qualifications and skills, which gives the recruiter the possibility to compare applicants . By the way, it is also possible to give certain requirements a different weight in the score calculation. For example, some requirements might be more important than others, e.g. 5 years of experience in programming might be more decisive than a degree in computer science with a scarce background.

In the final step of the process, the user is asked by a recruiter to upload a personal video of them explaining why they want to work for the company (or another question, related to the company culture and values ). This creates a personal experience for the potential employee, which can also lead to a unique branding effect. The candidate's video is the perfect opportunity to create a good impression and show whether they are a good fit for the company.

 

Step 4: Integration of the Chatbot application into the regular recruitment process

 

All information from the application will be send to the recruiter or the ATS

 

After the user has finished the process, the chatbot collects all the information (see picture above). Now there are different ways to deal with received applications. It can either be sent directly to a recruiter via email or integrated into an application tracking system (ATS). The score given by the screening questions could help the recruiter prioritise applications.

I hope this has given you a good insight into how to create an application process for a chatbot. There are obviously many other ways you can set up the process. Please share your thoughts in the comments on how it could be improved or optimised. This was also the first blog post on HR chatbots.

HR Excellence Award 2018 in HR Tech and Data

 

BOTfriends' Jobs & Career Chatbot for Porsche won the Human Resources Excellence Award 2017 in HR Tech & Data.

If you want to know more about voice and chatbots and how Conversational AI projects can be successfully implemented, then get our white paper: