The recruitment process is one in which people play the key role. But some activities are just very hard to do effectively with our human brains. In this blog we give an outline of the different technologies used in Intelligent Recruitment Automation.
In our earlier blog we outlined the The Rise and Rumble of Intelligent Recruitment Automation. In this blog we go through the technologies that are used in Intelligent Recruitment Automation.
| Part 2 of the blog series on Intelligent Recruitment Automation
The recruitment process is one in which people play the key role more than in any other business domain. People that you potentially want to hire appreciate it when you resonate your human side as an individual and as a company as a whole. If we look at outreach for an example, the most effective outreach is still almost always one with a creative and personalized approach. But some activities are just very hard to do effectively with our human brains. Take processing millions of talent profiles, filtering big sets of data down to relevant information and remembering (or storing) data.
A key characteristic that useful intelligent automations share is richness of data. Data obviously helps to make better decisions, but just data doesn’t bring you anywhere. You need the right technologies and the right application of those technologies to make sense out of data and put it to work effectively. Below is an outline of the different technologies used in Intelligent Recruitment Automation so you can better apply these technologies yourself and to make a better informed decision in working with third party suppliers.
Artificial Intelligence (AI) is an umbrella term of several artificially intelligent technologies like Natural Language Processing (NLP), Machine Learning (ML) and Predictive Analytics. These technologies are generally aimed at mimicking the smart things we do with our human brains. Artificial means unnatural or made by human beings. Intelligence literally means the ability to learn, understand and make judgments based on reasoning.
In all industries AI is changing the way people create, grow and run businesses and support their work related activities. In the recruitment industry, AI is getting more and more popular and fortunately not only more popular but also more effective.
The increased accessibility of these technologies is driven by more clear applications for AI and an increased user friendliness of third party tools. But it is important to say that intelligent automation technologies do not always yield the desired results in recruitment. Highly generalized algorithms can be causing depersonalization and bias, the black box effect can cause machines doing things that are hard to trace back and control and some systems simply break because of technological and process dependencies.
Because AI is often overgeneralized in all the hype it is not commonly understood. Below we have given our description of the underlying technologies of AI which we hope can help to better understand what they are.
Data scraping is a general term referring to the extraction of data from external online sources. A common type of data scraping is (web) crawling which refers to the extraction of data from external web sources (websites). Arguably not the most intelligent technology, but it is an important requirement in getting the data necessary to do intelligent things with. Data scraping is done at a large global scale. Companies use their own technologies or third party tools to scrape data by requesting data from the target source (website or database), parse and extract the data (like job titles and descriptions) and download the data to their own systems.
Job platforms and recruitment agencies use small scale data scraping to find job posts and post these on their own job boards. But data scraping is also done at a large scale, take Google who crawls (almost) any website to make the information from these websites searchable in Google's search engine. Google uses similar technology to make jobs available in Google for Jobs.
Natural Language Processing (NLP) is a set of technologies used to analyse large amounts of human language data in the form of text, speech or signing language. NLP can extract, categorize and further analyse text and speech. As humans we speak and write in English, Spanish or any other language. But a computer’s native language is code, incomprehensible to most people. So for a computer to understand human language there has to be a technology translating human language to computer language (code). That’s NLP.
A popular use case for NLP is the use in chatbots. Chatbots ask certain questions to customers, receive replies from them and can understand the replies from customers to reply with logical follow-up questions or triggers. Recruitment chatbots ask candidates questions based on the recruitment procedure of the hiring company. Recruitment chatbots work very similarly to the customer support chatbots you talk to when you have a question about a product or service.
Machine Learning (ML) is the field of computer algorithms that learn from users using data. An ML usually starts with a set of training data like some defined categorizations (most commonly referred to as classifications). When the ML is trained, it automatically starts recognizing keywords and patterns so it can attribute classifications and scores by itself and help humans (like you!) in making decisions. An ML can for an example start recognizing patterns in keywords mentioned by let's say engineers in their profile and link these keywords to certain tech stacks and technology frameworks. ML can be useful in many more cases like analysing millions of candidate profiles available on the internet, suggest different terminology used in job descriptions and see patterns in profiles of successful candidates. Most companies don’t have the ML expertise in house and don’t have enough data to make it work so they make use of third party tools.
Hiring companies use third party tools like Textio to analyse the text of their own job posts and have the tool give suggestions for alternative words that can be used to better resonate with the target candidates for the job.
Predictive analytics is the use and analysis of historical data to help inform future decisions. For predictive analytics a set of techniques is used including the above mentioned technologies and additionally statistical frameworks and techniques. A highly sophisticated form of predictive analytics is used in self-driving cars. But in recruitment there are also plenty of use cases like job change prediction, candidate pipeline predictions and future hiring needs assessments. The word prediction shouldn’t be taken too literally because most of the times it will be more like an estimation than a prediction because calculations are made based on generalizations on data sets and plotted to individual data points.
Hiring companies use third party sourcing tools that use predictive analytics to predict the likelihood of a potential candidate to switch jobs. The probability is given for the candidate to leave their current job in for an example the upcoming 3 months. This is done by analysing historical data on for an example average tenure in tech companies and pattern recognition in the candidate’s job switching habits over time and doing predictions based on this historical data.
Marketing automation is a set of technologies used to automate the process of reaching the right people, on the right platforms and with the right message. Marketing automation can be interpreted very broadly and can include anything from programmatic advertising, to automated messaging all the way up till data automation of the recruitment marketing funnel. Because there’s a strong correlation between marketing and recruitment nowadays, automation techniques used in marketing for commercial purposes are very similar to those used in recruitment marketing. For these reasons the term Recruitment Marketing Automation makes sense. Just like in any marketing related activity you are building a relationship with a lead (potential future hire). Therefore you need to share the right message at the right time to the right person and adopt marketing automation technologies that don’t eliminate personalization and creativity in the process. Most marketing automation is not very intelligent in itself and primarily rules based. But combined with previous technologies it can be an intelligent application.
Recruitment agencies use LinkedIn and email automation tools to reach out to candidates with a semi-personalized message at a large scale. They use tools to send a message in bulk to a preselected set of candidate profiles based on a message template in which they use semi-personalization like [first name], [industry name] and [role] and sometimes even images of the candidates that are automatically pasted in their message template (stalkers!). Unfortunately these automations often are used in a wrong way causing depersonalization which can cause frustration on the candidate side.
Many of the technologies used in Intelligent Recruitment Automation are complex and the applications of these technologies are still relatively immature. It makes sense to start small and be specific when you talk about automation. Having a goal to do "something with AI" is close to saying you're going to lose "some weight". You have to be more specific when you're talking about automations and AI. Educate yourself on the possibilities but also start applying recruitment automation intelligently at small scale. In the next and final part of this blog series on Intelligent Recruitment Automation we'll dive deeper in the potential use cases and examples of available tools in the market.
Recruitment agencies can be expensive and it's not always clear what to expect, the alternative is to source and recruit talent yourself. This is how.
Writing a good outreach message seems simple, but it's one of the most crucial steps in the recruitment process. So here you go with some message examples that work.