Business Intelligence

Business Intelligence for Recruiting: How to Analyze the Right Metrics and Optimize Your Search for Talent

The human resources area, in general, has a large amount of data; what can make the difference is how we use them to forecast hiring, dismissals, performance, etc.

The importance and awareness of companies in making data-based decisions is gaining more and more relevance.

In order to make better decisions, data is collected, interpreted and transformed into insights in increasingly broader scenarios in the company’s business, such as: product development, optimization of marketing expenses, improvement of the sales process, improvements in the talent hiring process, among others.
The competitive advantage of having more decisions based on data and less on judgments is evidenced in this article written by Andrew McAfee and Erik Brynjolfsson for the Harvard Business Review, noting that:

“Companies that usually use data for decision making were, on average, 5% more productive and 6% more profitable than their competitors.”

When analyzing the use of data in the context of recruitment, if we search Google for the term HR Analytics, numerous articles, courses and books appear that talk about the use of data and clear metrics within human resources.

Until recently, companies used to rely on intuition and common sense when hiring people. Today, it has become a fundamental aspect in the life of a recruiter to use tools that help in the acquisition of talent, so that the process becomes more and more accurate.

However, even with a large amount of information provided by these tools, few companies are willing to operationalize and organize the data and make the metrics used in day-to-day decisions quickly and efficiently.

In 2017, we hired 246 people in just 8 months, which represents an increase of 61% compared to 2016, when 228 people were hired during the year.

As part of the company’s culture, the importance of metrics was already very clear to everyone — one of RD’s values ​​is to be data-driven, that is, data-driven for decision-making.

However, the rapid changes in processes undermined the quality and reliability of the data, resulting in excessive time spent on solving simple analyzes and even making those considered more complex unfeasible.

This post describes how the process of implementing Business Intelligence was implemented in the Talent Management team at Resultados Digitais and the results that have already been achieved in the first few months.

Before building graphics, understanding the problem is essential.

Before we started measuring in a disorganized way, it was necessary to understand the entire recruitment process: how the interactions with the hiring software are, how and where the data is stored, what the bottlenecks are, and what questions the data can help answer. For this, there was a constant exchange of knowledge with all stakeholders (recruiters, managers and people analysts).

After the general understanding, it was defined that the first project to implement Business Intelligence in the talent team would be the creation of a metrics panel with the objective of making decision-making more accurate, based on data.

In addition, with the help of BI software, it becomes easier to analyze the performance of the recruitment team and also analyze bottlenecks, define key performance indicators (KPIs) and objective and key results (OKRs).

The first step in building the dashboard was to understand in more detail the real pains in the area and the main opportunities that could be generated, both in the short and in the long term.

After some studies and benchmarking, Blue World City developed a questionnaire to guide us in this process:

  1. What is the purpose of the dashboard?
  2. What is not possible to analyze today?
  3. What actions would you be able to take with the new dashboard?
  4. What analysis and information would be necessary for a decision to be made comfortably?
  5. For each analysis, which metrics would help to take the defined actions?
  6. How would you briefly summarize the story the dashboard will present?

With this step carried out, it is possible to have clarity about the situation of the area in relation to the analysis process, how the data will help to solve the problems exposed and even whether it will be necessary to change an existing process.

Thus, the next step is to understand the responses generated by the questionnaire and transform them into graphical views, also known as prototyping.

Prototyping is a quick and easy way where you can iterate with users showing the first draft of what would be the dashboard. It can be done either on paper or on slides.

As a guide for this step, we suggest that you use this flowchart to understand the best way to visualize each type of data you want to show.

Don’t expect the first prototype to look exactly like the final version of the dashboard. As we’ll see below, it’s normal to make changes, as users often don’t have a clear idea of ​​what they need until they use it on a daily basis.

Therefore, not too much time should be spent on this step, as its objective is to see what data will be needed to supply the charts and what the sources are (ATS, Excel, Google Sheets, and database). This step is already used to prepare and verify if it will be possible to obtain all the data for the development of the dashboard.

If the possibility of obtaining the data is verified, this step is completed with the validation by the area manager of the prototype and the carrying out of final adjustments, if something was missing. If it is not possible to obtain some data, process changes must be made and some views may not be prioritized in the first delivery of the project.

Recapitulating what has been done so far:

  • We understand the pains and opportunities in the area
  • We graphically demonstrate the opportunities we find to resolve everyday pain
  • We check if it is possible to pull all the necessary data for the charts
  • We carried out a validation with stakeholders in the area

Putting your hands dirty

We now know everything we need to do and there is only one thing left: to do. It is recommended to use a reputable BI tool in the market, as it helps with integrations, visualizations, usability etc.

Currently, the main BI tools are Microsoft Power BI, Qlikview, Pentaho and Tableau. All of them have a free version and, today, Power BI has trial periods of the paid version of 1 year.

In the case of Digital Results, Power BI was used due to its attractive price and monthly improvements to the tool, gaining prominence in the Gartner magic quadrant — a report containing a graphical representation segmenting BI software into four main quadrants: leaders, challengers, visionaries and niche competitors.

Building the dashboard

No technical issues will be addressed in this post (if you have any questions in the future, you can contact us by email at thiago.rocha@resultadosdigitais.com.br or thomas.dobereiner@resultadosdigitais.com.br ), but once you have chosen a tool and you have it Once the data is imported into the software, one of the longest parts of the project begins:

  • Interpret processes based on data;
  • Check the reliability of the data, clearing the data if necessary.

There are countless cases that may appear at this stage and it will not be possible to address them all, but let’s exemplify a process error that harmed our data.

In Lever, our Applicant tracking system (ATS) tool, we have a field called “Company” that is right below the candidate’s name. However, the field is editable and there is no indication of what information should be there. It quickly became a great recruiter’s notebook, and when the data was imported, we couldn’t understand why the Company field didn’t contain company names.

This is just a real example of Digital Results, but there are many important fields that people who are not in day-to-day operations cannot quickly understand. For this, the help and constant interaction between the BI analyst and future users are essential.

Another common process error that you should be aware of is when the advance in the candidate’s selection process step is not correctly marked in the software. This error can affect all efficiency and speed metrics in the recruiting process.

A common phrase that highlights the importance of data quality and the importance of interpreting it correctly is garbage in, garbage out.

If you don’t clean your data and you can’t trust it, no matter what you do, your results won’t be good.

You can try to increase the mathematical complexity and use predictive models for the analysis, but the gains will be marginal. The big secret of a world-class analysis process is having quality data.

This step ends with the certainty that the data is making sense and that there are no anomalies. Relying on the data, we can build our charts in the BI software, following the designed prototype and making necessary adjustments to complete the dashboard.

It takes more than graphics

The biggest mistake made by people working with data is thinking that the project ends with data modelling and launching the dashboard. However, after the development of the dashboard, one of the most important parts of the project begins: the implementation of that dashboard in the team, educating them to use it on a daily basis.

The first implementation test to be done is what you can understand in 5 seconds. In this test, recruiters are asked to look at each of the graphs and, in a short time, say what they understand about the graph and what action they would take to improve the result they are seeing—if applicable.

In our case, when doing this test, we’ve already seen that half of the graphs are not easy to interpret for people who don’t analyze data on a daily basis. In these situations we have two main alternatives:

  • Educate users so that they can understand the charts more clearly: this alternative can be inefficient, as most of the times the user does not understand, it is the fault of the Business Intelligence analyst, who could have been clearer;
  • Make the visualization form clearer: here you must understand the cause of the lack of understanding and find out how this could be visually clearer. Most problems are resolved with this alternative and we strongly recommend working hard on top of it.

An important aspect is that users always have support from someone who takes care of this dashboard, making changes (if necessary) and answering daily questions. For the company to be truly data-driven, users need to see the value that available information brings to their work.

Finally, we come to our dashboard! It was divided into three main views being them:

  • Strategic vision: details the area’s general indicators, not for the purpose of thorough analysis, but to provoke reflection and independent analysis. It provides the area director with a macro view of the hiring process with key metrics, such as: forecast, hiring number, candidate to hire (or CTH), among others. These metrics can be filtered by vacancy or by area.
  • Tactical view: shows a view for the recruiting manager and makes it possible to do some more complex analyses, getting a slightly more micro perspective.
  • Operational view: shows a view that will be used by the recruiter for daily analysis, for example, which vacancies will be filled, how the hiring pipeline is currently, efficiency indicators, etc.

Benefits of using data

With all these explanations and steps, you may be asking yourself: is all this energy and time worth it? What has actually changed?

It is worth noting here that the dashboard is not the end or the means, it is the beginning. Therefore, we separate the main results acquired at Resultados Digitais in the short and medium-term and what is expected in the long term.

Short term (1-3 months)

The main change immediately was the changes to predictability. We developed a forecast calculation that uses our efficiency across the hiring funnel and the number of candidates in the pipeline to determine what the expected number of people we will hire by the end of the month. It’s a clear indicator that shows how close we are to hitting our hiring targets.

Another point that is worth mentioning is that we were able to fully understand our process, identify bottlenecks, not let candidates get lost in the selection process and have a database for complex analyses. This was especially useful to give us directions on the projects to be developed and to measure the established Key Results.

Finally, we made changes to the candidate acquisition channel tagging process. With clarity of value gained from the beginning, the recruiters themselves began to police themselves so that there were no process errors that could damage the information on the dashboard

Medium term (3-9 months)

This is the stage we are in at the time of publishing this article. With all the databases ready and processes much cleaner, we had data to justify the creation of new projects and also to monitor if the existing ones are being well executed.

In addition, after an analysis of bottlenecks in the hiring process and with ever-increasing goals, several initiatives emerged to improve recruiter efficiency with process automation—expected to reduce time spent per candidate by 40%.

There was also an initiative to improve the quality of hiring and the satisfaction of candidates. All this was only possible thanks to the clarity that the data gave us.

Finally, we are starting some studies and talking to several companies to find out how we can take advantage of machine learning and predictive analytics to further improve our hiring efficiency, speed and quality indicators.

Long term (9-18 months)

We’re not here yet, but the focus is to improve our process based on benchmarking with world-class companies and insights generated with data, providing inputs to scale and train our Talent Management team in a more predictable way.

Conclusion

All the changes we’ve had so far are a tiny part of the potential impact Business Intelligence can bring to an enterprise. It is clear evidence of the benefits that well-defined processes and careful data storage can bring.

The results achieved so far were only possible due to two main factors: data and people. Clean data was essential to base our decisions, and people from all areas of the company were largely responsible for this by being attentive to the processes.

These results shown in the post were only for the HR area, but, as stated at the beginning, the possibilities of applying business intelligence are for all areas of the company.

Imagine if we left analysis within each area, now having an easy way of centralizing data to cross-reference information from all sectors of the company. What is the competitive advantage and the impact it could have on the business?

I leave this as a provocative thought so that your company also starts making decisions based on data — and less on intuition. So that your company no longer has the famous drawer reports and starts to have truly relevant information, being able to respond:

  • What are the causes of certain behavior?
  • What to do to reach the goals?
  • What are the current problems and opportunities?
  • Given the recent past, what will happen?

Make your company evolve, making it more and more data-driven. Peter Drucker, considered the father of modern management, already said:

“What cannot be measured, cannot be managed.”

And 60 years later we can complement:

“You can’t manage if you’re not data-driven. You can’t be data-driven if you don’t have reliable data.”