Many companies worldwide have already passed the stage of HR-processes automation, and now they are facing the task of extracting additional value from the data accumulated in the automation systems.

The first steps are usually rather simple reports on the basis of this data, which then becomes more complicated. The next logical step for such companies is predictive analytics, the economic effect of which can be many times stronger.

Compare how a company can benefit from departmental turnover reports and forecasts of employee departures.

Think about this:

Back in 2016, only 32% of employers were ready to build a predictive analytics management model but 2018 has already seen that figure rise to 69% with companies actively taking steps to improve the way they view people data.

Despite a clear trend, there is, of course, a question of the size of the company and the amount of data it has and keeps accumulating.

In addition, the effect of improvements in business processes will usually be a tenth of a percent, if it has been optimized enough, respectively, you need a sufficiently large scale to pay off such projects.

In addition, it is too early and useless to start doing predictive HR-analysis until the company systematically collects data on employment and work of employees, has not put in order the data on vacancies, resumes, positions, KPI and employee evaluations in electronic form, does not count on them on a regular basis metrics with the use of conventional HR-analysis, has not started to make and apply specific decisions based on the results of conventional analytics.

The main limiting factor of HR-analysis development, in general, is the immaturity of the market, which, fortunately, gradually ceases to be a problem, as our industry is developing. This is reflected in the fact that there are specialists and numerous training events on this topic.

For example, the largest conferences for HR-specialists in the last few years are focused on the “digitalization“.

Accordingly, now the application of HR-analysis and the development of solutions on the basis of HR are already must-have for large companies.

Similar to pricing automation in e-commerce, today it is impossible to imagine, for example, a large retailer that does not use this approach – such a company will be uncompetitive and its losses in recruiting alone will amount in significant profit shortfall.

Who is it for?

In my experience, predictive analytics is usually used by large companies, first of all, from retail and banking, because it helps them to achieve a competitive advantage in the objective conditions of high staff turnover and large budgets for selection.

It should also be added that the traditional innovators in the HR-technology market are IT-companies, which also use different analytics, but here the effect of its use is not so tangible compared to retail and banks.

Individual companies in many industries are already starting to apply HR forecasts. As soon as it starts to affect the profitability and margins of a given business due to the fact that some of them attract and retain the best specialists much better, while others have to work with others, a “boom” will begin.

The HR-forecast boom will be connected with the mass spread of technologies based on artificial intelligence (AI) and machine learning.

In fact, the revolution is already happening, but it is not noticeable: AI is winning various HR fields and already makes many decisions instead of a person: who to call for an interview, to which training course to send an employee, who to keep and who not to. Unfortunately, at this moment it will be expensive to catch up urgently and it is too late for many catchers.


How to implement it?

In general, any company has several options for implementing such projects, whether on its own, through contractors or using off-the-shelf products.

Of course, there are mixed options as well. The cost structure will depend on which strategy the company chooses.

If a company believes that the scope of predictive analytics is one of its competitive advantages, it seems worth implementing the project on its own. At such an approach the basic expenses will lie on IT department specialists, including, analysts and data scientists, and also on hardware, especially if it is planned to use the big data.

For example, the companies can be important training of the sellers then it is possible to construct predictive models for the definition of the most suitable training courses.

If you buy an off-the-shelf product, the main costs will be spent on the product itself and its integration. The most common use case is the use of predictive analytics modules in HCM/HRM systems, for example, to predict layoffs and assess the damage.

A company can implement such projects on its own, but in any case, it will need people with the appropriate competences, who need to be taken either from the market or from the departments that deal with analytics, most often it is the IT department. In addition to consultants, it can be useful to attend conferences to exchange experiences.

To familiarise yourself with the features and costs I recommend to stud off-the-shelf solutions first: HR Analytics Software Or better schedule a demo or two 🙂

On a company level:

1) Get over old habits

In order for the tools of predictive HR-analysis to work as efficiently as possible, first of all, it is necessary to change the habit of HR people (especially executives) to make each decision manually. They consider it a foundation for their recognition, authority, irreplaceability as specialists, as it justifies their large budgets and numerous subordinates.

First of all, we should start counting ordinary analysts and KPIs, admit, but do not accept the fact that the current values can be improved, the obvious omissions and bottlenecks can be recorded. Then set such KPI target values for which it will not be obvious how to achieve them without a predictive HR-analysis, and then do not let “forget” or “fill” them with resources.

2) Set up infrastructure

As I mentioned above, it is important to have information systems that accumulate data. The more such data a company has, the better its predictions will be.

Both data quantity and quality are important. The data should be “clean” and it often takes more time to clean, prepare and normalize the data than to develop predictive models themselves.

In addition to the data, people and tools are needed that can benefit from them, for example, through machine learning. At the same time, the company can develop models on the machine learning independently (this is actively done, in particular, by banks), and work through outsource providers.

Can your in-house guy fixing computers and resting passwords do that? Probably not.

As an example, we can consider the task of ranking or scoring candidates for a particular vacancy. The company can make such a model on the basis of previous years’ data and cut off at an early stage those candidates who are not suitable and communicate more quickly with those candidates who are suitable.

Optional: continuous fine-tuning

It is important to consider that the lifecycle of models is just beginning: the company needs to continually assess whether it is worth investing additional resources in improving its models to make them more accurate and efficient.

Often such development can have an even greater effect than the initial implementation. I also refer here to the possibility of reusing models to solve tasks that might not even be expected when creating them.

Here are two examples from our practice.

We mentioned above a clever search and recommendations for resumes: while developing our Virtual Recruiter service, which provides lead generation of candidates for mass positions, we realized that we could add our machine learning models, which were previously used for searching and recommending resumes. Almost free of charge. Thus, our product gained an additional competitive advantage.

A similar story was with ClickMe’s advertising service, where we added ready-made models to search for and recommend vacancies.


To sum it up on predictive analytics in HR

Given certain company size and ability to adapt predictive analytics used for more efficient human resources management can be huge.

It also seems to be hindered by the fear of new things, something that requires us to recognize that everything is not perfect, to review the processes that have developed over the years with all their compromises and the status quo.

And even to realize that it is necessary to take off habitual, simple, half-life executed and perfected to shine routine which the programmer can automate for half a day, and to do something more useful and meaningful.

Written By
Dmitrii Borodin is the CEO and founder of GRIN tech – a full-service digital agency doing design, marketing, and development alongside white-label agency & media outreach solutions.

Related Post

Human Resources Today