People Analytics Market Growth: Ten Things You Need to Know

I’ve been studying the use of analytics in HR and L&D for almost 20 years, and it has been a fun but frustrating space. While many have believed in data as a critical part of HR and L&D, most of the conferences I attended were filled with analysts, statisticians, and passionate professionals who struggled for resources and support.  And even though there are dozens of books written about learning measurement, engagement, assessment, and other people analytics topics, the real use of analytics in HR has been limited to studying employee engagement, looking at the impact of HR programs, and doing studies on retention and job fit.

Well all that has now changed, and today “People Analytics” as a business discipline has arrived.  Our research now shows tremendous growth in this market, and a significant shift away from measuring HR toward a real focus on using people data to understand and predict business performance.

To help you understand this change, let me share ten things driving this marketplace.

1.  Employee retention and engagement has become a high priority issue, driving the need to understand what drives the employee experience.

As critical roles in business become more competitive, the competition for strategic talent is intense. Our newest research (Deloitte Human Capital Trends 2016) shows that 86% of business leaders are deeply concerned about retention and engagement, 89% about leadership, and more than 84% about current workforce skills. And thanks to tools like LinkedIn, Glassdoor, and others, skilled professionals can find jobs easier than ever (the average worker now changes jobs every 4.4 years.)

Given this challenge, coupled with shifts in the workplace toward younger and more diverse teams, companies are searching for ways to build a highly engaged workplace. Our research shows that Glassdoor ratings (“would you recommend your company as a place to work”) for the average company are 3.2 out of 5 (around a C), with almost a perfect bell curve distribution. The reasons for the variations are complex, and it takes deep levels of analytics (plus a lot of local data) to understand why some people love their jobs, others want to leave, and some are on the fence.

Fig 1:  Glassdoor company ratings, 20,000+ June 2016

2.  Organizations are redesigning themselves around teams, and need data to understand how people best work together.

The second finding from our recent research is the astounding fact that 92% of companies believe they are not optimally organized for success.

Why?  Today, driven by digital technology, business moves too fast and people want to work in more dynamic teams.

Figure 2:  A Network of Teams

This new way we work implies that the org chart, while easy to understand, doesn’t really represent how things get done. We work in a variety of teams and these teams change often. Some are customer projects, others are internal programs, others are simply special assignments.

Today’s HR software gives us almost no visibility into these teams (this is changing), so we have to analyze relationships and use what’s called Organizational Network Analysis (ONA) to understand how work gets done.

ONA, a relatively arcane discipline used by organizational design experts, is now becoming mainstream. And guess who makes sense of the data?  People analytics teams.

Research by Rob Cross, a leading researcher in ONA, found for example that highly connected people are among the least engaged in a company. So some of your most valued staff, the ones everyone talks with regularly, are often buried in the organization and both underappreciated and over-worked. Only analytics will explain this to you.

Soon we can expect ONA type of functionality to come embedded in our email systems, so we will be able to see how we get things done, how decisions are made, and where we have bottlenecks. One company, Starling Trust Sciences, has actually developed algorithms that help us identify fraud and likely compliance problems simply by looking at who the trusted people are in the organization.  You can expect People Analytics to be predicting loss, fraud, accidents, and many other important things using these tools.

Fig 3:  Organizational Network Analysis (Images compliments of Rob Cross and Deloitte)

3.  The People Analytics function now has a clear mission.

For many years, the terms “People Analytics,” “HR Analytics,” “Talent Analytics,” and “Workforce Analytics” have all been kicked around. These changing labels have been confusing, making it unclear what we are trying to do. Are we talking about using data to measure the effectiveness of our HR programs? Measuring workforce productivity? Engagement?

Well today we have more or less arrived:  People Analytics means bringing together all the people data in the company (and there is an ever-expanding amount) to understand and address specific business problems:  sales productivity, retention, fraud, customer satisfaction, etc. It often means measuring the effectiveness of HR and L&D programs, but more broadly it means understanding all the employee data you have and its impact on business performance. And in some of the exciting new applications we’re working on, even measuring the marginal value of different roles, leaders, and other business investments.

 Fig 4:  Redefining People Analytics (Bersin by Deloitte)

As you can see from this picture, we are now looking far beyond the green “HR Data” sources of information. Organizations now have HR data, business data, and data about individual people at work. This latter category includes patterns of communication, location, feedback (ie. from pulse surveys), testing and assessment data, and soon heartbeat and other biometrics. While this all seems rather daunting and expansive to consider, it’s clear to me that this domain will continuously expand as we get better and better at building models.

One of our teams in Deloitte, for example, does people analytics for sports teams. This group of highly trained analysts can actually look at how much money a team is paying a player and figure out whether they are over or underpaying them. I believe over time we will be able to do this in many business roles as well.

There are quite a few implications of this redefinition of the role:

• First we have to consolidate HR and people data in a meaningful way. This means bringing together all the system of record data and creating a data dictionary so we can accurately and reliably answer questions like “what is our turnover rate” and “how many contractors do we have?”
• We have to look at many forms of HR data:  tenure, salary history, job mobility, location, training history, performance rating, and more – which means we have to pull data from many different systems.
• We have to consider psychological data like engagement, mood, and surveys. This is typically the domain of Industrial and Organizational Psychologists, who have to be part of the People Analytics team. Much of this data is often in small surveys systems or even on paper.
• We need to look at organizational network data, data about the organization itself (who reports to whom), and understand the role of structure, location, and team size.  (This is why I believe the organizational design discipline in HR belongs in the People Analytics team.)
• We need to look at external data (or data collected during recruitment) like job history, schooling, experience, and educational history.
• We have to be ready to look at new sources of data like location, travel schedule, commute time, and now even fitness, heartbeat, and more.

This means the people analytics function has to build an open and scalable infrastructure, and become very good at understanding data of many dimensions.  (“Dimension” in the context of analytics means looking at how the data is organized, its primary organizaing principles, and how we would structure it to mix with other information.)

The days of focusing primarily on building a retention model are coming to an end. While most People Analytics teams seem to study that particular issue, they quickly find that employee retention is really a surrogate for dozens or hundreds of other issues:  management, leadership, work environment, rewards, job fit, and more. And we now have retention predictors built into most core HR platforms.

4.  The People Analytics profession is now growing rapidly.  Staff skills and expertise are available.

Our research (Deloitte Global Human Capital Trends 2016) showed that the maturity, investment, and skills of People Analytics in HR has rocketed ahead this year.

 

Fig 5:  People Analytics Progress 

This growth is an outcome of many years of evolution. Today we have reached a point where many companies now have cloud-based HR platforms and they can now access reliable data in a relatively easy way. Yes, most companies still have very dirty, inconsistent, and fragmented data about their people – but the progress over the last five years has been huge.

We had nearly 30 people join us a day early at our IMPACT conference in April to discuss analytics for over two hours. I was very excited to see that nearly 30% of the attendees were from finance, marketing, or operations. They had volunteered to work in People Analytics because it was such an exciting new area.

Growth in this area has been a very long journey. Almost 20 years ago I started studying analytics in L&D and used to attend learning analytics conferences regularly. Those events, which had 100-200 people in attendance, have now been replaced by dozens of People Analytics conferences around the world, including conferences at major universities (Wharton and MIT).  UC Berkeley now offers a Masters degree in Data Science which is regularly sold out.

5.  Data Management still remains a challenge.

I’ve talked with many dozens of companies in this area, and almost all agree that their HR data is “bad.”  HR and people data is often inconsistent, unclean (not correct), out of date, and located in many places. Many large companies still don’t even know how many salaried or contract employees they have at a given time, so these teams are dealing with a big data integration problem.

Fig 6:  Bersin by Deloitte Talent Analytics Maturity Model

Several years ago we developed our talent analytics maturity model. While the data is slightly old (we are updating it this year), I’d venture to say that 2/3 of companies are still at levels 1 and 2.

Our research shows that most big companies still have 5-7 “systems of record” for various types of people and talent information.  So reporting and data integration remain a challenge (and this will always be true).

6.  Predictive modeling is becoming common, but really learning how to apply this information is very tricky.

We analysts love models: a great model that predicts retention, a model that predicts the right paths to leadership, a model that tells us how much salary to pay high performers, etc. These models are incredibly valuable … but as the book and movie Moneyball pointed out, “having the answer does not make results happen.” If you read or watched the move Moneyball, you remember how difficult it was for the Oakland A’s to change the behavior of their scouts and talent teams once the data became available.

I remind you that some of the most difficult challenges in People Analytics is implementing the changes recommended by the model. This means your analytics team must be surrounded with good change, HR, and leadership consultants.

Let me share an example: I talked with a large company that did an analysis on high potential turnover and found they were underpaying their highest performers and overpaying their mid-level performers. (This type of problem occurs often, and is usually created by an attempt to be “fair.”)

The analytics model clearly showed that the highest performers in this company needed to be paid at 120% or more of “market wage” to stay, while level 2 and 3 performers could be paid at 95% of market.

While this data was clear and compelling, this particular company had a 20+ year culture of “fairness” and “equality of pay.” So while the data was easy to understand, the HR team found it very difficult to teach managers to pay people differently. Ultimately the HR team had to change guidelines, systems, and communications to make this change “stick,” and it has taken several years for this new “meritocracy” to take effect.

One might think we can take useful data and simply share it with business leaders to get results. In fact, this can be very risky.  Let me share a few examples:

• Imagine you have a model that predicts retention, and you find that a particular engineer or sales person is a retention risk. What do you do with this information? If you tell a manager about this finding, he or she may disclose this to the employee and the person may feel spooked (and quit). What if the manager says “well if he’s going to quit anyway, I’ll just promote someone else.” Or maybe he bends over backwards to promote or reward that person when it’s really not needed. And how will he treat the others on the team?
• Imagine you have a model that predicts leadership and you tell a manager that he has someone who is a HIPO (“high potential”). Will that manager now act biased toward this person and assume everything they do is right? What happens to the other people in the team who may also be management potential?

The models we develop are dangerous when put in the wrong hands. We have to learn how to use them for systemic changes and very carefully use the data with line leaders. Remember most managers are busy getting work done, and they have plenty of their own opinions about who is performing well. Data should be “helpful” to them, or serve as a “nudge” – be careful you don’t give them a sledgehammer which might in fact break the team they have carefully put in place.

Remember also that human beings are very complex. While People Analytics models are useful and often predictive, but they are never 100% right in all cases. Think about using people analytics data like the “ice warning” light in your car during cold days – it tells us to be vigilant, slow down, and watch for problems. But it should never be used as a signal to “do something immediately.”

7.  Tools and platforms are here, but there are no Unicorns or Gorillas yet

As with all hot new markets, there are dozens of new tools, technologies, and platforms available to help you analyze people data. I wont list them here, but suffice it to say, none are Billion dollar companies (Unicorns), and none are Gorillas (read Geoffrey Moore’s Crossing the Chasm to understand that concept).

As Moore’s book points out, when the market gets hot (we enter the Tornado), the smaller companies that don’t have clear “market leadership” get shaken out, and big companies (like the big ERP players) take over.  Almost every major ERP software company is investing in this area (Workday, Oracle, SAP, ADP, IBM, Ceridian, Ultimate, Saba, Cornerstone, and more) – and there are dozens of tools companies (text analytics, retention analytics, sentiment analysis, and even companies that analyze your physical location, your heart rate, and your exercise). One very successful vendor in this market, Visier, actually does all the data cleanup and integration for companies and delivers dashboards “out of the box.”

Despite all this activity and investment, however, the market is still young, and I believe most of the analytics you will want to perform are not “out of the box” models you can buy from vendors. One manufacturer I just talked with told me they are analyzing unplanned absences and have been looking at commute times, birthdays, and many other work-related drivers. They found, for example, that one of the biggest drivers of someone not showing up for work is “having a performance appraisal the week before!”  I doubt any off-the-shelf analytics tool would have figured that out.

My advice is to seek out tools and platforms that are easy to use, extensible in their design, and can easily accommodate many forms of data. Remember that no vendor will give you everything you need, so look for a “toolset” you can use for many purposes.

8.  People Analytics is not a small, central team any more: it must extend into the organization.

When I started studying analytics 15 years ago, I would often find one person in the company who had this job. He or she was a statistician or HR technology specialist, and he ran reports, did analysis, and stayed busy trying to build dashboards that people could use.

Today the People Analytics mission is so big and broad that you have to think of it as a Center of Excellence. It is both a central group that understands data management, statistics, visualization, and reporting – and also a set of embedded business partners who can help line managers learn to use the insights you find. In one company the analytics group is fully distributed through business partners, and these business partners are being trained and certified to perform their own analysis locally (on centrally managed data).

Personally I believe this is the future of this domain:  right now we see heroes out there speaking at conference and delivering academic papers. Very soon this domain will become a core business discipline, and this kind of analysis will seem as common as looking at customer retention, profitability, and churn.

One more point as this market continues to grow:  while we need data scientists and statisticians to analyze and build models, the real discipline of People Analytics is multi-disciplinary. Successful teams include process people, consultants, OD experts, I/O psychologists, visual designers, as well as core IT professionals. You’ll need all these skilled people to work together to drive results, and I believe they should report to a senior executive (CHRO or head of operations), not be IT or HR Technology.

Fig 7:  The People Analytics Center of Excellence (Bersin by Deloitte)

9.  Build an open, extensive architecture to be ready for growth.

We have to think of People Analytics in an expansive way. The days of simply analyzing payroll, HRMS, and time and attendance data are over. That isn’t nearly enough.

If you consider the talent challenges I mentioned above, we really should look at employee engagement survey data, email history data, employee location and sociometric data (even voice patterns and who you have lunch with), and all the data which will come from wearable devices. Today employees are walking around with video cameras and GPS devices all day, so much of the data we will look at over time will be based on location, time, and visual identity.  (One vendor just announced a lathe that identifies its operator by face, and then automatically readjusts its equipment for the person.)

Think about the rapidly expanding world of employee and candidate feedback. Companies are changing the way they survey people, now opening up the floodgates to pulse surveys and “always-on” listening tools to understand what’s going on in the workplace. These new tools, which we discuss in our research, open up many new streams of data for analysis and predictive models. You have to think broadly in your analytics architecture. Topics like workforce absence rates, accidents, fraud, theft, and customer satisfaction are driven by dozens of people-related factors. The more source of data you have the more likely you are to find the right solution.

 

 

10.  Security, privacy, and confidentiality is now a critical issue.

Finally, we have to put security and confidentiality on the front burner. Thanks to the various credit card and other security breaches we have witnessed recently, employees and legal departments are nervous about the analysis we do. HR departments and People Analytics team have to take a crash course in data security, privacy, and identity protection. People are beginning to get worried about what this data is being used for.

Our research seems to show that most employees (72% according to Conference Board) understand that their employer is capturing data about their activities at work. The real issue you face is how can you manage this data carefully and securely, without alarming employees or committing a breach of trust?

My discussions tell me three things are important:

• First, tell people that any data you collect will be confidential and only used to help make their work life better. Let them “opt-in” wherever possible and don’t give employee-activity data to managers without very careful thought.
• Second, make sure your data team is well trained in the issues of Personally Identifiable Data and you have security measures in place to make sure the data stays private.
• Third, avoid using “anonymous” data in any feedback or analysis you do. While anonymity may make some sense on the public internet (although it is becoming less common every day), it only gets you in trouble at work. This tells people that when they fill out a survey or make a comment to someone else, the data is kept confidential but you do know where it comes from. Again this stresses the ethics and expertise of your internal team, but it’s the right thing to do.

People Analytics has finally become one of the most exciting trends in business, one which all companies can adopt. I welcome your comments and look forward to hearing about your experiences in this fast growing market.