From Prediction to Prescription: The Emerging Experience of People Analytics – An Interview with Balakarthikeyan Nagarajan of GE


Receive this article in PDF. By entering the email address below, you agree to our privacy policies. More info about privacy policies at the bottom of the page.

 

Balakarthikeyan Nagarajan is an Employee Experience Leader at GE based in Boston. He has extensive experience in a number of HR roles and a significant background in developing, implementing and people analytics practices. Bala is passionate about leveraging data and analytics to drive talent decisions in GE, and is responsible for creating intuitive and intelligent systems that offer “Personalized” learning, career and connection recommendations to employees. He is a regular speaker at industry and profession events around the world. To connect with Balakarthikeyan, please feel free to send him a connection request on LinkedIn.

 

Ben Churchouse: Good Morning Bala, it’s great to have an opportunity to speak with you. You’ve spoken with us at some of our previous events – the 2015 Workforce Analytics summits in Singapore and New York – what have you seen changing in the people analytics profession since then?

Balakarthikeyan Nagarajan: I would compare it to the Gartner’s hype cycle if I may. Around 2014 – 2015, the whole people analytics effort was at peak power, the peak of inflated expectations. There was a lot of buzz, people were talking about it, they were wondering what it could do, and in many cases it was hyped up more than it should be – from then on I think there’s been a slight fall. However, I think is probably a good thing.

Gartner Hype Cycle for Emerging Technologies, 2017

If I look at the next few stages, that’s where we’ll get into the enlightenment stage, and finally into what’s called as the plateau of productivity. In simple terms I’m saying that a lot of the hype around it is starting to vanish and people are starting to think about what is the real value for the HR function and as a business. This is also probably driven by one big change, which is that the function is now starting to open up a little bit more. Back then, it was essentially reserved for the geeks, for the data scientists and the PhD graduates, and now it’s starting to become less dependent on individuals and is evolving as a function. That’s a great thing.

BC: What are some of the most common current misconceptions that you see from HR and business leaders about the adoption, especially in the development of a predictive or even prescriptive people analytics function?

BN: I think the function is seen as something that will give magical solutions to people. You know, right from the get-go you launch something and people think that the recommendations are going to be perfect and everything is going to be in place, and that’s as far from truth as it can be. Whenever we are developing these predictive or prescriptive models it needs a lot of patience. In the first few iterations it’s not going to behave well, it’s going to be like a child that needs a lot of nurturing to help it evolve over a period of time. Occasionally it might present some virtues that sound like wisdom to you, but that’s not going to be consistent at least in the initial phases.

Over a period of time when you strengthen your feedback loops, when you start using it in a broader scale, then these systems are going to be giving some good consistent insights. What we also see is that a lot of people especially in the HR function and the business leaders think that the models should be decision making systems. So, they’re concerned that it will substitute the traditional way in which the humans make a decision, and that’s not going to be happening any time in the near future. In fact, with people analytics I would say, it should not happen ever. These systems will continue to exist as decision support systems and not decision-making systems.

For example, we are currently working on some models around identifying who are the top talents in the company and who we should develop more and invest further in. For the first iterations, you’re not going to get it right, but with continuous effort and patience you would start to be able to make recommendations, or provide options to people in terms of who is the top talent – therefore supporting the conventional HR system.

BC: I know that you’ve been looking into using data support systems for making better talent decisions, maybe you could tell me a little bit more about why this is such a challenge?

BN: Honestly the challenge has never been around developing the model or creating a solution. The challenge has always been around adopting it in terms of changing the behavior of the HR function or the people leaders – and by people leaders I’m talking about managers and business leaders who make a lot of day to day and long-term talent decisions in the organization.

Traditionally we have depended heavily on experience, assessments, and to some extent even perception and gut instincts. So, moving from that to data-based decision making, or data enabled decision making is a big change in behavior and that’s where the challenge is. It’s even more important when you have intuitive thinking that is being questioned with data, that’s where the propensity to shrug it off comes into picture.

For example, we are doing some work around recommending successors for key roles in the organization, we are using some predictive analytics, and in most cases you walk into those discussions having some pre-conceptions of who could be a successor and who is unlikely to be a successor. When the model gives you a different set of options that you had not considered or you had ignored, then the propensity to take it seriously and work with it is a real challenge.

The other part is explain-ability. When you have data, the traditional knowledge is to say that ‘okay I have data and I should be able to explain the recommendations’ but in some cases where we are using things like some machine learning techniques, it’s difficult to really explain the final outcome because there are multiple factors, multiple dimensions in those models and so the change in behavior along with the ability to explain recommendations is a challenge.

BC: As we start to see AI and robotics become more and more prevalent, I think it’s very important for people to understand that natural human flaws and biases are here to stay, but people analytics can potentially minimize some of the grey areas. When we’re thinking of HR leaders making important decisions about the business and the lives of all stakeholders, what do you see as the evolving reach of the function?

BN: The topic around how do you minimize bias using people analytics is important, but perhaps more important is how you ensure that the human bias in our decision making doesn’t creep into the analytics that we are developing. It’s a big area and we are heavily focused on ensuring that our models are truly agnostic of these biases.

One example I can give around how these people analytics systems are going to help us is from the space of attrition management. What happens typically when a manager is looking at employees who are potentially at risk of leaving the company, in most cases the HR leader is influenced by maybe one or two factors about the employee and because of these two things, they remark that ‘I think this person is at risk of leaving the company’. So, let’s say, they’re not getting the promotion that they were aspiring for, and so the manager starts thinking, okay maybe this person is at risk of leaving and starts to take some actions to engage the employees better. When you look at a model, it doesn’t go by just one or two of those factors. It looks at multiple factors, and it also looks at how employees across the organization have behaved in those situations. It would look at if somebody has applied for a job internally and hasn’t got it, then what is the likelihood that similar people have left. It gives a multi-dimensional view of the attrition risk, which is something that the individual is not able to do. And in many cases the individual is biased by what their opinion of the risk is, and that I think is the biggest advantage of using these models.

In some rare cases there might be one or two critical determinants on the employees’ propensity to leave or stay – which the manager might certainly know about. In those cases it’s important that the manager recognizes that the model layout building is actually a support system and should therefore be comfortable over-ruling it saying, ‘okay I get what the model is saying, I have taken some snippets out of it, but this is my decision and this is what I’m going to do’. That’s where multi-dimensionality plays a huge part in limiting or reducing the bias that starts creeping in.

BC: A lot is being written on the emergence of AI and intelligent learning systems, and how they’ll impact HR. We’ve spoken in the past that we should have a lot more patience with the self-learning models. How much patience is too much when it comes to self-learning?

BN: It is very critical that you have patience, but patience should not mean that your progress is slowed down and one way we do this effectively at GE is we start really small and we try to fail fast. We also learn quickly and then look at scaling once we’ve learnt enough, once we have grounded the risks we try and scale it really fast.

For example, we were working on integrating our performance management system with our online learning system, so when an employee gives development feedback, we consider it as an insight within the company. Along with that feedback, we’re trying to recommend learning assets so that when you get feedback, you’re not thinking about how to defend it, but you’re starting to think about how to use that feedback to grow further and develop yourself. In this case we’ve made some very basic assumptions that we wanted to test, such as whether we are even able to make some relevant recommendations, and secondly even if we make the recommendations, are people willing to go and look at it or leverage it for learning? So before we went ahead and invested in a proper system, what we did was first do it offline, using paper tests, in some cases emails to validate some of these assumptions and we did it in three or four cycles.

So, every time we tried to also improve the relevancy of these predictions and once we came to a stage where we felt comfortable enough in terms of being able to make relevant recommendations, that this is actually resulting in some change in employee behavior, then is when we started moving it into a full-fledged production into our systems. What we found is when you make learning recommendations at the right moment that matters to the employee then the learning actually goes up three times than it would have otherwise. And that’s something that we have used and I think that’s a good example of how do you be patient but also you progress with the right speed and ground your risks as you go further.

BC: Digital HR has become a big trending topic. But one of the things I’ve always been skeptical about is what actually Digital HR is? It doesn’t seem to be very well defined at the moment; it’s a very broad umbrella term. I think we’re going to see people analytics become even more widely embedded within the “Digital HR” super-structure. How do you see digital HR being influenced by data science, and what is the relationship between people analytics and the emerging ‘digital’ trend that we’re seeing at the moment?

BN: There’s lot of effort and activity happening around digital HR, and slowly we are seeing the signs of it impacting business in a very real way. Everybody sees it from their own lens, so I’ll restrict it to how we approach Digital HR and how we leverage people analytics. For us it’s about what impact it is making to the employee. That’s very important for us given our huge focus on employee experience. And then taking it one step higher, what impact does it have on people leaders or the HR function as a whole, and then what impact does it have on the organization in terms of the business impact?

I’ll just talk about it briefly at the employee level, where we see it’s primarily about hyper-personalization. How can you use data science and the digital tools and technologies that you have to really give a personalized experience to the employee? For example, I talked about how do you get learning experiences and assets to employees when they are in need? That’s one example of personalizing it. And then the other example is around serials. We think the days of the traditional serial path as disappearing, and people are more thinking about what’s the next immediate experience. In those cases, how do you showcase the next big opportunity for the employee using a set of digital and analytics tools so they understand what the organization offers is critical? That’s what we mean by personalization.

For the organization and the people leaders, we think more than prediction, prescription is important. Let’s take the analogy of health. You’re not really interested in knowing when your blood pressure level is going to go beyond the limit. What you need more importantly is – what should I do to prevent it from rising too high? So far, we have focused a lot around predictive analytics, that we’ve been playing this game of “hey you got it right” or “hey you did not get it right” type of thing. You need to move the needle from there to saying; what are the things that our people leaders or HR leaders must be doing for the employees for talent, for the organization?

So, for us it’s about focusing more on prescription. Getting this information into the hands of people leaders and HRM’s so they know in their busy schedules they know when to focus on what. Those are the two things, personalization and prescription that we have focused on heavily.

BC: How important is effective communication within predictive and prescriptive analytics? As we know, people are not always the best communicators; sometimes we all make mistakes, miss our communication cues and forget to copy someone on an important email. So, how important is effective communication and what are the skills that are required for HR and analytics professionals moving forward?

BN: That’s a very important topic for discussion, and we can talk about this for hours together but in a nutshell, I would say that the technical expertise is a given. The challenge or the area where we really want a data scientist to focus more on is listening. So, when we started this podcast, I mentioned that there are a number of professionals coming from a non data science background who are moving more into this function. So as a data scientist you should be able to listen to their perspectives and then be a data whisperer, translating your findings through data into a more understandable language to your business partners in HR. I think that’s where the criticality is and that will really help in getting this function moving forward.

The other aspect is that there’s two types of data scientists; people who are solution lovers, and people who are problem lovers. The one that works very well are the people who love the problem rather than the solution. The solution lovers typically tend to be experts in one or two approaches and they try to hit all the nails with the same hammer and that is a big challenge. So, we are looking for people who are really problem lovers.

BC: How does an organization find the right talent to take their function further? There’s no shortage of talented technicians who possess the skill set, but what will set them apart? HR is becoming this big melting pot of data scientists, generalists and organizational psychologists – all kinds of professionals being mixed into one. So, what are the things that an organization will look for in particular when it comes to the people who will transform HR?

BN: So, I just talked about the skill sets that we really value in the data scientist community and that holds good. What really attracts these data scientists to come and work for GE or for any other company is giving them tough problems to solve. That’s the number one thing that we have seen keeps them excited. On top of it if you have good data, they’re just going to be excited about it. At GE we had our HR systems in place like 20 – 25 years back and now we’re looking at moving to cloud, but if you have good data they’re excited, if you have great challenges they’re excited, and those two things keep them really engaged in the organization. But that’s more on the data scientist side.

The real thing that we want to focus more on is on the rest of the organization – the rest of HR – which needs to interface with these data scientists well. So how you feed data scientists, how you ask the right questions, and when do you go to them is where most of our capability development and training is focused on at this stage.

BC: I think we will see a number of organizations drifting towards an HR wide analytics function, where it’s a part of a core competency for everyone. It’s very much the future, and it’s an attainable future given the availability of training and development programs for HR and business professionals currently.

BN: Let me throw you on the spot here, you’ve been asking my view so, let me ask you, what is your advice to the people analytics leaders of the HR function? You have the real benefit of talking to folks across many companies in various forums, so what is your advice that you would give for people like me?

BC: I think the most important thing is to keep asking questions. I know that sounds really basic, but having spoken to a lot of HR and business leaders over the last couple of years, the most important thing is to be inquisitive and to always keep searching for new ways of looking at things. In my view, business is fundamentally about asking questions. At the moment, there’s just so much great material and content out there that can help you find the evidence-based answers to a huge spectrum of questions, there’s almost no excuse these days not to be inquisitive – especially when it comes to emerging management models, systems or tech.

Perhaps more specifically towards HR, I would look focus on really understanding the business itself. Understand who it is that you are working for in terms of the people around you, and the goals that the business is setting. The HR business partner model has been very well established but, in some ways, I still feel that a lot of HR professionals and people management professionals don’t fully understand the ecology in which the business operates. That’s probably the most important thing for HR; understand the business, understand the people that are in the organization, understand the consumers of the business product or service.

BN: That’s very well said, it’s always good to get a perspective from someone who is external – but really sitting on the fence and seeing what’s happening in the field, thanks!

BC: I think what we’ll also really begin to see is the rise of employee experience not simply as something that’s nice to have, but the end in itself. The goal for most businesses is to create profit and value for shareholders and stakeholders, but at the same time now we’re beginning to see the currency of experience emerging. It’s a view which is a bit of a luxury to have in the developed world, but I think many will start to see how people in the company are the real core of driving value creation.

So, to wrap things up, you’re going to be speaking at The HR Congress Brussels in November. Can you tell us about the presentation that you will be presenting this year?

BN: I’m very excited to be part of the HR Congress this year! I was just looking at the lineup of speakers – an incredible range of HR and thought leaders in this space. I appreciate the opportunity to share some of the examples and learnings from our own journey within GE, so you can expect to hear a lot around what we have done, what we have learnt and how we have pivoted where required, specifically around areas of employee experience, that’s something we believe a lot in. In fact, I personally believe that the best way you can add value to the business is through your employees, giving them the right experience, having the right talent at the right time. That’s the best way you can add shareholder value and that’s the expectation from HR function. I’m really looking forward to meeting some wonderful thought leaders and practitioners in Brussels.

BC: Indeed, and I’ll look forward to hearing your presentation! It’s been a pleasure to speak with you today!

BN: Absolutely, thanks for the opportunity, I hope the listeners have enjoyed this and have gained some insights. I’d be interested to hear some feedback and thoughts as well from them.


Balakarthikeyan will be speaking at the 3rd HR Congress in Brussels on 27-28 November 2018.

2 Comments

Add yours

+ Leave a Comment