In this episode of the Digital HR Leaders Podcast, David Green is joined by people analytics visionary, Prasad Setty, to explore the proven frameworks that drive impact to the transformative potential of people analytics in HR. 

Drawing on his extensive experience, Prasad shares actionable frameworks and strategies to help HR leaders turn insights into outcomes, tackle adoption challenges, and embrace AI’s transformative potential. 

Join the conversation as they explore: 

• The key ingredients that helped Google’s people analytics team achieve unparalleled impact, including unique organisational structures and actionable insights 

• The 4 E Model—Effectiveness, Efficiency, Experience, and Equity—and its practical application in people analytics 

• The ‘Arc of Value Creation’ framework and how it bridges insight, action, and outcomes 

• Strategies to close the adoption gap and embed people analytics into decision-making 

• How AI and GenAI are reshaping people analytics and HR strategies 

• The skills modern people analytics teams need to thrive in today’s AI-driven world 

Whether you’re looking to optimise your people analytics function or curious about the transformative power of AI in HR, this episode, sponsored by TalentNeuron is packed with actionable insights and thought-provoking perspectives. 

TalentNeuron is the future of workforce transformation. From strategic workforce planning to skill gap analysis, TalentNeuron combines external talent intelligence and internal data into one seamless platform. 

Why not join global enterprises already leveraging actionable insights to boost organisational performance and readiness and visit talentneuron.com today. 

[0:00:00] David Green: I'm David Green, and today on the Digital HR Leaders podcast, we have a very special guest joining us, Prasad Setty, who founded and then led the People Analytics function at Google for 14 years.  Under his stewardship, Google became synonymous with people analytics, publishing groundbreaking research, such as Project Oxygen, on the attributes of good managers, and Project Aristotle, on what makes teams successful.  This work inspired many organisations to build their own people analytics functions and encouraged numerous people to enter the field.  Prasad continues to champion the field of people analytics as an advisor to companies, Chief People Officers and technology founders, and as a lecturer at Stanford Graduate School of Business.   

Personally, I'm particularly excited about this episode as together, Prasad and I will be discussing Prasad's 14 years of experience leading people analytics at Google; two of the frameworks that supported this work: the 4E model, effectiveness, efficiency, experience, and equity, and the arc of value creation to turn insights into action; the current challenges around people analytics adoption; and how AI can close that gap.  And we'll also discuss the essential skills needed for people analytics teams in today's environment.  This is a conversation you won't want to miss.  So, let's get started. 

Prasad, welcome to the show.  Really looking forward to our conversation.  For our listeners, please can you introduce yourself and the journey that led you to where you are today? 

[0:01:46] Prasad Setty: Hi there, David.  Great to have this conversation with you.  What's there to say about me?  I got into the space of people analytics more than 20 years back quite unintentionally.  The area of analytics has always been of interest.  I studied chemical engineering, went to business school for my MBA, did strategy consulting, and so the field of analytics was very interesting.  The people stuff, perhaps not, but once I got into this space, I just saw how much impact it could have on individuals and the amount of time that we spend at work.  There's certainly things that we could all try and do to make them better.  And it sounds very cliched, but I think it was the draw of analytics that brought me into this field, but it was the people that made me stay in this.  And here we are 20-plus years later, and I still continue to be enamoured by making the experience of work better, and to help organisations create really thriving environments for their people to succeed.  And so, that is something that I continue to be passionate about.   

I'm perhaps most known for founding the People Analytics team at Google.  And I was there for 15 years, and I saw Google grow more than 10x in that time.  And we were able to contribute a few things, certainly to the growth of Google and to the benefit of Googlers, but also hopefully to the world outside.  And these days, I do a portfolio of things.  I advise a bunch of companies as they grow on scale; I advise HR tech companies as they think about new products; and I also teach a course at Stanford Business School on helping students form startup ideas around the future of work.  So, that is me in a nutshell. 

[0:03:37] David Green: So, Prasad, let's start with your time at Google.  When you reflect back on your time there, what do you believe were the key ingredients that enabled you and the People Analytics team at Google to thrive and to have so much impact in the organisation?   

[0:03:52] Prasad Setty: Organisationally, Google wanted to not be bound by convention and not be bound by best practice, but to purely, and from the ground up, think about first principles for a whole bunch of areas, along the business, along product development, but also on the people dimension.  There was just a firm belief that people were the success, the secret to Google's success, and there was always a view toward, we need to be the place that can attract and retain the best talent.  And so, in a lot of organisations, I think people analytics becomes an optimisation exercise.  So, you're trying to think about how you make the most out of certain resources that you have and get to the best answer.  And those are very interesting problems for sure.  But at Google, in those days, it was much more of a maximisation exercise, in the sense that resources were abundant, our aperture was very, very wide.  And so, the thought was, if you could do anything at all to make this an amazing organisation, what would you do?  And so, that kind of unconstrained freedom that we had to pursue big, bold, radical ideas was incredibly useful too.   

Then, there are many other factors, in that over time, I started taking on more and more responsibility for other functions, including compensation, performance management, benefits, and so on.  And all of those helped me understand what were important decisions to be made for those programmes and processes that we could use analytics for.  But I'll say one of the first functions that I took on in addition to people analytics was the communications function.  And that turned out to be an incredibly important element of making our story heard.  Because I think people like hearing stories, right?  If you just bombard them with data, it's very hard to cognitively relate to it.  And people like stories.  How has it helped individuals or teams get better?  And the communications team did an amazing job of helping shape our stories.  If you just heard the stories without data, they wouldn't be as impactful.  But again, if you had the data without the stories, people would just gloss over them.   

I think when you see great journalists and you look at publications like the New York Times, etc, they use that kind of storytelling format to create advantage.  They start off with the protagonist.  They talk about a person or an individual who you can relate to their problems, and then they sort of aggregate it up to a much wider population.  And so, that's the kind of storytelling that we worked on, and that became tremendously useful both internally and externally.  So, I would say those are some of the things that were all contributing to the depth of our impact and the ambition of our approach. 

[0:06:58] David Green: Yeah, and really interesting, because actually you start obviously with a CHRO in NASA at the time, and being part of the leadership team in HR.  But also getting access to the management team means that you're more likely to be working on stuff that's actually really important to the organisation.  Then obviously, the fact that the organisation was bold and prepared to think big around analytics.  But then the third piece, that storytelling piece helps you move beyond just having great data and insights to actually drive action and deliver outcomes, doesn't it?  So, it's kind of that flow through there from identifying the right priorities, and then actually delivering outcomes as well.  And that's what it's about, is we have to think about people analytics as that cycle really. 

[0:07:48] Prasad Setty: That's exactly right.  And I wouldn't say that all of this was by design right from day one.  I think that we had to figure out what these approaches were over time, and then in hindsight, they all make more obvious sense.  But hopefully, the lessons that we and others have learned and shared, and you have certainly been at the forefront of helping bring a lot of these stories to light, hopefully we can accelerate the time to impact for many of the teams that are setting off on their journeys, or are midway in their journeys now. 

[0:08:21] David Green: Yeah, I certainly hope so.  I think, I mean I have seen organisations actually get that time to impact a little bit shorter than maybe it was 10, 15 years ago, when many companies were just getting started with this.  So, and actually that moves us quite nicely to the next topic, Prasad.  So, you recently kindly spoke about our Insight222 Global Executive Retreat just outside Amsterdam, and you shared two models that really helped guide you and the Google People Analytics team, and I'd really like to talk about each of those in turn.  The first was the 4E model, so effectiveness, efficiency, experience, and equity.  I'd love it if, for our listeners, Prasad, if you could outline the model and maybe provide an example of how you applied it in your work at Google. 

[0:09:08] Prasad Setty: Sure.  I really do think it's one of those models that I come back to very regularly.  No models are comprehensive or perfect, but this one is really useful to answer a question that I think is often on people's minds, but they cannot articulate.  And that question is, what does good look like?  If you had to evaluate a certain programme or the success of a certain process or a principle that you have, what does good look like?  And the reason that is hard to answer is because it is multifaceted, and we're usually not just dealing with trying to solve for one outcome variable, but a multitude of them.  And so, the 4E framework helps to illuminate that and helps have a conversation about which ones are most important and how you might want to trade off across multiple objectives that you have, and answer that question of what does good look like.   

So, the first E is on effectiveness.  And this is really about, is the programme or process that you're trying to put in place or evaluate achieving the outcomes that it intends to do?  And so, if you're talking about something like a recruiting process, for instance, you'd want to make sure that your selection criteria that you're using, the entire interviewing and reviewing process that you have for how candidates are assessed, are yielding new hires who are going to stay at the organisation for a long time in a highly productive capacity.  That is the outcome variable that you're trying to solve for.  And so, if you're able to measure that your recruiting approach does that, does yield those outcomes, then it's effective.   

Efficiency is really about how much time, effort, resources are you putting in to get that outcome.  So, how much time are you demanding from your interviewers?  How much time are you demanding from your candidates?  How involved is this process and so on?  The experience is, how good is this for everyone who's involved?  How good are the recruiters feeling about it?  How good are the candidates?  How good are the hiring managers feeling about it?  The overall experience, the time that it takes, for example, to make this happen.  And then equity is ensuring that no demographic is unevenly treated or left behind as you go through this process, right?   

So, they all are really important in measuring different elements of the overall programme or process, and I don't think any one of them is going to win out.  I think it's a combination of all of these.  The trap that I see often is that people often look at efficiency because it's one of the easiest ones to measure.  And I think that is a trap because I would actually say effectiveness trumps efficiency, experience trumps efficiency, equity certainly trumps efficiency, but you need to be able to afford that, right?  And you need to be able to afford that, you need to fight for it.  And it's also the case that you cannot maximize all four of these at the same time.  Something has to give.  I'll give you some examples where we were able to, because we were in a maximisation kind of framework early on at Google, we were able to not care as much about efficiency and really focus on solving for the highest levels of effectiveness and experience and equity.   

Talking about hiring, when I joined Google, there were no hiring budgets.  The only limitation on how many people we hired were how many people got through our hiring process.  And so, there was always that focus on, how do we ensure that our selection criteria keep getting better and better?  Initially, we used to use things like GPA to select for general cognitive ability.  And then we found over time that GPA wasn't a great measure, and we came up with our own internal ways of assessing for general cognitive ability.  So, we were always trying to tweak up the effectiveness measure.  And then, we would look at the experience as well.  We would look at experience to see, hey, even the most people that get interviewed or reviewed by Google wouldn't get a job offer.  But how do we know what that experience has been like for them, because they're going to tell their friends, they're going to tell others, and we wanna make sure that it continues to be a great experience.   

So, the key outcome measure on experience that we would look for is, even for people who didn't get through at Google, would they recommend that others interview and apply for jobs at Google?  And we would always see that around 80% or so, and if it ever dropped below that, then we knew that we had some work to do, and so we would figure that out.  And so, it was things like that that would allow us, first of all, create metrics around each one of these categories, and then think about what the right set of dials would be around where we want each one of them to be.  And of course, over time, efficiency became much more important for Google, just as any other large organisation is doing.  And so now, we had to do more optimisation, not just maximisation. 

[0:14:32] David Green: So, the second model that you shared was the arc of value creation, and this really resonated I think with attendees there.  So, this encompasses optimising HR processes, maximising organisational health, and then realising individual potential.  Once again, Prasad, I'd be really grateful if you could maybe outline the model and explain how it can help people analytics and HR organisations make those critical steps from insight to action to outcome. 

[0:15:51] Prasad Setty: This is one that I thought long and hard about as I thought about the portfolio of activities and projects and initiatives that a people analytics scheme takes on over its evolution over time.  In most organisations, it is likely that the people analytics function sits within whatever they call the human resources function, although in some cases that is perhaps not the case.  But for the majority of them, where they're sitting within, let's say, the HR function, their primary responsibility and accountability, perhaps because they're reporting into the CHRO in some shape or fashion, is to make that function be better.  And so, that turns out to be the first set of activities that you take on.   

Typically, HR functions are responsible for recruiting programmes or compensation and benefits programmes or performance management or learning or diversity initiatives.  And so, you're trying to look at what is within the expanse of all of these areas that belong to the HR function.  And of course, there are a lot of resources and time and investment that go into it, and can I help the function be better?  And so, that is the optimisation exercise that you typically go through, and there is quite a bit of value in that.  Some of the examples that I shared earlier, absolutely.  I mean, there's always some space in the portfolio for that.  But I think the mistake is if you only focus on that part of it, because as useful and impactful as the HR function is, it is likely only 3% to 4% of the overall organisation in terms of budget, resources, and so on.   

So, that gets us to the next piece of the arc of value creation, which is shift your attention to organisational outcomes.  Primarily, the broad category of things that are categorised that are organisational health, and that's a wide spectrum of things, but it could include things like how healthy and thriving is our organisational culture, how efficient our information flows, how easy and quick and high quality are our decision-making practices and protocols, how good are our people managers and leaders at getting the best out of the organisation, how well do we work in teams?  So, there's like a wide spectrum of things that you could all characterise under organisational health.  And I think this is where you can truly have a lot of business impact. 

Then the third element of the arc of value of creation is making the lives of individual employees much better.  And at Google, we would always have a certain set of initiatives that we were always focused on, what are we doing to help each individual Googler be the best they can be?  And so, one of the efforts that we took on more than ten years back, David, was an initiative that we called GDNA, where we had signed up more than 10,000 Google employees, and we would call them or survey them or in some shape or fashion reach out to them every six months or so.  And what we wanted to do, this is a study that was modelled after the Framingham Heart Study set up in Massachusetts, and that's been running for several decades, and that's hopefully the same thing that we'll do at Google too, that we would run this over several decades to understand over the span of one's career, what contributes to career happiness and life happiness. 

Those are the kinds of things that we would always find some room to take on.  And I think if you are a people analytics function, and if you have initiatives that span across this portfolio of solving for the HR function, helping with organisational health, and ensuring that each individual employee in your organisation gets some value from the work that you do, that is what makes you a great people analytics team.  Because for me, naming the team as People Analytics, when I got to Google, that was the first decision that I did.  And for me, it was really about saying, this is not an HR analytics team because we are not just trying to be analytics for HR, but we are trying to be analytics for the people of Google.  And I think it becomes a very, very rewarding sort of approach. 

[0:20:23] David Green: And that naming convention seems to have been adopted by most, and there are lots of different names for people analytics, but people analytics seems to be the most commonly used now.  So, yeah, you started that process.  And I think that last project that you talked about there, the GDNA initiative, is just the example of the big, bold approach that you did at Google.  So, obviously, some of what you're doing is trying to sell for the here and now, but you're also thinking about something in the long term as well, which I mean obviously, we won't be around, but it'd be great to see the results of that in 50, 60 years' time.  And actually, I think what you've talked about in those three arcs, I mean, you can immediately see how a people analytics team that is focused on that arc of value creation, particularly with what you spoke about earlier and getting the context and access to the senior leaders as well, you're going to have impact.  And if you have impact, then (a) as a people analytics professional, you're enjoying what you're doing, because it's likely to be impactful work and you're helping the organisation and you're helping individuals within the organisation.  And (b) you're going to get more investment as you grow and deliver impact as well.   

[0:21:44] Prasad Setty: It is certainly a self-perpetuating cycle for sure, yeah.   

[0:21:50] David Green: We recently published our annual People Analytics Trends Report, and one of the key findings highlighted a gap between the democratisation of people data and insights to HR professionals and people managers and leaders predominantly using technology, and their adoption.  So, for example, although democratisations increased year on year, this year 71% of the participating 348 companies told us that they share people data and insights with leaders and managers across the enterprise; but only 47% had a high level of adoption in HR and an even lower 28% of adoption outside HR.  So, kind of two-part question, Prasad, and discussion perhaps as well, what advice can you provide to HR leaders and people analytics teams on how to close that adoption gap; and how do you envisage maybe AI supporting this in the near future? 

[0:22:48] Prasad Setty: First of all, I'm glad that you are looking at what the current state is in terms of democratisation of information and how it's being used.  This is something that we've been talking about for a few years, how do we get a whole bunch of valuable people, data, and insights in the hands of more people within the organisation, not just limited to leadership?  And so, that is great.  I'm glad that teams are trying to do that.  I'm not surprised that now, the next challenge is going to be about how do you actually have adoption, not just sharing of information, but how do you actually have adoption of this information.  And so, that's like the next set of challenges that we set out to do.   

I thought about it in terms of a few different sort of checkpoints to understand and assess how good your democratisation initiatives are.  One of my friends talks about dashboards as a great way to hide insights rather than reveal them, because they just throw so much information at you, and it's unclear sometimes how you are supposed to make sense out of it.  And so, maybe in a way of helping make this information be useful, I think about it in terms of five things.  I'm full of frameworks, as you can see, but maybe I'll just focus on a few of them for brevity, but let me just give you the broad set of things that I would test against.   

So, I would look for simplicity.  I think making sure that you have the simplest viable set of information that is easily explainable and clear.  So, simplicity, clarity, and explainability, I think all fall into one bucket for me.  And particularly for business leaders or managers who might not look at this information day to day, it is really important for them to understand the explainability and the clarity of information.  So, rather than if you have something that has like 10 or 20 metrics, you might want to think about, "Hey, here are three broad themes for you to know, and let us tell you what the linkages are between A and B, so that if you are trying to come to this like a dashboard, or like something that I'm pushing out to you every three months, I'm not giving you too much of a cognitive overload around why this is important and why you should be paying attention".  So, that is one aspect, particularly for managers and leaders who might not need this information day to day.   

Then, there are perhaps others who might use this information day to day.  And I again go back to people like recruiters or HR operations folks who would benefit from seeing just-in-time information, so that on day-to-day decisions, how should I think about this particular set of candidates, or how should I better search for this particular job opening that I need to fill?  I think ensuring that information is embedded into the tools that they use day to day, rather than being somewhere else, I think will truly help with adoption.  So, that is one bucket.   

Then, the second bucket is really around drilling down.  I think all these organisations are incredibly complex.  And so, you want to make sure that it's very easy for people to drill down to understand what exactly is causing this high-level outcome to be off of threshold in some way or fashion.  And sometimes, those drilldowns might be drilling down based on different demographics or location or job levels, or any one of these type of criteria.  But other times, the drilldowns might be much more in terms of explanatory power, right?  So, the reason A is trending down is because B or C six months back were in this kind of a place.  And so, that again requires you to have a very clear model of what is your theory of change, or what is your theory of explanation around this, right?  And so, in the absence of that, you're just letting the user interpret this data as they see fit.   

Then, the third aspect that I think about and perhaps the most important is actionability.  And this is where I think the people analytics teams need to provide space and information so that they can get feedback from whoever they are reviewing this data with or providing this data to, for them to say, "Hey, what did you interpret about this data?  What did you take away from it?  Did it change your mindset about something?  Are you committing to doing something different?" so that they can keep a track of those sets of actions or beliefs; and then six months later or three months later, whenever they go back to them, they can now say, "Did you follow through on any one of those areas?  What was the impact?  And can we now have a much better feedback loop around information sharing, actions taken, and whether that has resulted in any change to the outcomes that we wanted?"  So, I think trying to think through some of these dimensions will hopefully address the adoption gap over time. 

[0:28:29] David Green: That's fantastic.  In the end of the day, it's actionability, isn't it?  If people will use something, if it's useful and it helps them make better decisions and they see the impact of those decisions.  I wonder, because when we spoke a couple of weeks ago, Prasad, how do you see AI helping with adoption potentially?  I know you're working with a number of AI startups at the moment, and obviously given your time at Google, you're well up on what is possible now and what will maybe be possible in the not-too-distant future, around maybe how people interact with dashboards.   

[0:29:04] Prasad Setty: Yeah, I'll give you an example of one of the startups that I'm close to, it's called Take2, and what they're trying to do is conduct job simulations for early sales career roles, and be able to provide high quality simulations where the candidate is trying to sell a certain product or a service, and that is what the hiring organisation is trying to assess them on.  And so, in this case, they're able to use AI to take a first pass at not only creating the simulation, but also analysing on topics like the ability for the candidate to communicate clearly, how many filler words are they using, how clear is the grammar, and so on.  And in addition to that, they're also able to think about more advanced topics like active listening and whether the salesperson is expressing empathy or they're able to handle rejection in a thoughtful way.  And so, all of these are things that they're able to do using a wide variety of machine learning libraries.  For example, there's a flesh readability test that can assess written words; there's a whole bunch of language libraries on assessing for clarity of spoken communication; and so on.  And so, they're able to do quite a bit of that.   

So, I think one step is for people analytics teams to start testing with and experimenting with these tools to make it much easier to interpret information that is coming in.  The second is to try and, especially because a lot of our interactions are on using text and communications and the way we collaborate, I think there's a lot more that you could do to collect information better.  And so, a standard way that we typically had for collecting information is through surveys, and those still remain the best way to think about attitudes and perceptions and intentions.  And for the longest time, one of the questions that I would ask any PhD that we would hire into our people analytics team was, "Hey, how would you do your job if you couldn't conduct surveys?"  And I think with AI tools now, perhaps you can get to a better place around that.  Because I do think while surveys, etc, are very useful, they tend to flatten people, they tend to compress them down into monotonic, unidimensional folks, whereas we're all much richer, much deeper in terms of our perspectives, and we're certainly not just single-dimensional.   

So, I think there are quite a few new ways in which you can gather information.  So, let's say you're trying to collect information for a 360 process, and so on, or even for performance reviews and feedback, you can now use AI to gather information, just like you and I are having this conversation, David, we could make it about that.  So, if you're trying to gather feedback on a particular people manager, and you're trying to focus on how well do they set strategy and how well do they set goals for their teams, then you can have perhaps a simulated conversation with their team members and gather information that way, rather than through surveys.  And so, I think, again, there are a lot more opportunities for collecting data in very different ways.  So, these are all some things that companies are already doing, and I think that kind of interaction might make it much better, even when you're sharing results too.  And certainly, when you're sharing results, when you're talking about adoption of information that you're putting out, you could certainly use AI already to summarise information and provide it in a way that the person that is receiving this information views it best.  Some people are very visual, some people are much more open to written communication, and some people are much more open to sort of quantitative information.  And so, trying to understand who your receiver is and what the best way to put this information in front of them might all be great things to test and experiment.  Certainly, I would want to do that if we were running a team today.   

[0:33:55] David Green: What would your advice be to a people analytics leader or people analytics team that is getting that request to kind of help put the AI strategy for HR together?  I guess it's probably back to what you were talking about earlier, about making sure you're connecting it to something that matters for the organisation, rather than it just being about HR, I guess, but I'd love to hear your thoughts on that Prasad.  

[0:35:09] Prasad Setty: Yeah, and I think part of it for me is, again, understanding we all must have some beliefs about the theory of change out here.  Otherwise, we're just going to be constantly reacting to inbound data, whether it is from the large language model providers or a whole bunch of other startups who just want to sell us their best and shiniest thing.  But if we started off with first principles, I'm enamoured by generative AI, and I'm constantly trying to think about what is the way in which this could impact knowledge work, particularly because that's the area that I've spent a lot of time in.  And I've come back to two broad ways in which it could do that.   

One is to solve for time poverty, and the other is to solve for thought poverty.  And time poverty, because we're all stretched by having to do too many things, and therefore we're trying to be better at them; and then thought poverty, because we are perhaps too stretched, we end up doing a lot more shallow work where we are just trading text messages or emails, but not really getting the time to go deep into any one of these areas.  And so, you might have your own beliefs about how this is going to approach, but if you, again, think about the time-poverty and thought-poverty angles, then you might be able to say, "Hey, within HR, where are the areas where I could absolutely address time poverty by using generative AI best?"   

So, one of the other startups that I've worked with, they're called CadenceAI, and they focus primarily on helping HR organisations address benefits questions, which are really hard to solve and certainly very difficult if you have hallucinations and so on, right?  It's not a great experience.  So they said, "Let us tackle that big problem", and they're seeing a ton of positive reception and adoption from many companies, and they're ending up with a very personalised user experience, and this kind of offering is available 24/7.  And so, this is one that is freeing up their HR operations and benefits teams to focus on much higher order queries where they're thinking about, "Where do our programmes seem like they need help, rather than answering each and every ticket that comes in?"  They're able to see a much more summarised aggregate view of what is on their employees' minds.  So, that is one kind of way in which you can solve for time poverty. 

Then, I think you might want to think through, how do I address thought poverty?  How do I make sure that people are much better prepared?  How do I ensure that they are learning?  How does the range expand?  Because with the new types of technologies that we now have, you or I, David, could write code, because what we need is people who understand great problems that need to be solved and to have a view of what good looks like, and with our experience, we could do that.  And then we could, say, develop a prototype based on my vision and my thinking, and so the coding is not as important now as the thinking is.  And so, I think it opens up everyone's ability significantly.  And so, I would want to be doing experiments like that.  I would want to be thinking about, can I expand people's range; can I expand their ability to exercise better judgment; can I expand their ability to think deeply and come out with two or three really well-reasoned-out and well-thought-out possibilities in a particularly difficult problem area?   

I absolutely do think that people analytics leaders are best positioned to help their HR functions and their CHROs with this, because they already bring the technical chops, they probably understand the limitations.  They also hopefully have the context of what the right problems are.  And then, they can fashion these quick experiments so that you are accelerating the learning cycle of your organisation.  So, I think people analytics leaders and teams are amazingly well positioned to help with this. 

[0:39:31] David Green: I think that turns us nicely to the skills needed in people analytics teams.  So, I was fortunate to attend the Wharton People Analytics Conference in 2023, where you and Dawn Klinghoffer were with Matthew Bidwell on stage, looking at the past and future of people analytics.  And I think you said then that the four most important skills in people analytics were behavioural science, data science, consulting, and product management.  Can you outline maybe why each of those four skills is important, and highlight any additional skills that you think now, 18 or maybe 2 years later, you'd think of adding, given we're definitely in the age of AI now? 

[0:40:14] Prasad Setty: This is certainly one more thing that I would say was a reason for our success and for the wide range of things that we were able to take on.  I've certainly learned through experience and absolutely believe that bringing people from diverse backgrounds and perspectives to solve problems is incredibly powerful in terms of the innovation they unleash, as well as thinking about how we resolve various types of conflicts and move forward, right?  And so, when we set up the people analytics team, we would think about this three-thirds model.  One-third of us should be from academic backgrounds, IO psychology or management or sociology, and they bring with them research skills.  They're in tune with the latest academic developments of what is working and what is not in the broad field of organisational behaviour.  And so, that is a very useful set of skills to have. 

A second skill is about data fluency.  And it's certainly about, I think over time, it became about data and technology too, but it's certainly about how you ensure that you always have your hands on data that is of very, very high quality, how do you make sure that you're able to make it accessible to a broad set of folks, and so on.  And then, the third set was about consulting skills, because ultimately we want to not just share information, but we actually want to help the organisation move forward; and to truly be able to make decisions, you need to be able to put yourself in the shoes of business leaders who have much broader and much more expansive context, perhaps.  And so, that skill of translation between insight to something where I can showcase multiple options.  So, for example, I would spend a lot of time with the board, I would spend a lot of time with our CEO.  And broadly, my intent was always to say, "Here's where I, as an expert, am going to give you three options that solve for each of these different areas that is the problem that we are trying to solve, and here are three options.  And I'm going to recommend option B as the best approach, but I wanted you to see this broad view, and look forward to your decision-making".  And so, that was a model that worked really well when I would work with Sundar and with the board. 

I would say this has always been important, but I think increasingly important now that we have more and more AI tools.  And so, the additional element that I would add to those four these days, David, would be certainly a focus on responsible AI, responsible data management, ethics, right, something around those lines that has always been important, but I think it needs to become much more prominent now.  I think as you think about the deployment of AI, certainly in the people space, one of the key dimensions of responsible AI is going to be around fairness and bias.  And so, to evaluate the data that has gone into these models, to ensure that you're using the appropriate bias correction methodologies, and to make sure that you build trust with your products and with the data, I think is really key, and I don't think it's something that we can take for granted.  And so, that is an area that I would certainly want teams to invest even more so than they did in the past. 

[0:43:54] David Green: Right, Prasad, we have two questions.  We're going to look at the future, and then we're going to do question of the series.  So, let's see how ambitious we feel about this one.  So, if we fast forward to 2030 or maybe even to 2035, go forward ten years, how do you envisage AI and social sciences playing a deeper role in the work of people analytics across maybe those three elements of the arc of value creation as well? 

[0:44:19] Prasad Setty: I think broadly, the world of generative AI should have profound impacts on work, and particularly knowledge work over time.  And I would look for how much work is actually required to be done to come to a certain outcome today.  And that comes all the way from goal setting on a particular area, assembling the right types of teams, the way those teams work together, then you think about how they get decisions made, how they actually deliver an end product.  I don't think, even with all the tools that we have today, David, I don't think we necessarily understand how much work goes into the system to get a certain outcome done.  And so, I would look for AI to truly help with that, to provide a systems level view of how that work is done.  And then to think about things like cycle times.   

I was having this conversation last year about, what are the ways in which organisations could truly benefit from AI?  And I think I come back to things like cycle time, which are, I think, a really important measure of work.  I think cycle times get stretched today because of the amount of shallow work we are doing, because we just don't have enough context, and because leaders perhaps get very, very bottlenecked in terms of decision-making.  And so, trying to come up with better ways of understanding the amount of work that it takes to get something done, to reduce cycle time, those are the kinds of areas that I would look for people analytics to have a much better view toward.   

I also think that we will certainly have much better ways of thinking through this multidimensional functionality.  I think, over the last ten years, we've sort of reduced everything down to five-star ratings, and that's certainly been a part of what our consumer mindset has been, whether it's on Amazon or Uber, but I think that has sort of cascaded down into the people function as well.  And I think that kind of reduction into single dimensions leads to a whole bunch of gaming behaviour, it doesn't lead to a very high-quality signal.  So, I'm really looking for how AI can result in much higher-quality signals by taking text, taking multi-modal information and helping people be much better.  And one of the ways I think you do that is through simulations, much better simulations.  And if you look at like a sports analogy, you always have sports people talking about how many repetitions they do.  And I think that equivalent for knowledge work, I think, is now possible.  And a lot of adult learning talks about how you learn very little in programmes and much more on the job, and on-the-job experiences.  And the reason is that over the 12 or 18 months of being in a new job, not only are you learning the main skills, but you're also learning to face multiple contexts in different situations and build up your muscle. 

[0:47:54] David Green: That leads us to the final question.  So, this is the question of the series we're asking all our guests in this series of the podcast, and it's another area where people analytics is increasingly having impact.  So, Prasad, how do you leverage people analytics to inform strategic workforce planning initiatives? 

[0:48:13] Prasad Setty: When I started off my journey in this space, I actually led workforce planning at Capital One, and that's how I got into this space.  And so, I think it's always been related in some cases.  There's a lot of it that people do that is all about workforce planning.  And I think it's sort of related to some of the things that we've been talking about in the last couple of questions.  One is thinking about the work itself.  Before I do workforce planning, particularly with generative AI in the mix, I would want to think about work planning, right?  To achieve certain outcomes, what is the work that I think needs to get done in the organisation?  And increasingly, which of that type of knowledge work do I want to be done by people, to be supervised by people, but done by machines, etc, right?  And so, the first element of workforce planning is going to be this kind of a work planning exercise.  And you see, for example, people who are on the outlier of this, like Klarna, where the CEO says, "I haven't hired a single new human over the last year.  I've been trying to take out a whole bunch of work and make it…" you know, their customer service is run by ChatGPT or whatever, right?  Now, there's probably always a little bit of embellishment around these kinds of claims, but I think it's always useful to look at some of these outliers and say, how far, how extreme can I take this?  And so, I would start off with that.   

Then I would think about skills, which I think is another topic that we have touched upon.  I would think about having a clear assessment of what skills have we been good at in the past?  What skills are we good at now?  And if we trace that arc, where would we be in terms of our skill fluency as an organisation over the next few years?  And what is the gap between that and a clear overt conversation that we have about what we want it to be?  And this is, again, going to be a hard exercise, because people are not very good at understanding what skills we need to be successful for the future.  And so, we'll probably have to draw a line in the sand and we'd have to have the executive team and the people leadership team say, "This is our intent, right?  This is where we want to be".  And so, I think those kinds of exercises are incredibly useful.   

Then you can sort of think about what you have in terms of your supply in your current organisation.  What fraction of the people can you put into this kind of a learning mode to get to the skills that you want it to be?  You want it to be very clear to them, "Hey, these are the kinds of skills that we want in our organisation to succeed in the future, and are you up for the journey, because we are ready to invest in the appropriate infrastructure, technology, etc to do that". 

[0:51:10] David Green: Well, Prasad, it's been a wonderful conversation from my perspective certainly, I've learned a lot over the last hour or so, and it's always a joy to speak with you. I know that our listeners will appreciate your wise words too.  But before we end the episode, Prasad, please can you let listeners know how they can get in touch with you, maybe follow some of the great work you're doing.   

[0:51:33] Prasad Setty: Absolutely, David, always eager.  People analytics continues to be a love of mine, so absolutely always happy to connect with folks in this space, and I certainly look forward to all the episodes and the writings that you publish for sure.  But for folks who reach out to me, LinkedIn is a great way to connect with me.  And I'm always curious to see where this field can go in future.  And so, I really look forward to hearing from folks, and I'm at a stage now where my focus is on amplifying other people's success.  So, if I can help anyone at all in any stage of their journey as they think about their own experiences and adventures in this amazingly rewarding space, drop me a line and if timing works, I'm really happy to connect with them. 

[0:52:30] David Green: Thank you so much for being a guest on the show, Prasad, and I look forward to seeing you again, hopefully in person, in the not-too-distant future.