Episode 159: How To Scale People Analytics Through Experimentation and Productisation (Interview with Ian O’Keefe)
In this episode of the Digital HR Leaders Podcast, David is joined by Ian O’Keefe, Head of Talent Analytics, Amazon.
With an impressive background working at renowned companies like Google and JP Morgan, Ian brings his wealth of experience to the table as he shares his insights on the evolution of people analytics teams and the critical role of a product mindset.
Key topics covered in this episode include:
Reflections on the evolution of the people analytics discipline over the years
Different types of operating models observed in various organisations
The role and importance of experimentation in people analytics and HR, and the significance of a product mindset in scaling people analytics
Skills and capabilities required for successful productisation of people analytics
Practical tips and learnings for HR leaders looking to embrace experimentation and productisation
Predictions on the future of the people analytics field and the potential role of generative AI
Support from this podcast comes from Worklytics. You can learn more by visiting:
www.worklytics.co/DigitalHRLeaders
David Green: If there is one question that the team at Insight222 get asked more than any other by CHROs and other senior HR executives about people analytics, it is, what are the best companies doing? So, in the study we conducted with 184 global companies in our recent people analytics trends research, we sought to answer this question. We found that leading companies, those companies that impact business value to a higher degree than non-leading companies, display seven key characteristics.
One of those seven characteristics is that the people analytics function in leading companies has a strong focus on scaling people analytics to employees through personalisation and productisation. So, in today's episode, I am delighted to be joined by Ian O'Keefe, Head of Talent Analytics and Data Science at Amazon, where Ian and I will be discussing how to productise and scale people analytics in order to drive organisational success. With a wealth of experience working in people analytics leadership roles at companies such as Google, JPMorgan, and now Amazon, Ian has witnessed first-hand how as people analytics teams mature and the need for insights expands, a product mindset becomes essential.
So today, we have the privilege of tapping into Ian's wealth of knowledge, where we will be exploring the key principles and best practices of productising people analytics, the role of experimentation, common use cases, and how HR leaders and people analytics leaders can successfully implement it in their organisations. So without further ado, let's get into the conversation with Ian.
Today, I'm delighted to welcome Ian O'Keefe, the Global Head of Talent, Science and Analytics at Amazon, to the Digital HR Leaders podcast. Ian, welcome back to the show. Obviously, we recorded an episode back in 2019, I think it was, when you were at JPMorgan. Can you share with our listeners a little bit about yourself and your role at Amazon?
Ian O'Keefe: Sure, David, thanks. Good to see you again. It's been a while since we had that pre-pandemic interview in 2019. I guess in short, I've been in analytic roles for 20-plus years, leading people analytics teams for the past 10 to 12. I started my career in "human capital", in consulting at Deloitte and some other boutique firms, then went on to a couple of different corporate roles and spent time in different sectors and financial services at American Express and JP Morgan, and retail at Sears Holdings, and more recently in tech with Google, and now more recently with Amazon where I'm leading analytics.
David Green: As you said, Ian, you've been in the field for quite some time and worked at some very prestigious organisations as well. I'd love to hear your reflections on how you see that the people analytics discipline has evolved over your years working in it.
Ian O'Keefe: I've always worked at the intersection of data, process, transformation and technology and I think people analytics has kind of evolved in concert with those three spheres, if you will. But over the years, I guess I'd point to three big themes that I've seen. One is related to the talent of those in people analytics and where they're coming from, right? I think historically you've seen people analytics practitioners come from HR centres of expertise like operations and tech and the like. And fast-forward the tape to now, I think you're still seeing some of that, but more and more you're seeing teammates come from data teams and a lot more, and the mature functions of the organisation like marketing and pricing and supply chain optimisation, and things like this.
So, I think the talent funnel has definitely broadened and you also see that in education, by the way, right, when it comes to undergrad and graduate-level university degree programmes, more and more in analytic and data science. And you see MBA programmes even having infusions of statistics and kind of the more analytical disciplines that we would have hung our hat on for the past ten years. So, I think that's all positive, and I'm really happy to see that.
The second thing that I would point to is what I would just call kind of generally breadth, so the types of organisations that are in people analytics. If you were to go back ten years ago, I think you would see generally very large, high-margin organisations with technology at their centre were at the vanguard of people analytics, and still are in many ways. Now, you have organisations from pretty much every sector, and you also see organisations at mid-size and smaller-size orgs starting to invest in people analytics, single person teams, small tiger teams that are taking on some of the challenges of foundational data that we've seen and talked a lot about over the years. So, I think the breadth of organisations that are getting into it has definitely widened.
Then, the last thing I'd point to is the depth of complexity of problems that people analytics teams are solving and the solutions that are being generated. So, if you think about the way that we have always had mathematical and statistical depth right at the centre of people analytics teams, I think you're starting to see more and more those skill sets and those focus areas being augmented with product design, with technology integration, and having that be not a nice-to-have, but more of a force multiplier for people analytics teams as they think to scale and deploy their solutions throughout the organisation. So, I think the adjacencies, when it comes to operations and tech that people analytics has worked alongside for many years, is starting to blend and even in some ways become part of people analytics teams organisationally or strategically. So, that's pretty exciting to see.
David Green: Yeah, I agree with all of that. And actually, also we're going to be spending quite a lot of the time today talking about productisation. And I guess, you know, part of that is that the insights and people analytics, or the benefits of people analytics, may be reaching more people in the organisation than maybe it reached ten years ago. Maybe it was a bit more of a white-glove type of service for senior people within the organisation to give them insights, hopefully support better decisions. But you're right, I mean, the way it's transformed in the last ten years is incredible.
I know when Jonathan and I were writing the book, we could see that the growth had been significant. And if anything, the pandemic helped accelerate it even further because suddenly...
Ian O'Keefe: This book you mean?
David Green: That book, yeah!
Ian O'Keefe: Yeah. It's on my desk for quick reference. I think you say it well in the book that people analytics is not about analytics, it's about the business. And I would tweak that to say that it's about the business and it's about the employee experience, the people that are driving the business every single day, going direct to them with helpful information and recommendations; and all the things that can be triggered and derived out of people analytics at scale vis-à-vis products is an exciting development, I think, in recent years that we've seen.
David Green: Before we get to that, I'm going to take advantage of the fact that you've worked in some of the leading companies in people analytics, or certainly with big teams, Google and JPMorgan you mentioned, but obviously at Amazon as well, and you've seen various different people analytics team structures and operating models. For listeners, what are the different types of operating models that you've seen; and how is your team organized at Amazon?
Ian O'Keefe: I think there's this age-old question of starting a team, analytics or otherwise, when you have component parts already present in the org, to what extent do you centralise or leave certain things decentralised? So I think we've seen a number of operating models, depending on the org, and I don't think there's any right model, to be clear. People analytics teams reporting to the CHRO has clear advantages and also disadvantages, depending on where the real opportunity areas and the problems are, from a data and a systems and a tech perspective.
So, I think leadership is really, really important. That's a forehead-slappingly obvious statement, but the HR leaders that can help incubate and curate the capabilities for people analytics, they need to be analytical and analytically literate, but we see that in some orgs in compensation, we see that in talent, we see that in operations, and we see that at the C-level as well, at CHRO level and such. So, I think where the team reports into is very important, but it's important to align that structure to where the opportunities and the problems are most pressing, where you can get the most velocity and momentum.
Amazon works in a highly kind of portfolio-based way. So, we have a team that looks after analytics on the talent side. There are other teams that look after analytics and science and other parts of the employee lifecycle; we work in collaboration. Other organisations, like JP, the org that I built and led, we centralised more or less the entire HR data model into the workforce analytics team there.
I think an overarching philosophy that I've come to really embrace at Amazon is, two is greater than zero, but one is greater than two is greater than zero, meaning if there isn't a solution to your problem, two versions of a solution arguably is better than no version of a solution. So, if you have two projects or two teams or two whatever working to solve a problem on behalf of the customer, that's pretty hard to argue with being bad. Ideally you need to get to one, I think, to manage ongoing and not have different versions of the truth and different versions of a solution out there. But I think when it comes to the federated versus centralised argument, it's helpful to bear that two is greater than zero, except when one is greater than two is greater than zero, in mind, and not to get too hung up on that.
David Green: And I guess, in many respects, whatever structure you have within your people analytics team, it needs to reflect how the company is organised. I know when I was at IBM, IBM's quite a centralised organisation, so mostly quite a centralised team, sounds like it was similar at JPMorgan. But Amazon, I mean, Amazon is an enormous company, isn't it, and it's grown rapidly in the last few years. So, trying to centralise everything in Amazon probably wouldn't work.
Ian O'Keefe: I think that's a fair statement. I think you need to meet whatever org you're in, leading a people analytics team, you need to meet the org where the org is. And if you have a preconceived notion of how to structure your own team to get the maximum result, and that's out of sync with how the org operates, the way that the operating principles have been in and out of HR and the business, you're going to be probably setting yourself up not so well. So, I think you need to meet the org where the org is.
David Green: And thinking about, again, not necessarily talking just about your current experience, but what are some of the advantages of centralised and decentralised models from your experience? Obviously, you've worked in several organisations, and I guess there's good things about one and there's good things about the other, but yeah, it would be good to hear some of those actually.
Ian O'Keefe: Yeah, I think decentralised formats give you the benefit of having embedded experts close to the different processes and programmes and businesses that ultimately you're trying to serve and improve. I think there's an increased coordination tax that goes along with that. It creates more kind of matrix-based complexity, where that's less so in a centralised format. A centralised format gives you the benefit of having perhaps a singular technology and analytic thesis across the entire HR data model, which if the org is ready for that, that's certainly a viable approach.
If an organisation has data production from X different products and Y different systems across Z different geographies, maybe a centralised format isn't where the org is at the moment, and need to think about how to understand the problems locally to then solve them maybe centrally and globally before you centralise. So, I think it's not one or the other, all or nothing; I think you have to think of it on an arc of evolution and treat it as a journey.
David Green: As I said, I'd like to spend most of the time, or the rest of the time, talking about two things: experimentation and productisation in people analytics. I know these are topics that you're passionate about as well, Ian. So, for the listeners out there, what does experimentation in people analytics mean at Amazon, and why do you think this is such an important element in people analytics, and frankly HR in general?
Ian O'Keefe: Yeah, let's talk about experimentation in HR, generally speaking. Let's take a minute first to reframe the notion of experimentation in HR. I think over the years, we've heard this word kicked around and it's created maybe at times somewhat of a visceral reaction, like experimentation in HR and in the workforce, that sounds creepy, that sounds like an overstep of the social contract that companies have with their employees, a breach of trust. I think a guiding principle when you think about testing something, experimenting testing, is that you do no harm. You can't introduce a policy, a process, a programme, product, anything that would create an inverse impact to one group and a more positive outcome than another.
Again, this is hopefully at this point in the people analytics space, a very obvious thing to say, but experimentation, I think when you kind of look at it from a do-no-harm standpoint, well, what does it mean? I think it really means that you're giving people more choices, you're creating variations on information that's helpful for people, and you're working backwards from the managers and employees that you're trying to serve to create a more personalised, helpful experience for them to bring their best to work every day.
So, if you think about working backwards from managers and employees, and then testing what might be useful for a particular persona or cohort of managers to receive when it comes to making hiring choices or putting people up for promotion, or making compensation administration decisions, or thinking about the equity of talent outcomes within an organisation, these are all really, really important, in some way one-way door decisions that you cannot get wrong. And I think a lot of times people analytics teams can get hung up on what's the single best way to solve this problem and to do this, and we can debate decimals all day long.
When you think about, for instance, the most compelling engaging metrics on the workforce and team to put forward to managers on a monthly basis, easy example, right? You probably can agree on 90% of it really quickly and then you could spend months debating 10%. Why not create variations on that 10% that do no harm, that don't kind of stray strategically from some of the needles you're trying to move and measure, but measure the actual engagement of one, two, three, slightly different variations on a data product that ride the rails of a singular technology infrastructure and stack.
You can do the same with content and words that might be in a recommendation that appear to someone in a transactional workflow, that might suggest the data shows that X, Y, and Z are true of employees who are similar to this person you're looking at. What's the psychology that you're trying to instil in that recommendation; are you trying to instil action, thoughtfulness, immediacy, urgency? Let's test a couple of different versions and get the feedback on what our customers think is the right way or the wrong way to do this while doing no harm. And you can draw up a number of examples when it comes to data science approaches and assumptions.
We have, I think, found that technology, availability and cost and complexity has come way down in years. Conducting experiments sounds like a really time-intensive, daunting thing, and it is time invested, but I think if you test three versions of something in a way that does no harm and creates more choice and personalisation, you get feedback and you can test, learn, fail fast, and create some implementation momentum in the process. So I think, in my mind, that's what experimentation is all about. It's about choice and it's about implementation and momentum and, by the way, meeting people where they are by hearing from them.
David Green: And I'm guessing an important step towards productisation as well, because you don't want to embed something and build something potentially for either the whole organisation, or certainly a significant part of the whole organisation, without understanding are people going to use it, number one; number two, does it help you achieve the outcomes, and the employees the outcomes, that they're trying to achieve? So, it's an important step in the whole productisation process as well, isn't it?
Ian O'Keefe: Critical, absolutely critical. When you're pushing a model into production and then programming product treatments on top of whatever that model might be signalling or triggering, you should experiment with the words on the screen and thresholds at which certain recommendations might be pushed to people. I think you do that in a way that is transparent, that's traceable, auditable that creates again more choice. I think it's a very good step to take from a change perspective and it's a necessary one to take from a productisation perspective.
David Green: Which helps us move nicely to productisation, Ian! So, I was recently at the Wharton People Analytics Conference and Prasad Setty and Dawn Klinghoffer were in the opening session, and they were reflecting back on the previous ten years of people analytics and looking forward to the next ten, because it's the 10th Wharton People Analytics Conference. I know Prasad is one of your erstwhile colleagues at Google and he highlighted four key skills for people analytics: funnily enough, consulting, data science, and behavioural science were the first three, and then a product mindset was the fourth.
His view, which is backed up by our research at Insight222, is that a product mindset enables people analytics to scale, and I'm pretty confident that you're also a firm believer in productising people analytics. Can you explain what productisation of people analytics means to you and means to Amazon, and why you believe this is such a critical component of scaling people analytics?
Ian O'Keefe: The one word definition of what it means and why it's important is scale. For me, introducing science-derived insights into transactional workflows and software applications that people use every day to manage their teams and make important decisions across the org, is the name of the game to scale your influence and an insight dissemination out to the organisation. To be clear, I think there will always be a need for the one-stop shop that HR colleagues, managers, etc, can go to, to research, seek, understand where their orgs are on a backward-looking, forward-looking basis.
I believe that to scale the impact that we can have in surfacing those insights in moments that matter, based on different points of the employee life cycle and career journeys that people are on, helpful information bits that are embedded in the workflows related to hiring, transferring, to retention decision when it comes to keeping star performers from leaving who might be at risk, you can understand some of the individual team or events and data streams that might create a scenario whereby it's helpful for David to know now that someone within his organisation looks more ready for promotion than he might realise, because he just took over the team.
As an example, that's hard to put into a static report and hope that someone sees it on the off chance that they go to your site or your product that you built. But if you're embedding that into the tools or the products or the apparatus that you might have to manage promotions in that example, that's useful, I think, in my view. So that's where the working backwards from the customer's experience in the product, taking a product mindset, would be where I would definitely agree with Prasad.
The technology integration that comes with literally integrating people analytic productised outputs into software interfaces, that's a build from a software developer perspective with people analytics teams on a product-specific use case. And so, I think the union of the analytics product and tech for a better employee experience is one that is necessary for impact and for scale.
David Green: It's interesting, actually, because listening to you there, Ian, you said something earlier that people analytics isn't just about the business, it's about creating a better employee experience, and it's almost like productisation is the way that you can do that at scale, as you said, by personalising, by putting it in, helping managers make decisions, help managers, I don't know, whether it's promotion, whether it's hiring, whether it's even doing a catch-up with one of their team, giving them some prompts and some information that support that discussion; it really is taking it to the next level, isn't it?
Ian O'Keefe: I think it is. I don't think it's the answer, but I think it's a really important part of the formula to do those things you mentioned, David, and to, in many ways, think about how you might teach culture at scale, "Here's how we think about managing performance or paying people or hiring them, and here's how we teach managers through our products in those moments that matter", where you rely on high judgment and perhaps experience of having been in this org for five, ten years. Once you've had a couple laps around the track, you get it culturally, but especially now and during the pandemic especially, and what I would dare call post-pandemic, we can debate that one, I think it's really important to think about not just the product side, but the humans that are using your products.
The managers, in large part, are culture carriers for you and if you're relying on the manager's interpretation of data and insight to inform his or her judgment to make a better decision, management capability, leadership capability, the culture we want to carry forward in that conversation is a really important piece of it, too. So, I think productisation turns the conversation of management effectiveness and culture into one that you can measure a little bit more so than you used to, and you can help improve perhaps a little bit more tangibly and tactically in ways that previously were probably more theoretical and maybe out of reach for some orgs.
David Green: Can you share some examples of people analytics products that you and the team have delivered at Amazon, or maybe some previous companies where you've worked; obviously, only the ones that you're able to share with listeners?
Ian O'Keefe: Yeah, I'll do my best here. I think one that I really enjoy talking about and is super-relevant, I think, for any org is on the recruiting or the talent acquisition, talent curation side. When you think about orgs of any shape or size or maturity, for that matter, there will probably be key roles that are important for that org to hire often and to get right, and in the tech space, you could point to software developers or software engineers as that; sales, frontline healthcare workers if you're in healthcare; tellers at banks if you're in a financial services organisation. And if you're hiring for a critical role and you're hiring with volume and velocity, that's a real challenge to recruiters, right? And so a lot of independent software vendors have created some data-science-oriented solutions to this.
At JPMorgan, we built an ML product that plugged into the front end of ATS with Taleo as JPMorgan's an Oracle shop. So if you have 100 applications to a role and there's must-have critical skills, experience certifications that you can write down and label and look for, and you would rely on recruiters to do that and sift through a pile of 100 or 500, can you look for said must-haves with natural language processing and understanding, and create a simple scoring rubric to sort, rank resumes in ways that are useful for recruiters to turn a pile of 500 into a shorter pile of 50 or 100, or a 100 pile into a pile of 10? So again, not making decisions for recruiters, not auto hiring, not doing anything that you need humans to be the arbiters of, but to shortcut that process and create 33% gain in efficiency when it comes to time from resume review to phone screen or phone screen to offer, and do those who are made those offers stay longer?
We had a great success story at JP Morgan where we saw recruiter efficiency go up, they reviewed fewer resumes, they got higher quality hires who stayed longer and performed better. So, you can embed that type of intelligence and validated IP throughout the hiring process when it comes to interview rubrics, Google is famous for this, and as well as the full array of talent decisions that you'd make when it comes to onboarding, promoting, paying, retaining, mobilising, growing your workforce. So, I think the solutions, they can range from I'd say productised and complex when you think about applied data science and product integrations, and you can also think about the more simple, maybe within reach kind of analytical solutions when it comes to like metrics, forecasting, and how that lands on the desks of HR leaders and business leaders in a consistent, single-source, hard-to-argue-with type of way.
So, for that reason, I think that back to the breadth comment earlier, I think orgs of coarse size and scale and maturity would probably relate to some things I'm saying here; but orgs that are newer in their journey, perhaps smaller in size and might consider people analytics out of reach, I think that's definitely something we should debate more and I would say that it's not as far out of reach now, it's more in reach now than it ever has been.
David Green: Yeah, I agree with that. I mean, we're seeing organisations with around 1,000 or even less now, you see them, they're hiring people analytics professionals, even as you said, even if it's one or two people, because there's always challenges that data can help shine a light on and help make more informed decisions about. So, yeah, it does seem to be that if we'd been doing this conversation ten years ago. We probably would, as you said, just been talking about some of the big organisations, particularly in the technology sector, but maybe some of the big banks as well. But now it is definitely more and more organisations, as you say.
Ian O'Keefe: You could argue, just to put a finer point on it, David, that organisations that can immediately identify and name who their customer is, if you have a branch banker who's your customer, I think that might be an easy answer that might be more immediate and different than if you ask a programme or product manager at a tech organisation who their customer is. We all have answers to those questions, but I think working backwards from improving your customers' experiences, that's very, very powerful. And I think that orgs that have more of a front-line presence, perhaps they can answer that question in ways that orgs that don't have that presence might be able to answer, and pros and cons to each, but again I think it's a really important focus for teams to have when they're thinking about putting these types of analytic capabilities into effect.
David Green: And if we think about the skills and capabilities required to productise people analytics, what do you think organisations or people analytics leaders such as yourselves, what do they particularly need to pay attention to?
Ian O'Keefe: Yeah, I think there's four or five capability areas. I've always built my teams around four or five common capability areas that I think any analytics team worth its weight in salt would embrace. One is data engineering, where's your data coming from; how do you land it; how do you organise it; how do you securely store it, create permissions, and create transforms that are science- and business-analyst-ready for inspection and organisation? Second capability is that analytics layer. Your analysts, your business intelligence engineers, people building reports, rapid response to executive queries, dashboards, building those self-service data products that leaders and teams can use, super-critical.
Third, and this is in no particular order by the way, I would point to your applied data science. This is model-building, predictions, forecasts that can then be put into production and integrated into products, such as existing HR software or something you might build in concert with the software development team. I think that applied data science piece, we could talk about software development as an extension. I think really, really talented applied scientists and data scientists, they index into some of the software development skill-T areas, but I think that's usually a partnership in any org I've ever seen.
The fourth area is what I would just call primary research. Think I/O psychology, think listening, survey work, focus groups, interviews. If you're pulling a number of data streams together, organising them on the back end with your engineers, transforming reporting on that with your analysts and model-building and predicting, ultimately to influence an outcome in the org, are you talking and researching your hypotheses with human beings that will ultimately use your product? If you're not doing that, I think that's a miss. And your researchers are often extremely talented partners when it comes to working with product teams and helping to translate findings into words and statements and takeaways that leaders can absorb and get behind. So I think there's adjacencies with product management and programme management that often come with your research capabilities.
Then the fifth area, something that Amazon is pretty famous for is its focus on applied economics and looking at causal evaluation measurement, experimental design, so that's something I think, depending on what your org is, is absolutely worth looking into. Something we had in place at JPMorgan that was a fifth area is your kind of business liaisons and translators. So, those are points of contact that can assess the current state when it comes to the analytic needs and opportunities in a particular line of business, based on what their people plans are to deliver to the business, and then to in turn point to pieces of your people analytic portfolio, whether it's on the engineering side or on the reporting side or on modelling side, to deploy perhaps against certain problem statements that a business is expressing.
That can take on a couple different I think permutations organisationally. Sometimes that's part of the people analytics team, sometimes that's embedded within the lines, and you kind of build processes and mechanisms to source and crowdsource perhaps that information on an intake basis. But I think that's really important to stay connected to the business and then translate that into how you prioritise your own roadmap, quarter on quarter, year on year, et cetera.
David Green: Yeah, I guess it's that important step, as you said, translation is one both ways, but it's identification and prioritisation, isn't it, linked to the business unit that they're closer to maybe than the core analytics team?
Ian O'Keefe: It is, 100%.
David Green: Really good. So, Ian, for those listeners that are looking to do maybe more experimentation, more productisation in people analytics, what are the key learnings and tips that you would offer to help them, based on your experience of doing this for a number of years?
Ian O'Keefe: Yeah, I would say that you really need to understand and identify those partner teams that you're going to have to work with; who are the product owners of the software applications that you hope to create an integration with; who are the tech teams whose roadmaps you're going to need to be on; what does that ecosystem look like in your work; do you have a tech team that's part of HR or a tech team that's borrowed, so to speak, from a central technology team that kind of runs up and down laterally in the organisation?
So I think understanding people analytics teams will not get this done on their own. You'll need to partner with legal compliance, product teams, tech, line teams, your HR leadership and line organisations from a communications implementation standpoint. So, rather than presuming to build this all out on your own and get it right from the people analytics team on out, there's I think a very necessary partnership approach that needs to be taken, and then kind of mobilise according to the vision that people analytics leaders can bring forward to connect the dots and be a bit of connective tissue across these partner orgs.
I'd also say I think it's really important to start small and scale fast, pick your use case and your problem wisely, narrow it down, make sure it's highly measurable, make sure you know what success looks like, make sure that you're able to build momentum and manage what will be a more complex undertaking than you realise if it's a first-of-kind undertaking. And the more you contain that within a group of importance, a group where you have good kind of buy-in and executive support, I think that helps you build that productisation flywheel and credibility frankly in ways that would be very, very difficult and challenging if you were to go out to many with sweeping recommendations and suggestions. So, that would be my advice.
David Green: And actually, what you've highlighted there, Ian, I think is something that's so important in people analytics, particularly for people analytics leaders I guess, is the ability to manage several stakeholders within the business, both right up to and including C-level, general managers in the business, your own team, IT, finance, legal, CHROs, HR leadership teams, it really is. It maybe is sometimes an undervalued skill, isn't it, that ability to manage those senior and, well, a range of stakeholders; it's a big part of the role?
Ian O'Keefe: Yeah, I think it is, and I think it puts that much more importance on the vision. If we all work together, here's the kinds of experiences and outcomes we can expect. If the org happens to already be organised in a way that cuts laterally across all of those functions, that's fantastic, you have an in-built mechanism and advantage to work with. But otherwise, and more often I think is the case, that we find that those domains tend to be organised as such and you need to plan to cut horizontally across them in ways that are, not to be corny about it, but that are inspirational, but practical and achievable. And I think you'll build this with vision, inspiration, and enthusiasm, and less so with organisational mandates and wielding a sledgehammer. So I think it's important to look at that as a partnership and less as a mandate, as it were.
David Green: Good advice. So, moving to the penultimate question, I can't believe we've got there already, but at the beginning of our conversation I asked you about your thoughts on how people analytics has evolved during your time in the field. Now as we come to the end, how do you predict or how do you think that people analytics will evolve in the years ahead? For example, what do you believe the role of generative AI will be in people analytics in the coming years? I couldn't not ask you that question, Ian, I'm afraid!
Ian O'Keefe: The generative AI question; yes! I think the jury's obviously still out, but I do think that generative AI and just technology advances that we've seen can create more highly participative notions of how people can share inside an org what they think, and kind of create new data streams in ways that we just haven't considered or seen before. So I don't think there's any area of HR, how we manage and think about the workforce, that generative AI couldn't be applied to. I think, again, I would probably go back to the notion of what problems are you solving and why; how are you working backwards from the customer?
I think applying a new, bright, shiny tool or object or approach without first considering what you're solving for and why, while doing no harm, in ways that you can start small and scale fast on, I think we have to put those principles around any new technology innovation that comes our way and make sure we're doing that responsibly. So, I think generative AI does really, really dial up the notion of very intentional ethics and legality when it comes to people analytics, and just applying data to inform decisions that affect employment in some way, shape, or form. There's a lot of laws and regs obviously emerging around that coming out of California and Europe, and that's a pretty fastly evolving space.
But again, I go back to working backwards from ultimately your customers that are going to be receiving and putting into action the insights and suggestions and all the good things that are coming out of your analytics programme. Start there and do no harm, and I think those are really good orientation points.
David Green: And how else do you see the field evolving in the coming years?
Ian O'Keefe: Well, I think the way that in-house teams, third-party software providers, and vendors, academics, even local, whether it's state or kind of market-based educational institutions, work together, I think that there's a lot of opportunity there that we haven't realised. I think the build versus buy argument and equation gets a lot more interesting as the start-ups -- there's a lot of really, really compelling Series A startups out there that are just pressing the boundaries within every kind of nook and cranny of the employee lifecycle that we can think of; and so, how you think about the problems that need to be solved and the opportunities that can be realised by looking at how HR can grow and cultivate and get the most out of its workforce, and have that be a great experience for people.
There are things that in-house teams perhaps are better positioned for, there are things that perhaps third parties might be better positioned for, and as we see advances in, whether it's AI or new types of data science, those are going to have to be taken into account and measured as part of how you approach problems in the toolkit you use to do that.
David Green: Yeah, very good. And finally, Ian, this is the question we're asking everyone in this series, and I think everyone's going to come at it with a slightly different perspective and background. What steps can HR leaders take within their organisation to humanise the work experience?
Ian O'Keefe: Well, I think we've heard this is age old, but it's important. People analytics is about people and it does affect lives and careers and we're not measuring widgets, we're dealing with humans. And I think sometimes that it doesn't get lost, but it could get out of focus, depending on the forces at work in the world and at companies, whether it's pandemic related or macroeconomic or return to office, layoffs. I mean, there's just so much that affects people and we have no idea really what people are bringing into the workplace every day. It used to be their lives and perhaps not settling for that, but being mindful of it, I can't see us not doing that and being kind of evangelists to that effect when it comes to what we might in turn measure and deploy and do to help make this part of the experience better.
So, everybody's coming to work with their own circumstances and those in recent years have gotten, I would think, a lot more complex just given everything going on in the world; and putting a little bit of empathy and humanity on top of that when it comes to our people analytics teams is the absolute least we can do to be better stewards of the craft.
David Green: Yeah, it's back to what you said earlier, isn't it? Do no harm.
Ian O'Keefe: Yeah, it is; do no harm.
David Green: Do no harm. A good way to end our episode, I think, Ian. Thank you so much for being a second-time guest on the Digital HR Leaders podcast. Can you let the listeners know how they can get in touch with you, follow you on social media, maybe find out more about your work?
Ian O'Keefe: Sure, LinkedIn is, I think, the best place. If anything I've said is peaking curiosity or you can relate within your org, I'm always happy to compare notes or to benchmark. So feel free to hit me up there on LinkedIn.
David Green: Ian, it's always a pleasure talking to you. Thank you so much for sharing your time, expertise, and knowledge with listeners, and I look forward to hopefully seeing you in person at some point soon.
Ian O'Keefe: I hope so too David. Good to talk to you, my friend, thanks.