What happens when HR stops guessing and starts predicting with AI and analytics

Guesswork is expensive. We dig into how HR and L&D can turn everyday signals—surveys, onboarding chats, training feedback, and attrition patterns—into reliable foresight with the help of AI. The goal isn’t flashy dashboards; it’s fewer surprises, smarter hiring, clearer development paths, and decisions that hold up under pressure.

TL;DR:

• Why HR data is underused and how AI compresses time to insight
• The difference between insight and foresight in people analytics
• When agentic AI helps and when it harms trust
• Lifecycle touchpoints to capture meaningful signals
• Asking better questions leads to better decisions
• Building analytics confidence and skills inside HR and L&D functions

We break down the difference between insight and foresight in plain language, then show how to move from reactive firefighting to proactive planning. You’ll hear a concrete turnover example where simple analytics smoothed recruitment cycles, reduced cost spikes, and protected onboarding quality. We also explore where AI earns its keep: structuring messy qualitative feedback, highlighting hidden drivers across teams and seasons, and shaping the same truth for very different audiences—executives who want scenarios and risk, managers who need timing and actions, and employees who want relevance and growth.

Along the way, we look at agentic AI and digital labour with nuance. Some tasks are perfect for automation; others demand a human touch. We unpack why exit interviews shouldn’t be your only listening point and make the case for onboarding interviews that capture expectations when they’re fresh—and influence the week‑one decision that often sets retention. Finally, we push on the skill that changes everything: asking better questions. When questions improve, data quality rises, insights sharpen, and the organisation gains a calmer operating rhythm.

If you’re ready to trade gut feel for grounded foresight, tune in and take notes. Subscribe for more practical people analytics, share this with a colleague who lives in spreadsheets, and leave a quick review telling us what question you’ll ask your team next.

Exciting New HI for HR and L&D Professionals Course:

Ready to move beyond theory and develop practical AI skills for your HR or L&D role? We're excited to announce our upcoming two-day workshop specifically designed for HR and L&D professionals who want to confidently lead AI implementation in their organizations. 

Join us in November at the beautiful MCS Group offices in Belfast for hands-on learning that will transform how you approach AI strategy. 

Check here details on how to register for this limited-capacity event - https://kierangilmurray.com/hrevent/ or chat https://calendly.com/kierangilmurray/hrldai-leadership-and-development

Chapters

• 0:00 Why HR Underuses Data
• 1:40 AI As Analytics Superpower
• 2:40 Insight vs Foresight Explained
• 3:45 Predicting Turnover Patterns
• 5:25 First Steps: Data That Matters
• 6:13 Smarter L&D Measurement
• 8:00 Presenting Insights to Audiences
• 9:45 Agentic Labour Pitfalls & Promise
• 11:20 Onboarding Interviews Over Exit-Only Focus
• 13:20 Lifecycle Data Touchpoints
• 15:05 Week-One Retention Leverage
• 16:35 Adaptive Induction With AI

Transcript

SPEAKER_00: 0:03

Baines say that if executives, particularly HR leadership executives, Claire, if they use decision insight, in other words, great data analytics to make good decisions, not just their own gut, then there's a 95% correlation between a good decision and a good business outcome. So why is it that data appears to be such an underused asset or such an underused skill, data-driven insight, when it comes to HR and learning and development professionals, or have I got that wrong?

SPEAKER_01: 0:38

I think you've absolutely got that right. If you think about the scope of the role of a human and resources professional in an organization or an LD professional in an organization, it covers a lot of facets. There's a lot of areas that they have to action and a lot of hats that they wear in the business. When it comes to data and data analytics and taking that beyond what's the data showing me to what's the data telling me is going to happen in the future, how do I move to that predictive analytics? That takes time. So if you're reprioritizing based on demand in your certain role or things that have happened in the business, those professionals don't necessarily have the time to get into the detail about what the data actually says. And beyond that, that then needs to be formatted for your audience. So presenting that internally to your team or an operational team is very different to presenting that to a boardroom executive panel who need to make critical business decisions on the basis of what you're presenting and what that means. So for me, I believe that AI really does change the landscape of that because it speeds up those analytics. It actually, I always say this, it supercharges us, it's our superpower because it gives you the insights the human eye can't see. Remember, this is about artificial intelligence. Humans have a fantastic creative mind. Specifically, HR and L and D individuals have been trained to be critical thinkers. But when you layer AI into the data analytics that you're seeing in a business, it expediates what you can find, what the underlying information shows, and moves it more into a predictive base where you can start to look at okay, this is how we've got to where we are. Show me what the future landscape is going to be on the basis of the data that I'm now seeing. And for us, that's a skill that needs to be developed. We need to see how we can introduce AI into those processes so we can deliver really data-rich, information-rich analytics that actually inform those decisions. For me, the most valuable points for HR and LD are those critical employee lifecycle areas. So, why are people joining our business? You know, what is the sediment of those new starts? What is our training and development like? Is it actually useful? Are we actually training and developing the people to the needs that they have personally for their career progression, but also to the business direction in the future? What are absence rates like and how does that look over a year? Is that influenced by particularly high characteristics or certain areas? And what layers do we need to add within the analytics, maybe that we're not even thinking about? And I think the AI expands that thinking for me. Uh would you think the same, Kieran?

SPEAKER_00: 3:20

Yeah, I do. Uh you made a great point a moment ago, Claire, talking about what I call insight and foresight. So, insight, you there's a set of analytics called descriptive and diagnostic. Sounds very complicated. There's not really uh it's sort of describing what's currently happening and why is it happening. You've mentioned a variety of reasons there, you know, seasonal turnover. You might know it, but there's certain details you might not see. Let me give an example. I worked in an insurance brokerage years ago, and we kept reacting to people leaving the business. And of course, people would leave, everyone would panic, you'd suddenly have to go through a hiring process, all the plans that you had laid out to get projects done and training done just disappeared as everybody firefought. We we used analytics to look at you know recruitment and exits over the year. And to your point, we saw very predictable patterns. And then, you know, at a low level, then we started to investigate well, why was this? And then, of course, you get your usual, you know, annual review pay time, someone wasn't happy, they left bonus time, but there's a whole host of other reasons during the year, but very, very consistent. Once we were able to see that, then what we were able to do is start to put plans in place, not to over-recruit or overhire, because that creates operational costs that most businesses can't afford. But when we started to see, you know, and I'll make up the days, but every September we were going to lose three or five people. Well, we started to work back from that, you know, 12, 18 weeks to go, okay, when are we going to hire? Who do we want to hire? You know, can we get everybody lined up, ready to go? So we don't have, you know, managers busy on one project when we need them to interview and onboard and everything else. It really smoothed operation, smooth cost, made things affordable, didn't interrupt managers or other team members when these people were onboarding and learning. The predictability helped us. Now that allowed us to see what was happening, when it was happening. And to your point, the predictive analytics, where you would use AI, is telling us, well, look, if this happens this year, what's going to happen next month and next quarter and next year? And when you combine foresight with insight, you just made better decisions. But I suppose, Claire, you know, but people listening into this will look at AI and go, well, hold on, I haven't even got the first lot ready in the first place. So before you start talking about the future, what about now? So uh to be transparent, you know, data analytics and AI is not the natural general home of HR and L and D professionals. So, what are those first steps? What are those first valuable data points that teams should focus on? And whose support do they need to potentially get those, do I say, dashboards, insights, whatever in place to allow them to start to make more data-driven decisions or data insight-driven decisions?

SPEAKER_01: 6:13

If we think even about the learning and development landscape and the sources of data and information that they have access to, even something as simple as your training feedback. So, post-delivery, where you're trying to measure that employee's output, what's their thoughts and feelings, and maybe even three to six months down the line, how have they embedded that knowledge and dissipated that across an organization? That can be really difficult to put a pin on, to measure, to put a number against. And most of the time that is qualitative information that is coming back, but some of it is quantitative. Taking that as your first step and then applying AI and how you frame the data is sometimes the easiest step to make. We're so used to working in images, so the human brain works in pictures. So when we're presenting information or analytics, we have our thoughts around what that should look like, or the executive team have ideas about how they want that paper presented, that presentation to be delivered, what models they typically use. Think about typical analytics, you're talking about bar charts, line charts, pie charts. But what if they could be dynamic and what if they could be really interesting and really grab your audience? So it's not only about how you're manipulating and measuring the data and actually looking at the underlying information, it's also how you're modelling that and presenting that as well. And AI can be used at all of those points to try and get the right message across to the different audiences that potentially L and D are going to use. And I think giving them also avenues to regularly update that information. We know that you know, staff surveys can go out, they're measurements at a point in time, but if they're not regularly reviewed, and again, something that takes time, they're useless, they're not being acted upon, they're not they're not giving two set points to measure between, and they're not predicting any future outcomes and changes for the business. Again, AI can be used to expediate that. So that's something that would have taken LD and HR professionals time to build those surveys, to manipulate the data, to analyse the information. But if they build an underlying measurement that's there, where it can be regularly reviewed and refreshed and updated and present that back actually to the employees, they can start to see, well, actually, this is how my input is being valued by the organization, and this is how it's now going to be used to formulate future training and development plans throughout the company. And small things like that, that's something that is a small element of what those professionals do. That has ramifications in the likes of career pathways in an organization. So it's starting to carve out directions for the business and how they're investing AI in particular areas that actually lead to measurable outcomes for the organization, but also for the employee. And it's having that happy medium between the two. We want AI to be used responsibly and ethically, but also that those that are using it know why the businesses are using it and how it's actually going to impact them directly. So I think it'll help as well. HR and LD professionals speak that language of the boardroom. So it actually allows them to target their individual audiences and see what's actually relevant for this area and how do I do this or manipulate it in a different way. As I said before, what you present operationally is going to be very different to what you present to an executive. And it's it's seeing what those audiences want to see, but using AI and its application to actually deliver those measurables for them too.

SPEAKER_00: 9:41

Yeah, I suppose that's the challenge, isn't it? Because you have different audiences and everybody wants it presented in different ways. So you end up with the green version and the yellow version and the blue version. And I would say, you know, the correct prompt will allow you to do that. The correct AI will automatically build the dashboards as well. But you also made an interesting comment earlier on, you know, are we actually capturing the information that we need in the first place? I had a conversation with a with a large group last week who were talking about exit interviews and they were talking about using agentic labor, digital labor to perform the exit interview on their behalf. Now, you have to be careful with that because statistically, you know, when you put an agent, a digital worker, performing the exit interview, and this can sound very emotionally intelligent and do a great job, you click on the link, it'll do it whatever time of day or night you want. With a younger audience, you get a 23% uplift. With an older audience, at times, and this is very general, not specific, they're frustrated that they're leaving the company, they're already upset, and the last thing that you're doing is not even bothering to speak to them. You know, so again, to implement some of this technology, you really need to think it through as well. Now, the bit that I said to them, they were very keen to do exit interviews, but you you made this comment earlier on. What about your entrance interview? So, how many people actually perform an onboarding interview? You can record it using, you know, Teams, Otter, Fireflies, anything, subject to the person in front of you agreeing, of course, because we want to do this responsibly. If you did 10 or 20 of those, put that transcript into ChatGPT, or ideally a secure uh version of it, like co-pilot in your own business, then you can get the AI to give you an assessment, a breakdown, a word cloud, insights that you may have missed over a trend of individuals, and then all of a sudden you can start to adjust your workplace to make sure you do the things that keep those employees in the business because they've left somewhere else and have been very frustrated with what was happening there, rather than repeat the same frustration. Let's just get it right from the first time. So I wonder, Claire, is there a sequence of points in the HR lifecycle where we should actually be creating data of some sort to allow us to make better decisions? Because there doesn't appear to be enough hours every day to collect all the information that's available to present it to all the different audiences in the huge variety of ways that they want it. Or do we just need to be a little bit firmer as HR and L and D professionals and say no?

SPEAKER_01: 12:18

Well, there are only so many human hours in a day, but when you start to layer in a gentic labour, and I know we're going to talk about that in another podcast, it really starts to grow your thinking around it, and you start to think, okay, so if I could split myself in two and take all those administrative tasks that I generally do and require time, dedication, focus, is AI appropriate for that? If it is, okay. So if I sidetrack that hour and a half, there's an hour and a half a week that I can then roll up into six hours, maybe even eight hours a month. And when you measure that out throughout a year, yes, you start to kind of grow that you call them as sometimes digital twins, you know, those other parts of you where you're using AI to replicate the role. But coming back to your point about at which points in that employee lifecycle do we start to gather data? Honestly, at every opportunity, because you can start to connect different sources of information through the likes of Agentich AI and really look beyond the data and see what is it telling me about the business and our people, not just about how does this new start feel one week after they've started with the organization. Um, and funny, you're talking about questions as well. Yes, we can. I think that the typical example for HR is you know, use GNAI to generate a job description. Yes, that's great, but think beyond that. There is so much more that you can apply this to. I I read an article actually recently where it did say, and it was talking about exit interviews, and someone was asked the question generally, why are you leaving? Okay, where are you going to? But they were asked the question, what will we need it, what would we have needed to change to make you stay? Which I thought was so interesting. So it's not only about using the AI to capture the information, it's about changing how we think about what we're asking. Because if we're going to get all this information and data, the questions need to be good for our employees. We need to understand why, what's impacting their feeling, how how engaged are they in the business, you know, how much do they want to stay before they're actually making the decision to leave. And that's at various touch points throughout that life cycle. I think it's within the first week, um, 90% of employees will make the decision within their first week's experience whether or not to stay or leave an organization. And you think about the resources that it's currently taking organizations and the candidate permanent marketplace to attract and actually get that person into the front door the cost of that, and then you have you have the ability to control that in the first week, the cost of that can be colossal for organizations. So why not invest in some form of AI or some form of process that's associated with AI to try and start to capture and engage that employees and individual and generate the excitement about what the existing employees be advocated at in the first week, and then there's other touch points throughout the role. You know, when you think about induction, for example, that new start will have an LD package that's been associated with them. But if you start to make that really interesting and role-specific in using AI engagement, you know, we've seen some of the models where you can actually have live conversations and it will give you feedback on those. I've also seen examples where that induction training, as simple as sometimes some of the areas can be, we can start to see feedback given to the employee to say, great, you know, you you've progressed through this much of your training, this is what you still got to go, but here's what you've learned so far, and this is why it relates to your organization. Here's our organizational goals, this is our strategy, and this is where you sit in the scheme of our employees. So you you really start to layer it once you've got the basics in place. Um, but again, it's back to skills and abilities and knowledge. So, how confident are you in actually exploring that particular process and how AI might actually fit into it? And I think that's why we need to start to really drive that capability in HR and L and D and increase the confidence and skills of those people so that they can then assess to make those strategic decisions.

SPEAKER_00: 16:25

Yeah, I think that goes back to, isn't it, it's not just the skills of the employees, but it's also the skills of the HR and L and D leaders inside of the business who then need to pass the skills on to everyone else. Well, look, there's good news, or two pieces of good news. What you're describing there is if we can ask better questions, we can ask better, we can get better insights of the data itself. And you and I are running a course, part of the course is actually data analytics and AI for HR and L and D professionals, so that you can learn to ask better questions. And by the way, Centure said if you learn better questions, not only are you more effective, but it makes you a more interesting person as well. We'll teach you how to ask those questions to get those insights to allow you to come up with the things, the hidden things that nobody else can see to allow you to improve every aspect of the learning cycle. Claire, until we talk next time, let's see how folks get on becoming more interesting through data.

SPEAKER_01: 17:20

Thanks, Karen.