HR teams are ready to embrace AI. Their data is not.
Jump to section
Everyone is rushing to use AI. And it’s easy to worry about falling behind or getting it wrong. But we’ll let you in on a secret: those worries are a trap.
Because when everyone’s racing towards the same goal, speed, spending, and access to similar tools won't differentiate you. What sets teams apart is the foundation AI is built on. Right now, that’s where HR has the most work to do.
HR is all-in on AI, but it's only skin-deep. 89% of HR leaders use AI in HR, but only 14% have it deeply integrated with their people data.
AI underdelivers when the data is weak. Organisations without accurate data are eight times more likely to say AI fell short of their expectations. But only 21% have largely complete data.
Employees are going it alone. Just 31% of employees get adequate support with AI. 23% are using it without their organisation's knowledge.
Workforce Pulse is the largest survey of HR leaders in Europe. More than 3,000 HR decision-makers and 6,000 employees told us where AI is paying off and where it’s stalling.
Data is what separates AI that delivers from AI that disappoints. Every team can close that gap, and this report shows you exactly how.
Key findings at a glance
67 %
of HR leaders rate business performance as high or very high
49 %
of HR leaders say it's still hard to recruit
51 %
of employees are looking for a new job — and 26% plan to quit within the next six months
21 %
of organisations have largely complete and accurate people data, versus 37% with significant data gaps
65 %
of HR leaders have made a big decision without full confidence in their data
89 %
of HR leaders use AI in HR processes, but only 21% say they have the complete, accurate people data to support it
14 %
of HR leaders have AI deeply integrated with their people data
8x
HR teams with incomplete or inaccurate people data are eight times more likely to say AI fell short of expectations than those with complete, accurate data
31 %
of employees say their organisation provides adequate support on AI
23 %
of employees are using AI without their company's knowledge
26 %
of employees rarely or never get a clear explanation of how decisions are made
Foreword
This is the third year we've run Workforce Pulse, which I’m proud to say is the largest survey of HR leaders in Europe — over 3,000 took part, alongside 6,000 employees. The results are eye-opening, especially where AI's concerned, and three big opportunities stand out to me this year.
The first is in the budget conversation. Budgets, recruitment, and pay expectations have all cooled since last year. The one budget item holding firm is AI and automation spend, with 62% of HR leaders still expecting it to rise. For me, the opportunity is in using automation to get more out of our other budgets, rather than waiting for them to grow. In practice, that means automating manual work, simplifying how we do things, and enabling managers to take ownership of tasks we used to hold onto too tightly. It's about generating capacity on the team I already have, so we can spend our time better.
The second is in how we treat our own knowledge. Only 21% of us have complete, accurate people data, with 37% falling into the ‘data-deficient’ camp. I think this is a symptom of how HR has historically stored our information: processes are documented retroactively, then data is kept in a million different systems. I think we have to go through a paradigm shift in the HR function, treating our knowledge and our processes the same way technical teams treat code. That means it all needs to be stored in one place. There needs to be clear governance of how changes are made and who approves and reviews them. That is the critical foundation for making AI agents truly work for HR.
The third opportunity is how we talk about our decisions. About half of us have made a big call this year without full confidence in the data behind it. That alone is no surprise. I'd argue that's simply part of the job: making decisions with incomplete information. But here’s the big opportunity: Only 20% of employees say they always get a clear explanation of decisions that affect them. HR has good reason to be careful with people’s data, since much of it is genuinely confidential and sensitive. But this can also quickly turn into a convenient excuse for us to not explain our decisions to the degree we could — and should. Sharing the rationale and considerations behind a decision opens you up to scrutiny, feedback, and criticism, which can feel uncomfortable. But I believe it is a vital mechanism that ultimately will lead to better decisions, higher trust, and buy-in.
If anything, I see this year’s insights as an enormous opportunity: to do things differently, to reinvent how we work, and to use the current spotlight on people leaders and teams to take our impact and contribution to the next level. That's the habit I want to build into my own team this year.
Philipp Richter
Chief People Officer at Personio
Regional breakdowns
42%
of UK employees use AI in their role
26%
of UK employees use AI without their employer knowing
42% of UK employees use AI in their role, the highest of the regions we surveyed. However, 26% use it without their employer knowing, also the highest. UK employees are motivated by the new tech, but their black market AI usage poses a unique threat to organisations: data security.
The compliance backdrop of the UK makes this a trend to watch carefully. Detection of unsanctioned AI usage is a challenge for all businesses, regardless of whether or not they encourage employees to use approved AI tools. It seems UK organisations are suffering from engagement that has outpaced governance.
This tension is unlikely to resolve on its own, as employees will likely continue to find gains in shadow AI. For HR in the UK, the challenge is clear: turn sanctioned AI use into the easier and more fruitful path.
2025 might have felt like the comeback year for HR. After years of second-guessing every move under constant pressure, things felt like they were on the up. Healthier budget lines beckoned, and fewer redundancies were on the horizon.
Things felt like they were on the up. But what we now know is that 2025 was the exception, not the rule.
Compared to 2024, 2026 is still feeling like an optimistic year. But there's definitely that old familiar hint of caution coming in on the breeze.
How do you expect the following to change in the next 12 months?
% of HR leaders expecting an increase
HR budgets
51%
53%
44%
202420252026
Overall recruitment
50%
52%
43%
202420252026
Average pay
54%
55%
47%
202420252026
Redundancies
60%
39%
38%
202420252026
Business profitability
51%
56%
50%
202420252026
Business performance
56%
69%
67%
202420252026
HR's year of guarded confidence
Expectations for hiring, profits, pay, and budgets in the year ahead all fell compared to 2025. Among these, fewer leaders expect to see increases to HR budgets and recruitment efforts, which both slid nine percentage points.
But while numbers slumped a little compared to last year's high, HR's overall sentiment is still well above 2024 levels.
Redundancy expectations have stayed steady at 38% compared to 60% two years back. Business performance also kept its footing, with 67% of HR leaders rating it as high or very high.
The economic landscape and the business constraints at the moment means budget, recruitment, availability, profitability is right at the front and centre of the table. That means recruitment will continue to be harder. It's more expensive to make mistakes.
Talent pressures are easing, but it's still hard to hire
Compared to 2025, fewer HR leaders believe that talent and skills shortages will be their biggest barrier to business success in the next 12 months (33% versus 39%).
But that doesn't mean hiring is easier across the board. Almost half (49%) are finding it hard to hire, suggesting that pressure is becoming concentrated around a few key capabilities and candidate profiles.
If you can't hire great talent, the logic follows that you need to develop it from within. Trouble is, a significant proportion of employees are planning to quit. And they're planning to do it soon.
In 2026, employee churn intentions are still high, but they are steady. Over half (51%) of employees say they're actively seeking a new role. 26% are planning to quit within the next six months. For most teams operating with budget and skills constraints, that's not a lot of runway.
What do you consider as potential barriers to your organisation's success in the next 12 months?
What good AI looks like: Automate workforce and scenario planning
The problem: Headcount planning is fragmented across spreadsheets, finance, and hiring managers, and it's not responsive to changing business objectives.
The use case: Use AI-assisted tools to build a workforce scenario planner that models different organisational structures, geographies, roles, skills, levels, and costs.
The outcome: Talent mapping and decision-making became more visible and easier to share and change.
One company I worked with was scaling across multiple jurisdictions and hiring into different departments. They built an easy way of modelling different scenarios over time — what levels, skills, and capabilities would be needed, and how much that would cost in terms of headcount growth across different departments. They used Lovable, then added Google single sign-on, a proper backend, and deployed it with Vercel.
That allowed the talent mapping to become much more open and obvious. They could map out scenarios visually, share them, and make edits quickly. They had a clear, centralised line of visibility for what the decision was and its reasoning.
Uncertainty about what the future holds is rippling into how HR leaders feel about day-to-day life at their organisation, too.
Compared to 2025, HR leaders are less likely to rank employee engagement, HR-function effectiveness, and trust in leadership as high.
Part of this problem could be down to data. In 2026, HR leaders are less likely to agree that they use employee feedback to inform their decisions, and their confidence in being a data-driven function has taken a knock (61% in 2026 versus 66% in 2025).
As HR data begins to sprawl across systems and platforms, leaders feel less able to act with confidence on the information they have right in front of them.
HR is now standing in front of a data firehose. We've all got too much of it. But you can't invoke change for the business when you're sitting with data you're not able to use. And there's a very big difference between having lots of data, and having decision-grade data.
Fractional global talent acquisition lead, Maverick Otter
How would you rate the following in your organisation?
% that said "high"
Employee engagement
65%
63%
20252026
HR function effectiveness
65%
62%
20252026
Trust in leadership
66%
62%
20252026
Use employee feedback to inform decisions
69%
65%
20252026
Confidence in being a data-driven function compared to two years ago
66%
61%
20252026
Trend 1
The AI reality check
Most of HR's AI problems aren't actually AI problems.
By now, AI adoption in HR is almost everywhere. Everyone already has the shiny new thing.
And in some organisations, it's really doing the work. The headcount scenario planning process that used to take three departments two weeks huddled around a spreadsheet now gets modelled in one afternoon. A manager wanting to run better one-on-ones spends more time on the conversation, not the prep work.
But in others, it comes with shiny new problems. Agents that don't talk to one another. Decisions made on unreliable insights. Work that gets done twice. There's a niggling feeling that it hasn't made their processes better or faster.
And maybe it hasn't — yet. But it will. To turn that corner, these HR teams need to adjust their mindsets: they've just been fixing the problems they can see, rather than looking for the root causes.
AI scales whatever sits underneath it: good processes or bad processes, strong data or weak data. It's down to us as leaders to build that framework and dictate what “good” looks like for us.
Fractional global talent acquisition lead, Maverick Otter
HR's appetite for AI is strong, but something's still holding it back
Knock, knock: AI has come calling, and HR has flung the door wide open. Nine in ten HR leaders already use AI in their processes. Adoption is clearly not the problem.
89%
of HR leaders already use AI in HR processes
14%
have AI deeply integrated within their people data
I see a lot of HR teams investing in all the AI platforms and ending up with agents and different tools that don't talk to each other. Then they have this whole AI landscape and ask, ‘What do we do with it?' We're trying to be more intentional, focusing on where AI can save time within the work people operations is already doing while also challenging how work gets done and driving continuous improvement across people operations.
Plugging AI into HR's data and existing tech stack is proving tricky. Only 14% of HR leaders say they have AI deeply integrated within their people data, leaving most with a frankenstack of siloed platforms that don't quite connect. But focusing on AI integration is the wrong starting point — the problem actually starts further back in how teams designed their processes before bringing in AI.
“The gap between using AI and deeply integrating it comes down to one thing: process redesign,” says Puren Ucar, founder and adviser at Milele. “Most teams I have seen using AI without results have simply added tools on top of existing workflows. The ones actually seeing results have redesigned the workflow around AI from the start.”
Which processes are HR teams supporting with AI?
Emerging trend
Integration impact
HR leaders said integration is the top reason they're not seeing value from their AI tools, cited by one in four (25%). And the data suggests that depth matters.
6x
Organisations where HR leaders said AI exceeded expectations were six times more likely to have the tech deeply integrated compared to those where it fell short (42% versus 7%).
7%
Only 7% of organisations where AI fell short said it's deeply integrated.
2x
HR leaders who rated their ability to make data-driven decisions as “high” are twice as likely to have AI deeply integrated into their data compared to those with weaker capabilities (17% versus 8%).
Is integration depth a good signal for AI success?
As a rough proxy, yes. Deep integration tracks with AI delivering value, but it's a marker of maturity, not a shortcut to it. The thing that matters most is where teams integrate.
Weaving AI into a process that's poorly designed or didn't need it, and all you've done is embed it in the wrong thing. And in a lot of cases, AI might not actually be the answer to the problem, but your workflow design might be.
“AI may be deeply integrated at some organisations, but it could be in the processes that don't actually need it,” says Adam Horne, co-founder of Open Org. “The bigger issue is that many people teams haven't clearly mapped out where their real problems are, and whether AI will add meaningful value in that specific use case.”
To identify where integrations are adding value, ask:
To identify where integrations are adding value, ask:
What problem is AI meant to solve in this workflow/process?
Are we seeing measurable results from integrating AI here?
Has it actually removed work, or has it displaced it elsewhere?
Could the same problem be solved by redesigning the process?
More money, same problems
If you've been wondering if everyone else's AI budget is plumper than yours, let's just say the grass isn't greener.
Budgets across the board are actually pretty modest, with UK organisations spending £15,550 on AI in HR per year on average. Over half (57%) of HR teams spend nothing at all, use AI features bundled into existing products, or keep AI costs under £5,000.
Bank-busting budgets or not, the best predictor of overall satisfaction with AI tools is the quality of your people data. When teams do spend money on AI, only 6% of data-ready teams (those with largely complete and accurate data) say their tools fell short, compared to 22% of data-deficient peers (those with largely inaccurate and incomplete data).
What does HR actually spend on AI?
AI only delivers when the processes underneath it are solid
Two organisations can buy exactly the same tool and have very different experiences depending on what lies underneath. But they might also see very different outcomes — the kind that impact employees' pay, progression, and experience at work.
Over a third of organisations that say they're using AI-generated insights (35%) or predictive analytics (37%) to make people decisions have significant gaps and inaccuracies in the data those tools depend on.
“AI is exposing problems that already existed — the shortcuts we've taken,” says Dr JooBee Yeow, adviser to founders at Learngility. “If organisations have tolerated inconsistent data, unclear processes or poor governance for years, AI simply makes those weaknesses more visible and sometimes amplifies them.”
Overall, only 15% of HR leaders using AI say it fell short of expectations. But that risk isn't spread evenly: organisations with the most complete people data are the least likely to feel let down, while those with the most incomplete data are eight times as likely to say the same (19.8% versus 2.6%).
What to do next: Make AI prove its worth
When teams integrate AI effectively into HR workflows, they're far more likely to see and feel a payoff from their spend, whatever the budget. The trick is working out where AI will genuinely add value, and then making every new rollout prove that it does.
1
Challenge your desired outcomes and metrics
Making your AI budget work harder starts with knowing what you're trying to achieve. If your reasoning is to make things faster or better, question why first.
“Most people start by asking, ‘What AI tools do I need?'” Theen says. “But when I ask what they actually need help with, the answer is usually, ‘I need this to be faster.' Faster in what? Maybe much of what you're doing is redundant. Take the manual work out first, then work out where AI actually fits.”
2
Map every single workflow, step by step
When you integrate AI into a bad process, you still have a bad process. Mapping your workflows to identify where teams are duplicating effort or finding friction is essential. But doing it in real time is even better.
“In a lot of the projects I work on, much of the actual work is just fixing the underlying workflows,” explains Matthew Bradburn, founder at Deepgrain.
“Screen record every workflow at an atomic level. Who clicks on it? Where are the friction points? Where do they have to move things around and shift between platforms? Do they even need this workflow at a human level before they even get to the AI side? Breaking your processes down like this exposes where HR teams perceive there are problems versus the reality.”
3
Use data quality as a requirement for every AI project
When data quality is one of the biggest predictors for AI satisfaction, it becomes an effective gate that every project pitch needs to clear before it gets the green light.
“Use the AI rollout itself as a forcing function for data hygiene,” Ucar says. “Most of the AI projects I have seen fail do so because the underlying data is incomplete. One company I work with now requires every AI project to commit to a specific metric target before it gets funded. That single rule made data quality a precondition for AI investment rather than an afterthought, and it changed the conversation faster than any data audit had.”
Trend 2
The data gap
A strong people data foundation relies on data quality and capability. Few teams have both.
In an ideal world, every decision HR makes is backed up by exactly the right data. Every data point updates in real time, and your tools all shuttle that data back and forth without you needing to lift a finger. Oh, and your AI features work like a dream because they rely on an accurate, single source of truth.
But this is reality. Things are messier here.
Decisions get made quickly in the flow of work based on the data you have, and the instinct you've spent years honing. Auditing the data you have versus what you need costs time the business can't afford. Updating the data keeps getting pushed down your never-ending to-do list.
It's not just you, and it's not a failing. Because as it turns out, almost everyone is working around data gaps somewhere.
Organisations are telling HR to buy the AI tools and plug them in. But if we don't have a fuse board to take the power, you'll end up blowing the fuse.
say their people data capability is “strategic” or “effective”
21%
report having largely complete and accurate people data
Only 21% of HR leaders have access to complete and accurate people data. Nearly double (37%) sit in the data-deficient camp, wrestling with significant gaps and inaccuracies in the data they rely on for everyday people processes.
Yet almost half (49%) rate their data capability — what they're actually doing with the data they have — as 'effective' or 'strategic.' And among those who say they use data highly strategically, 40% don't actually have the complete, accurate data foundation to back that up.
The other half acknowledge that their data capability is limited, and that they struggle to use it for decision-making. Viewed together, they feel like two sides of the same story.
“The HR function as a whole struggles with a data literacy gap,” says Monique Major, Senior Director of HR, People Experience, and HR AI transformation at Diligent. “But there's an undercurrent to that — HR leaders' perceptions of their own literacy. And that may even be the bigger gap.”
Few HR leaders have time for a beginner-level statistics course. But the gap is causing some to think differently about who sits on the HR team.
“One of HR's biggest challenges is that we have a vast amount of data, but using it effectively requires a completely different skill set,” says Katharina Kuhlmann, Global Director of People and Culture at Harren Group. “When I recruit for our team, I focus on the skills we need, where we want the role to take us, and what that person will add to the team. Experience in the exact same HR role is no longer a requirement.”
HR leaders aren’t feeling confident in their data
51%
of data-ready leaders have made a decision without full data confidence in the last 12 months
75%
of data-deficient leaders have made a decision without full data confidence in the last 12 months
HR's hidden data gap undermines decision-making confidence across the board. Three-quarters of data-deficient leaders have had to make a decision without full data confidence in the last year. But data-ready leaders (the 35% whose people data is complete and accurate, or close to it) aren't immune, either.
Clean data doesn't always guarantee an easy decision if it's not deep enough to tell the whole story. And by the time that decision gets made, the data might even be out of date.
“HR data changes daily, but the expectation to make decisions quickly remains,” says Aaron Windrum, principal HR consultant at A Human Resource. “Leaders are operating on the basis that imperfect data is better than no data. They're working with what they have.”
Employee survey data, manager feedback, and performance metrics are among the top data sources for HR decision-making. Leaders clearly have volume, but they might not have enough depth to inspire confidence.
“People data has so much context that you don't always get from a report unless you're very diligent about input,” says Yasmin Chalmers, Head of People at StackOne. “If a candidate rejects your offer, for example, you need to capture why rather than recording that they accepted another role. Otherwise, the data becomes too high-level to be useful when you come to analyse it.”
What sources of data is HR using for decision-making?
Data is a daily pain, but not yet seen as a long-term risk
While HR teams are aware of the challenges with their data, they don't always prioritise the knock-on business impact.
Only 15% of HR leaders consider a limited ability to make data-driven decisions as a key barrier to their organisation's success in the next 12 months. HR leaders were more than twice as likely to name talent and skills shortages by comparison.
It's a framing problem. Talent shortages are visible and immediate, always elbowing their way to the top of HR's to-do list. Meanwhile some inconsistent data here and there feels more like an operational annoyance, not a strategic risk. And that makes it much harder to connect to its impact on the business.
HR often measures the quality of HR data in isolation rather than asking whether it connects to the business outcomes they're trying to influence. If you can't clearly explain how your people data links to operational or commercial performance, then the data isn't yet strategic… regardless of how polished the dashboard looks.
Building a stronger data foundation starts with understanding exactly what kind of data you have, where it sits, and what purpose it's serving.
1
Work backwards from business objectives
A strong data foundation is built on curation. Teams need to find the metrics that matter to the business, and do those ones well before adding in everything else.
“The instinct is to collect more data,” Yeow says. “I think the opposite is often true. Start with the business decision you're trying to improve, then work backwards. What business outcome matters? What lead measures indicate progress? What people conditions influence those measures? Only then should you decide what data is worth collecting.”
2
Triage the data you do have, ruthlessly
“Half of HR leaders are admitting point blank, ‘We don't know what we're looking at',” Theen says. “But for the other half, some of that confidence may be coming from familiarity. If leaders are working from the same limited set of data points year after year, it's easy to mistake comfort for effectiveness.”
Review the metrics, reports, and dashboards your team produces or tracks by default. Evaluate whether each one still helps you answer a meaningful business question or make a decision. If it doesn't, consider if it's still strategically relevant.
3
Identify where you can enrich the context
Once you've established which data you need, look more closely to see where it's only giving you a partial read on what's happening, and if you need more information. This is the context that helps your AI tools work more effectively.
The stakes of the decision change the amount of detail you need.
“Whether the data is detailed ‘enough' depends on what you're trying to decide, what the stakes are, the timeframe, and what information you'll rely on instead,” Chalmers says. “For example with salary benchmarking, if you realise the data you have to inform you on market rate isn't detailed enough, the consequence of getting that wrong is too big if you go on gut feeling.”
To identify where your data needs more context, ask:
Why do we keep getting this outcome? Do we need more information on behaviours or reasoning behind decisions?
Do we have internal terminology or definitions that add meaning? (for example, what does quality of hire or high performance mean to us?)
Would different teams interpret this data point the same way?
Do we need situational context around when or where something happens (for example, which stage in your hiring funnel sees the most candidate drop-offs?)
4
Make data quality a KPI
Keeping your people data accurate and up to date requires ownership. Scaling that ownership across the HR team means embedding it into everyday behaviour.
“The most underused lever is also the most direct one: tie data quality to performance,” Ucar says. “People teams should have explicit KPIs around data completeness and accuracy, tracked, reviewed, and connected to how the function is evaluated. When clean, consolidated people data becomes a measurable expectation rather than a best-efforts aspiration, behaviour changes.”
What good AI looks like: Triage data gaps with agents
The problem: You're struggling to see data quality issues and inaccuracies.
The use case: Build an AI agent to identify your data gaps and inaccuracies, then run autonomously when new ones appear.
The outcome: Eliminate the manual data work, and improve accuracy of decision-making.
We track a lot of data, and it's sometimes difficult to know which is important versus what we're tracking because we always have. AI makes it much easier to say, ‘hey agent, flag gaps in our employee data'. Our primary goal is to reduce the manual work involved in auditing our data. The next step is to improve the quality so we can derive deeper insights.
HR leaders say decisions are more data-driven than ever. Employees say they still don't get an explanation.
Since the dawn of the AI era, HR leaders have been trying to answer one question: How do employees feel about the machine among us?
The media is full of screaming headlines about job loss and skills shortages. Whispered tales beside the HR water cooler tell of shadow AI use and intentional sabotage.
If this is our new reality, it feels stranger than fiction.
The good news, then, is that some of it may have been a little bit overblown. Because what employees are actually asking for is a whole lot simpler: trust, clarity, and transparency.
The human challenge is resisting the temptation to treat AI as another shortcut. Technology cannot compensate for unclear ownership, inconsistent decision-making, or weak operating discipline.
If HR adoption of AI is near-universal, employees are feeling a little more tepid. 38% of employees say they're currently using AI to assist them in their role. But adoption is by no means linear.
Desk-based workers (48%) are more than twice as likely to adopt AI compared to non-desk workers (23%).
AI-forward industries like IT and telecoms (60%), finance (52%), and sales, media, and marketing (50%) show the highest overall levels of adoption at a sector level.
Adoption lags in frontline-heavy sectors, such as healthcare (30%) and retail and catering (32%).
44% of employees aged 25–34 use AI in their role, compared to 28% in the 55+ group.
Employees who are optimistic about AI's potential outweigh those that aren't (40% versus 27%), with younger employees broadly feeling more positive than older peers.
Employees who agree that they're excited by AI's potential to improve the way they work.
But after a huge swell last year, AI use among employees is starting to cool off. In 2025, employee AI use almost doubled from the year before, springing from 24% to 43%. In 2026, that dipped to 38%. And 25% of employees are worried about losing their job to AI.
Employees who currently use AI
24%
43%
38%
202420252026
Without support, employees are spinning their wheels
If employee AI use is trundling along slower these days, it could be down to how supported employees feel in using it.
Only 31% of employees say their employer gives them adequate AI support at work. That's down 5% from last year.
Employees actively developing their AI skills fell from 49% in 2025 to 44% in 2026.
Employees feel less confident overall in their skills lasting the distance (66% in 2025 versus 62% in 2026).
Early on, adoption of AI tooling was relatively easy. Employees dived in readily, curious about what it could do, and how it could make their work better.
But there's a big difference in using AI, and using it well at work. The rush to adopt it on an organisational scale, under the organisation's rules, got ahead of both employees' skills and the support their companies were able to give. The easy wins that came thick and fast in the first months of adoption have slowed. And without the right support at the right time, enthusiasm and progress have started to wane.
“Employees are constantly hearing their company executives say ‘use AI' without any clear guidance,” Bradburn says. “They aren't seeing impact or actual proof that it's changing something meaningful within the business.”
Emerging trend
Mixed feelings
While employees are broadly positive about AI's potential impact, the overall story on how they're feeling isn't that clear-cut. Across several sentiment questions, agreement and disagreement were often closely balanced.
AgreeNeither agree nor disagreeDisagree
I know very little about how to use AI at work
36%28%33%
My organisation encourages the use of AI to improve efficiency
35%29%33%
I would like more support in the workplace for using AI tools
37%32%28%
AI tools have made me more productive in my current role
35%29%33%
I am concerned about the workplace changing as AI replaces jobs
34%27%37%
What's making employees sit on the fence?
While this data could read as indifference or neutrality, we think it's actually the opposite.
A large share of neutral and mixed responses reflects a workforce facing an uncertain future. And while there are broadly more advocates than those actively resistant, we're still a long way from the healthy enthusiasm needed to promote experimentation and novel use cases.
“We're in a moment when people are questioning themselves,” Horne says. “There's no benchmark on how good or bad people are with AI. No one really knows what AI is going to look like in 12 months' time. There are people saying it will change everything, and others saying it won't. Employees might be sitting in the middle because they just don't know.”
Employees and HR don't see eye-to-eye on data transparency
If AI still feels like a black box to employees, they also feel like they're being left in the dark on how their organisation uses data to make decisions about their future.
20%
of employees say they “always” get a clear explanation on how HR decisions are made
55%
understand how their data feeds into the decision-making process
61%
of HR leaders say their decisions are more data-driven than two years ago
Only 20% of employees say they always get a clear explanation of how HR decisions relating to them are made, including pay, performance, and promotions. Over a quarter (26%) rarely or never do.
Many don't understand the role their data plays within that process, either: just over half (55%) of employees know how their data feeds into the decision-making process itself.
But mirror, mirror on the wall, HR still perceives a more data-driven function than them all. 61% of HR leaders agree that their processes and decisions are more data-driven than two years ago. This transparency gap could become a bigger challenge once AI is integrated into more processes involving employees.
“Employees aren't necessarily afraid of AI,” says Dominic Joyce, fractional global talent acquisition lead, Maverick Otter. “They're afraid of what the organisation might do with the tool, what it means for their role, and whether decisions affecting them will become less transparent. They don't need to understand every technical detail of an algorithm, but they deserve to understand how important decisions are being made using it.”
Emerging trend
Will employees start demanding AI transparency?
Half of employees don't understand how their data is used in workplace decisions. Yet just 10% of employees want to know how AI makes decisions about them.
As HR teams begin embedding AI across the employee lifecycle, demand for transparency and explainability may rise quickly once employees have more knowledge about where it factors into processes.
The hidden costs of data opacity
HR and employees don't see eye-to-eye on how data impacts decisions. But for employees, that ambiguity has a sizeable impact on how they show up to work.
Employees who always receive a clear explanation on how their data is used to make decisions are more than twice as likely to be highly motivated, and over 2.5 times as likely to trust HR to a “great extent”. They also report feeling more productive, and are more likely to stay.
Trust HR “to a great extent”
40%
of those who “always” get a clear explanation
15%
of those who “rarely” or “never” get a clear explanation
Report that motivation is high
64%
of those who “always” get a clear explanation
31%
of those who “rarely” or “never” get a clear explanation
Report that productivity is high
72%
of those who “always” get a clear explanation
46%
of those who “rarely” or “never” get a clear explanation
Actively job-seeking
44%
of those who “always” get a clear explanation
60%
of those who “rarely” or “never” get a clear explanation
Emerging trend
AI in the shadows
AI has outpaced both readiness and enablement. It's also outpacing organisations' ability to govern it, leading to some employees taking matters into their own hands.
But that's not necessarily a support or training issue. At least, not in the way you'd think.
23%
of employees admit to using AI without their employer being aware.
5x
Employees who say AI has improved their productivity are more than five times as likely to use it without their employer's knowledge compared to those who say it hasn't (45% versus 8%).
3x
Employees who say they receive adequate AI training are three times as likely to use AI under the radar compared to those who don't (43% versus 14%).
What's driving the capability paradox?
One possible explanation is that employees who feel confident using AI and can see how it benefits their role are proactively integrating it into their work, whether they have their organisation's blessing or not.
But if organisations fail to enable this hidden cohort, they may need to worry about retention as well as risk mitigation. Shadow AI users are far more likely to be actively looking for a new role compared to non-users (67% versus 42%).
“Employee ambivalence often comes from not seeing meaningful use cases across the company,” Major says. “The people who have found them can become frustrated when organisations don't equip them, and that's often where shadow AI use comes in. These are also the same individuals who, when equipped, accelerate AI adoption. This is the organisation's opportunity to talk to those people, understand what they're doing and enable them.”
What good AI looks like: Employee-led performance review prep
The problem: Switching to a more frequent, employee-led performance review model meant employees were struggling with review prep.
The use case: Created a Claude skill that gathers recent evidence from the HR platform and systems where work is already documented (CRMs, AI notetakers, Slack etc.)
The outcome: Employees arrive at reviews with an AI-prepared and human-refined evidence dossier, resulting in a richer conversation.
We switched to a bimonthly performance review model that's more employee-led. To do that, we needed a way of making it easier for people to focus less on the fact-finding and more on conversation itself (i.e. the most valuable part). So we built a Claude skill that helps employees prep by pulling eight of their last weeks of activity from our HR platform and a lot of other connectors like Linear, Hubspot, Notion, Slack. Claude then anchors that in their job mission and outcomes, their previous performance review, the company roadmap and any relevant skills matrices to help spot what might become important in the next few months for the business.
What to do next: Build trust and clarity into your roll-out
When employees are being told to use AI without any guidance from the organisation, or clarity over how it impacts their role, it can lead to them abandoning the tech entirely or forging ahead on their own. Neither works for long-term AI integration.
Focusing on trust, clarity, and enablement will give employees the confidence to experiment and the knowledge to do it responsibly.
1
Create safe conditions for experimentation
Reaching AI maturity is as much about employee behaviour as it is about technical integration.
“You can't jump from employees having no exposure to AI to giving them a fully integrated system,” says Ankita Poddar, senior HRBP at Amazon Web Services. “Skipping that step reduces trust. You trust a system a little less when you don't know what it's doing or how it's doing it. Employees need to play around with AI, fail, and figure out how it works for them, their team and their organisation.”
Give employees clear guidelines on where they can and can't use AI, including what proprietary or sensitive data cannot be used in AI tools. Then, create safe places for experimentation, including hackathons, sandbox environments, and dummy data sets so employees can test and learn responsibly.
2
Turn hidden use into shared capability
When the trust and safe guidelines are in place, organisations need to create ways for employees to feel comfortable sharing use cases openly.
“AI is optimised for individual use, rather than at a team level,” says Aoife O'Brien, founder of Happier at Work. “People may hide how much they're using it because they fear that working more efficiently will make their role look less valuable. Shifting the focus to outcomes and improving coordination across teams will help organisations get more effective results from AI.”
O'Brien recommends organisations take an experimental approach to creating this environment to create mechanisms for knowledge-sharing that both match their existing communication norms and keep psychological safety high. Her suggestions include:
Creating a repeatable structure around learning, including immersion days, dedicated Slack or Teams channels, showcases in team or all-hands meetings, and building a centre of excellence for AI.
Holding retrospectives on all AI projects, especially those that failed.
Framing AI projects in terms of outcomes and the problems they're solving for the business.
Building a centralised repository of prompts, projects, outcomes, and learnings that everyone can access and refer to.
3
Give employees clarity on how data and AI shape decisions
If you expect transparency on how employees use AI, you need to offer the same in return. HR teams need to be able to explain how data factors into decisions made about them, and how or where AI has had input, if at all. This needs to cascade down at a manager level, too.
For every key decision, you need to know your data and your model well enough to answer:
What data or AI did we use to inform this decision?
Where does human judgement shape this decision/outcome?
How can employees question or challenge this decision if they don't agree?
4
Measure the impact, not the output
Organisations may be all-in on improving efficiency, but if that's all employees see, then using AI could start to feel transactional — especially if AI-related optimisation simply means their workload increases or intensifies.
This means looking beyond how much faster work gets done, or what costs are saved. HR leaders need to ask: are employees finding work more meaningful? How is overall employee wellbeing? How do employees feel about their workload over time?
“HR leaders ultimately care about the service and experience they deliver,” Major says. “Efficiency shouldn't be the only metric you measure. If you focus on it too narrowly, you don't see whether things are actually improving as a result of the changes you've made. Combine efficiency with experience, and you get a much more holistic picture of what's going on.”
What good AI looks like: Leading better one-on-ones
The problem: Strong managers spend a disproportionate amount of time on admin to prepare for high-quality 1:1s. When preparation is rushed (or skipped), bias creeps in, patterns are missed, and valuable context is lost.
The use case: Embed AI into existing pre-1:1 and performance prep to reduce admin, so the manager can focus on personalised coaching conversations.
The outcome: Time saved is reinvested into listening, coaching, and showing up with a consistent, high-quality, and personalised approach for each individual.
One use case really stood out to me: a manager who built his own AI solution to reduce the admin required for unbiased, data-informed, and personalised 1:1s — so he could focus on the conversation itself.
During every 1:1, he uses AI to transcribe the conversation. But he doesn’t treat the transcript as the notes. Instead, he takes his own notes alongside it: keywords, phrases, and moments that stand out. He then feeds those highlights back into the AI as a layer of his own judgement. Combined with the full transcript, this produces a far more meaningful summary because the AI knows what to look for.
Data-ready HR teams span different budgets, sizes, sectors, and levels of technical sophistication. But the thing that unites them in embedding AI successfully is consistency in the fundamentals.
What data-ready organisations do differently
Document most decisions
78%
Data-ready organisations
69%
Data-deficient organisations
Regularly review people predictions
64%
Data-ready organisations
28%
Data-deficient organisations
Embed AI across the organisation
38%
Data-ready organisations
14%
Data-deficient organisations
These seemingly small actions reinforce one another. Documenting your decisions creates a record of why choices were made. That breadcrumb trail makes it easier to test and review your people predictions, and course-correct the errors. And those checkpoints for quality control build trust in the data, meaning you can embed AI more confidently across people processes.
And when teams put those checkpoints in place, they feel more confident about the value of the work they deliver.
Data-ready HR teams are more likely to feel that they are more effective, valued, and heard. They're more likely to say employee engagement is high. And they're 1.4 times as likely to report high business performance.
How data-ready organisations see themselves
Report high HR effectiveness
79%
Data-ready organisations
50%
Data-deficient organisations
Feel trusted and valued
79%
Data-ready organisations
58%
Data-deficient organisations
Feel HR doesn't have enough of a voice
29%
Data-ready organisations
52%
Data-deficient organisations
Report high business performance
82%
Data-ready organisations
57%
Data-deficient organisations
For HR teams looking to close that gap, you're probably further along than you think. But the path to becoming data-ready is forged in the tiny things you do day to day. It's curating and maintaining the handful of datasets you can trust over the ten you can't. It's creating the workflows and gates that keep that data quality high over time, and making it a KPI for every new decision or project.
It's turning data into an everyday instinct.
You already have the data. What you're building now is the confidence to stand behind it.
Get your HR data in order for 2027.
Join our webinar to hear from HR leaders on improving data quality, preparing for responsible AI, and keeping employees engaged through change.
Our 2026 Workforce Pulse report features survey responses from 3,001 HR decision-makers and 6,000 employees at companies ranging from 10 to 2,000 workers across the UK, Germany, Spain, and the Netherlands. The data was collected between May 29th and June 10th by Censuswide.
