Overview
.svg){kind=small}
One AI
Data Mesh
.svg){kind=icon}
One AI Assistant
Data Stories
Enterprise AI
Employee Engagement
.svg){kind=small}
Talent Retention
-1.svg){kind=small}
HR Dashboards
Succession Planning
.svg){kind=small}
Talent Acquisition
Workforce Analytics
Why One Model
-2.svg){kind=small}
Whitepapers
Blog
-2.svg){kind=icon}
Events & Webinars
.svg){kind=icon}
Product Innovation Blog
-1.svg){kind=small}
Roles in People Analytics
.svg){kind=small}
Contact Us
-3.svg){kind=small}
Partners
About US
-3.svg){kind=icon}
Careers
-1.svg){kind=icon}
News
This blog was inspired, with permission, by Alexis Fink’s think piece on Substack.
As AI takes center age in all facets of business, we see rampant speculation about the fate of jobs–even entire industries. Questions like “Will AI replace me?” and “Is People Analytics dead?” simmer in the heads of analysts.
The answer to both: no. In this article, we’ll unpack why AI won’t replace people analytics teams.
The work will look different, yes. Initiatives that once took months will shrink to days. Menial tasks will be automated away, as they should be. Natural language queries will make it easier for leaders to access and query data (call it “citizen data science,” if you will). The impact of AI on HR analytics can’t be overstated. But that burgeoning efficiency doesn’t signal the death of People Analytics or the careers of the humans who oversee it. Here’s why.
1. Confidentiality Is Key, And It Requires Human Oversight
In a People Analytics context, confidentiality ensures the integrity and reliability of your data. Consider this: Survey responses furnish some of your most straightforward insights. The strongest predictor of attrition, for example, is simply the response to an “intent to stay” question. But if people doubt the anonymity of their responses, they won’t answer honestly, and the data loses its integrity.
The People Analytics team protects against that. They’re the presence that assures privacy. As workforce data expert Alexis Fink put it, “A People Analytics team is a crucial ethical moat that protects employees, while still helping leaders make informed decisions based on aggregated, privacy-protected insights.” If employees believe their private responses rest with an unmanned AI, it erodes trust.
This stretches beyond surveys. Performance review data, pay history, and sentiment data are all tightly held for the same reason, and they require human stewardship. A dedicated People Analytics team ensures that AI acts as a tool to process this sensitive information, not an unmonitored gateway that exposes it.
2. Asking the Right Questions Takes Human Judgment and Context
It’s easy to underestimate how difficult it is to scope and build a research question. Even harder to know what data is relevant and available, and which statistical technique to use. Even basic metrics like counts and rates can be misleading. Small choices, such as which denominator to use, can significantly affect the conclusions you draw. Without the team, who makes the call and defends it?
Consider this example from Alexis Fink:
“Let’s look at a seemingly simple question about cancer rates: Are they rising, flat or falling?
Yes. To all.
It depends on how you ask the question.
Obviously, they are rising. Cases have been rising for many years! If you simply count all cases, this is true.
Obviously, they are flat. The global population is rising, so you’d expect the occurrence of pretty much any human experience to also be increasing. So, we obviously need to divide the total number of cases by the global population. And cancer is pretty flat as a cause of death.
Obviously, they are falling. Cancer is also more common in older people, and life expectancies have been increasing. So, if you normalize the data by age, you can see very plainly that cancer as a cause of death has been falling.”
In this example, “correct” is relative. It depends on how you choose to approach the question, and the business logic of the entity asking it.
- If you’re a public health organization, you’re likely focused on the 3rd approach, cancer rates normalized by age.
- If, however, you’re a company that sells chemotherapy drugs, you are much more interested in the total rates.
None of the answers are wrong – the challenge is figuring out which frame is appropriate. Neither AI nor citizen data scientists are fully equipped to make those judgments without prior PA intervention. The data can also be downright contradictory in a way that prevents conclusive findings. This is why, as Fink says, "[Analyzing workforce data] responsibly and effectively requires more than an LLM license."
For example:
A company introduces a hybrid work policy to combat turnover. Throughout the year, the strategy seems like a massive success: when broken down by department, every single function successfully reduces its attrition rate. Individual leaders are celebrated.
The Twist: At the end of the year, the company-wide aggregate data reveals that overall corporate attrition actually increased.
The Paradox: While every leader successfully lowered their team's turnover rate, a massive corporate headcount shift overrode their progress. To meet new demands, the company doubled the size of its Customer Support team (which historically has a high 40% base turnover rate) while downsizing its highly stable Engineering team (which sits at a 10% baseline). So, even though Support lowered its attrition to 35% and Engineering dropped to 5%, expanding the high-turnover department dragged the overall corporate average upward.
The Conclusion: Both narratives are true: every individual leader improved retention, yet the company overall lost a higher percentage of its workforce. Without human business context to spot this "mix shift," an analyst looking only at aggregate data would wrongly declare the hybrid policy a failure.
3. Without the PA Team, There’s No “Shared Organizational Reality”
One of the important jobs of a people analytics team is carefully crafting a “shared reality” where organizational decisionmaking takes root. When disparate stakeholders have access to the data, or the reality hasn't been conveyed to the AI, that becomes more difficult.
When left to their own devices to explore data, citizen data scientists:
- Take time away from their own job
- Have no guarantee that they’re using the correct data for their analyses
- Have their own perception and biases that dictate the analysis they land on
Without even trying to, they’re each operating in a bubble universe with its own subjective reality. It’s like the “blind men and the elephant” problem: Three blind men feel a different portion of an elephant, from its trunks to its legs to its body. Each man arrives at a different conclusion about what object he’s touching (a snake, a tree trunk, etc), but none are right. Their perception is informed by the information they had access to, but lacks full-picture context.
Different versions of reality lead to clashing opinions. Energy that should be directed at building and scaling solutions instead gets squandered in quarrels over defining the problem. That’s not to say that there isn’t a place for citizen data science in People Analytics. AI will be a tool that helps PA expand their universe, democratize insights, and be more effective. But vibe coding and natural language based queries won’t be enough.
That said, the shared organizational reality that analysts bring can and should be codified in a semantic layer that applies agreed-upon business definitions consistently across reports, dashboards, and AI-powered analyses.
In Conclusion
People Analytics will continue to evolve alongside AI, but the expertise behind the function is becoming more valuable, not less. As workforce data becomes easier to access and analyze, organizations need people who can define the right questions, establish trusted business logic, protect employee privacy, and create a shared understanding of what the data actually means.
The challenge is ensuring that expertise can scale beyond a small group of analysts. AI can help democratize access to workforce insights, but it can’t automatically inherit years of institutional knowledge, analytical judgment, and business context. Those decisions must be made explicit and embedded into the systems that power reporting, analytics, and AI.
This is where platforms like One Model play a critical role. By providing a governed semantic layer, secure access controls, and an AI-ready people data foundation, One Model helps organizations operationalize the expertise of their People Analytics teams. The result is a workforce intelligence ecosystem where employees, leaders, and AI assistants can explore data confidently while remaining aligned to the same trusted definitions and business logic.
As we discussed in our recent article on why AI stacks can't rely on tribal knowledge, successful AI initiatives depend on more than access to data. They depend on making the judgment, context, and governance that live inside expert teams available at scale. Organizations that invest in that foundation will be better positioned to use AI responsibly, make faster decisions, and maintain trust in the insights that guide them.
We’d love to discuss your challenges.
Connect with us today.
