The first wave of HR technology gave people operations teams better ways to collect and display what employees already knew they felt. Engagement platforms multiplied, survey cadences shortened from annual to quarterly to pulse-weekly, dashboards became more visual. The underlying logic remained the same: ask people how they feel, aggregate the answers, present the results to managers.
That approach has real value. Sentiment data — collected carefully, acted on honestly — is a legitimate input into management decisions. But it is structurally limited by the same constraint that has always defined it: it is retrospective, aggregated, and self-reported. It tells you what teams felt in a recent period. It does not tell you what conditions are producing those feelings right now, or which conditions are trending toward a problem that has not yet registered as sentiment.
The second wave is different in kind, not just in degree. It applies behavioural signal detection to operational data — the data that organisations already generate continuously from their tooling stack — to surface the preconditions for people outcomes before those outcomes crystallise into survey responses.
What operational data actually contains
Every organisation running on a modern collaboration and productivity stack is generating a continuous stream of behavioural signals. Calendar data records when meetings occur, who attends, how meetings cluster or fragment across a work week, and how synchronous time is distributed across teams. Collaboration platform data captures response latency, message volume patterns across channels, which individuals are communication hubs versus isolated nodes, and how information flows between teams. HRIS data tracks leave patterns, tenure distribution, and internal mobility.
Individually, none of these data streams was designed for people analytics. Collectively, they contain more behavioural signal than any survey can capture — because they reflect what people actually do rather than what they report about what they feel.
The shift from sentiment aggregation to operational signal detection is a shift from asking "how do your teams feel?" to "what are your teams' calendars, communication patterns, and collaboration graphs actually showing, and what does that predict about team health outcomes in the next four to six weeks?"
Why this changes the job of a people ops leader
Consider the difference in practice. A people ops leader in the survey paradigm works primarily on two cadences: survey design and deployment, and intervention planning based on results. The core skill set is survey methodology, qualitative analysis of open-text responses, and stakeholder management to convert findings into manager action. The time horizon is quarterly or biannual — because that is how often meaningful new data arrives.
In the operational signal paradigm, the time horizon compresses significantly. If meeting load fragmentation data is updated continuously and displayed against engagement prediction models, the people ops leader knows — in near real time — which teams are trending toward conditions associated with disengagement, and which specific signal is driving that trend. The intervention conversation happens weeks earlier and is considerably more specific: "Engineering team 3 has had zero 90-minute focus blocks for two consecutive weeks, which is correlated with the context-switch spike we saw last month — we should talk to that team's manager before the next sprint starts" rather than "Q3 survey shows Engineering at -4 points in effectiveness dimension; recommended action: manager training on workload prioritisation."
The second conversation is harder to ignore, because it is specific, current, and predictive rather than historical and vague. But it also requires more analytical capability from the people ops leader — not just reading a survey dashboard, but understanding what the operational signals mean in the context of that team's structure and workload.
The limits of this shift — and where it can go wrong
We want to be direct about what operational signal detection cannot do, because the field's failure mode is overpromising the precision of behavioural prediction and underweighting the irreducible complexity of human experience at work.
Operational signals are structural indicators. They tell you about the conditions in which people are working — how much synchronous overhead they are carrying, how fragmented their attention is, how smooth the handoffs are between their team and adjacent teams. They do not tell you how any individual is experiencing those conditions. Two engineers with identical meeting load and context-switch scores can have completely different responses to those conditions based on their experience level, career goals, relationship with their manager, or personal circumstances outside of work.
This means that operational signals are inputs to judgment, not replacements for it. The people ops leader who receives a signal that a team's meeting load index has crossed the alert threshold and immediately sends an automated intervention message to the team's manager has made a mistake — not because the signal is wrong, but because they have confused a structural indicator with an individual diagnosis. The appropriate response is to open a conversation with the manager: "We are seeing this pattern — do you have context that helps explain it? Is there something we should be addressing in how this team's work is structured?"
Operational analytics at its best increases the quality and timeliness of those conversations. It does not automate the conversations themselves.
The GDPR dimension for European people ops teams
For organisations operating under GDPR — which covers all European companies and any company employing EU residents — the deployment of operational analytics on employee data carries specific obligations that are worth addressing directly, because this is an area where vendor claims often outrun legal reality.
Processing employee calendar and collaboration metadata for HR analytics purposes qualifies as processing personal data under GDPR Article 4. Legitimate legal bases typically include legitimate interests under Article 6(1)(f), which requires a documented balancing test, or contractual necessity, depending on the specific use case. Anonymisation and aggregation — ensuring that individual-level data is not surfaced to managers or accessible in ways that could enable individual profiling — is both a privacy-by-design best practice and, in most implementations, a necessary condition for the legitimate interests basis to hold.
Growing organisations considering operational analytics deployments should ensure they have a data processing agreement with any analytics vendor, have conducted a Data Protection Impact Assessment for the specific processing activities, and have made appropriate disclosure to employees about what data is processed and for what purpose. This is not a barrier to operational analytics — it is the proper framing for how it should be deployed. People analytics that is built on clear consent and transparent purpose has significantly more organisational legitimacy, and therefore more managerial uptake, than analytics deployed in a privacy-opaque way.
What the next generation of people ops looks like in practice
The people ops leaders who are ahead of this shift are not primarily data scientists. They are analytical practitioners who have developed a working literacy in what operational signals mean — which patterns correlate with which downstream outcomes, what the appropriate confidence interval is on any given signal, and when to escalate from signal to conversation versus when to monitor and wait.
They are also developing a new kind of relationship with operational managers. In the survey paradigm, HR and people ops often had a somewhat adversarial relationship with managers around engagement data — HR surfaced scores, managers disputed them or felt blamed by them. In the operational signal paradigm, the relationship can be more collaborative: HR is providing managers with earlier, more specific information about structural conditions affecting their teams, which most managers genuinely want and find useful.
The transition is not complete or universal. Most organisations are still primarily running on the first-wave model. But the analytical toolkit is now mature enough — and the gap between what survey data can tell you and what operational data can tell you is wide enough — that people ops leaders who build operational signal capability now are building a meaningful advantage in how quickly and precisely they can respond to the people conditions that determine organisational performance.
