People Analytics Trends 2026: What the Evidence Shows

Short answer

The clearest people analytics developments in 2026 are not a race to buy more software. US and UK evidence points to AI moving from a technology question to a workforce-change question: teams need to plan skills, redesign work deliberately, and test where current data and human review are sufficient. For HR leaders, the practical response is to test one decision, its data, its human review, and its limits before expanding a workflow.

1. UK evidence points to work and skills redesign

CIPD's April 2026 UK research surveyed 1,342 senior leaders and people professionals in January and February. It found a gap between confidence using AI tools and confidence in workforce planning, reskilling, and job redesign as AI changes tasks and workflows. This is a survey of its UK sample, not a forecast of every employer.

What this may mean: the 2026 question is less "Do we have AI?" and more "Which work and skills change should we prepare for?"

Test it: choose one task, such as preparing a manager conversation after a shift-handover change. Define the required context, the reviewer, a correction route, and a decision that could follow. If the team cannot define those elements, a wider rollout has no testable job yet.

2. AI skills planning is being treated as staged change

CIPD's April 2026 UK guide frames AI adoption as early-stage, intermediate, and advanced work, with a sequence to align, audit, analyse gaps, act, and adjust. The guide is practical guidance, not evidence that every organisation will follow the same path.

What this may mean: a dashboard or generated summary is only as useful as the work question, definitions, coverage, timing, and context behind it.

Test it: take one current workforce measure and write its definition, source, coverage, refresh date, known gap, and decision owner. Then ask a focused work question where the record cannot explain the context. Keep the account distinct from the measure and record what would confirm or challenge the first interpretation.

Fictional example

A fictional UK service firm sees higher internal transfers from one region. Its records show dates and roles but not why employees moved. The people lead asks recent movers where the handover process created avoidable friction, then checks rota and training records before deciding whether to test a change. The conversation supplies context; it does not establish the cause of transfers.

3. AI adoption remains uneven across HR functions

SHRM's 2026 executive summary reports a December 2025 survey of 1,908 HR professionals, managers, and leaders. In that sample, 39% reported AI already in their HR function and a further 7% intended to introduce it in 2026. Recruiting was the most frequently reported HR use area. Plans are not confirmed deployments, and these figures concern AI broadly, not demand for a particular people-analytics product.

What this may mean: the existence of AI elsewhere in a business does not establish readiness for employee feedback analysis. This is our operational inference, not a measured outcome of the survey.

Test it: inventory the HR workflows that already use AI. For the proposed feedback workflow, identify its purpose, data, reviewer, permitted decision, and correction route. Test the missing step before buying a wider system.

What is observation, and what is hypothesis?

The sources combine a UK survey, a UK practice guide, and research from US-based SHRM. Their populations and purposes differ; they are not a combined US/UK adoption estimate. They do not prove that a particular product will improve retention, that every employer needs the same AI capability, or that current employee input represents a whole workforce. Those are local hypotheses to test.

Use this short review before acting:

• What decision would this practice improve?
• Which population and data definition does it cover?
• What work context is missing from the record?
• Who can interpret the evidence and own the action?
• When will the team review the action and its limitations?

If the missing input is current employee context, explore Lontra’s employee conversations and manager briefs. Start with a focused question and judge the output by whether it helps a responsible person prepare a real discussion, not by a claim of universal accuracy.

Sources

• SHRM: State of AI in HR 2026, executive summary, pages 2–3: December 2025 sample and reported adoption. The summary does not establish a representative estimate for every US or UK employer.
• CIPD: AI skills planning, UK practical guidance on staged adoption and skills planning.
• CIPD: Futureproofing your skills, based on a January-February 2026 UK survey of 1,342 leaders and people professionals.