From Harvard Business Review · Max Blumberg · Nov 5, 2026 · 4m

HR needs R&D, not just data

Most companies collect vast employee data but rarely invest in research to answer the questions that matter most. HR should work like product development.

In brief

Most companies collect vast employee data but rarely invest in research to answer the questions that matter most. HR should work like product development. Companies should treat human capability analytics as research and development, investing in new evidence to answer consequential business questions rather than…

#human resources #people analytics #research and development #workforce data #business strategy #employee performance #organizational culture #data investment #rentokil initial #google project oxygen #business #management

HR collects data but lacks insight

Most companies gather enormous employee information from compensation to engagement scores, yet collect it through routine processes rather than to answer specific business questions.

The streetlight effect in HR

Companies limit analysis to information they already have, even when it cannot adequately answer the question at hand.

Treat HR analytics like R&D

Companies should identify a consequential business question, determine what evidence would help answer it, and invest in generating that evidence when the potential payoff justifies the cost.

What existing data cannot reveal

HR systems can show how many salespeople a company employs and how they perform, but may reveal little about underlying attributes that distinguish exceptional performers from average ones.

Google's Project Oxygen

In two thousand nine, Google asked: Do managers matter?Existing data suggested yes, but to understand what effective managers did differently, Google conducted double-blind interviews with highly and poorly rated managers and identified eight associated behaviors.

Google's research paid off

Google trained managers on those eight behaviors and developed twice-yearly surveys to assess them. Between two thousand ten and two thousand twelve, median favorable scores on the surveys rose from eighty-three percent to eighty-eight percent.

Why companies avoid new research

The costs of collecting new people data are immediate and visible, borne by HR.Potential returns take years to materialize and show up elsewhere in the business through increased sales, retention, or profitability.

Analytics teams lack research skills

Human capability analytics teams are generally better equipped to analyze existing information than to design new measures, conduct original research, or evaluate economic returns. They gravitate toward what they know how to do.

Organizational barriers to research

Employees may be wary of additional monitoring, and managers may resist research that calls their decisions into question.

Criterion one: financial consequences

The greater the cost of making the wrong decision, the more a company should invest in getting it right.

Criterion two: existing data gaps

Before commissioning new research, companies should establish what they already know and what they do not.If current records can provide a reliable answer, there is little reason to pay for new evidence.

Criterion three: actionable findings

Research has little business value if its findings cannot inform a decision.Before collecting anything, leaders should identify the actions they might take depending on the results.

Criterion four: responsible collection

New measures must be sufficiently valid to support the decisions they inform.Companies must consider employee privacy, legal requirements, and potential consequences of collecting sensitive information.Not every useful question can be answered responsibly or economically.

Calculating research value

A company considering a five hundred thousand dollar investment in a development program might commission a forty thousand dollar study with a one-in-five chance of revealing ineffectiveness.

Manage research as a portfolio

Each year, leaders should identify important questions existing data cannot answer, estimate their potential value, and allocate funding accordingly.For each project, record what was spent, what research revealed, and how findings influenced decisions.

Fund research by beneficiary

Funding should reflect who stands to benefit.If a study improves operations, that function should contribute to its cost.If it addresses customer share, marketing should help fund it.

Stage funding and protect privacy

Companies should fund research in stages, beginning with a pilot to establish whether proposed measures are reliable.Agree in advance on criteria for continuing, changing course, or stopping.

Rentokil Initial's sales research

Pest control company Rentokil Initial wanted to improve salesforce performance.Existing sales records showed who performed well but not why.

Rentokil's results and shift

Rentokil tested six validated selection instruments with approximately two hundred seventy salespeople in the UK and US from a global salesforce of about seven hundred.One personality assessment had a stronger relationship with sales performance.

In short

Companies should treat human capability analytics as research and development, investing in new evidence to answer consequential business questions rather than relying only on existing data.

Read the original on Harvard Business Review

How we make summaries