Ethical governance: preserving employee trust and privacy during relational analytics

Relational analytics keeps employee trust only when people agree to take part, results are reported in groups rather than about individuals, the data is used only for the purpose it was collected for, and the rules are written down before any analysis begins.

Network data describes relationships between people, which makes it more sensitive than most HR data. Handled carelessly, it can feel like surveillance and damage the trust it was meant to strengthen. This guide sets out a governance approach that BlissPoint follows and recommends.

A team leans over printed charts and graphs on a meeting table
Photo: Yan Krukau on Pexels

Why relationship data needs extra care

A network map can reveal who is isolated, who people avoid and who holds informal power. Even in aggregate, small groups can be identifiable. And relationship data about one person is also data about the people they name. These features call for stronger safeguards than ordinary reporting.

Six governance principles

  1. Purpose first. Write down the business question and the decisions the analysis will inform. Use the data for nothing else.
  2. Informed, voluntary participation. Explain what is collected, who sees it and how long it is kept. Make taking part optional, with no consequences for declining.
  3. Aggregation. Report by team or role, with a minimum group size so no individual can be singled out.
  4. Minimization. Collect only the relationships needed for the question. Avoid content of messages entirely.
  5. No individual scoring. Do not use network position in performance reviews, pay or promotion decisions.
  6. Retention and deletion. Set a date to delete raw responses and keep only the aggregated findings.

Privacy frameworks such as the NIST Privacy Framework offer a structured way to identify and manage these risks, and local data protection and labor laws may impose specific requirements; involve legal counsel and, where they exist, employee representatives.

Who should be involved

Roles in an ethical network analysis
RoleResponsibility
SponsorOwns the question and commits in writing to the use limits
Analyst or consultantDesigns the survey, holds raw data, applies aggregation rules
Privacy or legalReviews the plan against law and policy
Employee representativesReview the communication and the use limits where they exist
ParticipantsReceive a summary of findings and what will change

A note on passive digital data

Some tools build networks from email, chat or calendar metadata without asking anyone. Even without message content, this data can be intrusive, and employees may not know it is collected. BlissPoint's approach favors consent-based surveys and treats any passive collection as requiring an explicit purpose, transparency to employees, legal review and strict aggregation. See active versus passive ONA.

How this applies at BlissPoint

BlissPoint's network work is survey-based and reported in aggregate. Organization Viewer maps the formal structure: the org chart, reporting lines, spans and layers and succession readiness. It does not read email, chat or calendar data and does not run network analysis. Its manager and 360 surveys show peer and direct-report results only for groups of five or more. The same group-minimum rule applies across its people reporting, such as the turnover dashboard, which groups small teams so no individual is exposed.

What to tell participants

  • Why the analysis is being done and which decisions it will inform.
  • Exactly what is collected, and what is not, such as message content.
  • Who will see raw data, who will see aggregated results, and the minimum group size.
  • That taking part is voluntary and that declining has no consequences.
  • How long raw data will be kept and when it will be deleted.
  • How participants will hear about findings and resulting changes.

Sharing a summary of findings with participants afterwards is one of the strongest ways to build trust for future surveys.

Key takeaways

  • Relationship data is about more than one person at a time, so it needs stronger safeguards than ordinary HR data.
  • Purpose, consent, aggregation, minimization, no individual scoring and deletion dates are the core safeguards.
  • Write the rules down and share them before collecting anything.
  • Tell participants what was found and what will change; it protects future participation.

Questions

Should employees be told about network analysis?

Yes. Explain what is collected, why, who sees the results and how long data is kept, and make participation voluntary.

What is a safe minimum group size?

There is no universal number. Many survey programs use five or more; choose a minimum that prevents identification given your team sizes.

Can network data be used in performance reviews?

Responsible practice says no. Using network position to rate people undermines trust and the honesty of future responses.

Does Organization Viewer collect communication metadata?

No. It does not read email, chat or calendar data.

Who should hold the raw network data?

A small, named group, often the analyst or an outside consultant, under written rules. Leaders should see aggregated results, not raw responses.

Is anonymized network data safe to share widely?

Not automatically. Network structure can reveal identities in small groups, so share only aggregated findings above a minimum group size.

Sources

← All relational science and network analysis