The Survey Says Everything Is Fine. That May Be the Problem.

Amy Edmondson encountered a result early in her research that appeared to make no sense. Well-led hospital teams seemed to report more errors than less effective teams. The obvious conclusion was that the better teams were somehow making more mistakes.

The more important explanation was that they were more willing to report and discuss the mistakes that were already occurring. In teams where speaking up felt dangerous, the problems did not disappear, they became harder to see.

That finding should concern any executive relying too heavily on employee surveys to determine whether the organization is healthy.

What psychological safety actually means

Psychological safety is not the absence of standards, accountability, disagreement, or discomfort.

It describes an environment in which people believe they can take interpersonal risks; such as asking a question, admitting a mistake, challenging an assumption, requesting help, or raising a concern without being punished or humiliated for doing so.

It does not guarantee that every concern is correct, but it does make the concerns discussable.

That is essential in environments where leaders depend on employees to surface weak signals before they become significant failures.

The survey problem

Consider the structure of a common engagement survey. The organization asks:

• Do you feel safe raising concerns?
• Does your manager respond well to feedback?
• Can you challenge decisions without negative consequences?

For employees in a high-trust environment, these may be straightforward questions.

For employees in a low-trust environment, answering honestly may feel like performing the exact behavior they believe is unsafe. The organization is asking someone to report that they do not feel safe reporting. That creates a measurement problem.

The people most willing to respond candidly may be the people who already experience relatively high psychological safety.

The people with the strongest concerns may soften their responses, avoid written comments, or decide that participation creates more risk than value.

A favorable survey result can therefore mean several different things:

• Employees genuinely feel safe
• Employees do not trust the confidentiality process
• Employees believe nothing will change
• Employees have learned which answers create the least attention

The score alone cannot tell leadership which explanation is correct.

What behavior reveals

Psychological safety becomes visible when someone tests it.

• What happens when an employee disagrees with a senior executive?
• How does a leader respond to an unexpected problem?
• What happens to the manager who challenges an aggressive deadline?
• Does the person who admits a mistake receive support in correcting it or become permanently associated with the mistake?
• Does the organization reward early warning, or only successful recovery?

These moments establish the actual operating rule even thought the executive may not intend to punish candor.

A dismissive response, public correction, defensive explanation, or subtle withdrawal of trust can still teach the entire room that speaking up has a cost. The lesson spreads quickly because everyone is watching what happens to the first person who tests the stated value.

Microsoft Research: participation depends on conditions

Microsoft researchers examining participation in virtual meetings found that psychological safety affected how lower-status employees participated in technology-mediated conversations.

In the study, employees lower in organizational status who perceived greater psychological safety contributed more through meeting chat than comparable employees who perceived less safety. The research illustrates that the communication platform alone does not create participation; the surrounding interpersonal conditions shape whether people use the available channels.

Giving employees another way to speak does not guarantee that they believe speaking is safe.

That principle applies well beyond meetings.

Where Viva Pulse helps

Microsoft Viva Pulse offers a more agile alternative to relying solely on a large annual survey.

Organizations can use recurring, research-backed pulse surveys and configure minimum recipient and response thresholds. Results are displayed after the pulse closes, and authors generally do not see employee names or identifiers unless respondents include identifying information themselves. Microsoft notes that administrators or other privileged users may be able to identify responses in certain circumstances, which is why accurate communication about privacy matters.

The shorter cadence can help leaders detect changing sentiment earlier. It can also make feedback feel less ceremonial and more connected to current work. But a better feedback tool does not automatically produce a safer culture.

A six-person team may still worry that specific comments are identifiable. Employees may still doubt that confidentiality protects them from contextual inference. Most importantly, people will watch what leaders do with the results.

If leaders ask for feedback and then explain away every uncomfortable response, the tool teaches employees that participation has little value.

If leaders react defensively, the tool may lower psychological safety rather than raise it.

Measurement is not the intervention

A pulse survey can reveal where leaders should pay attention. It cannot substitute for how leaders behave when difficult information arrives.

Psychological safety is built through repeated leadership responses:

• Thanking someone for raising an issue
• Asking questions before defending a decision
• Separating the person from the problem
• Acknowledging uncertainty
• Correcting the issue without punishing the messenger
• Following up visibly enough that employees know speaking up mattered
• The stated value is established in the town hall

The theory-in-use is established in the response.

The disagreement may be the signal

Suppose a survey reports strong trust in leadership, but projects regularly remain “green” until shortly before they fail.

Suppose employees report comfort with speaking up, but meetings contain little disagreement.

Suppose leaders describe an open culture, but the same individuals deliver every difficult message.

The correct response is not necessarily to reject the survey. It is to examine the gap between the survey and the operating pattern.

One may reflect what people believe. The other may reflect what they have learned to do.

Both matter.

The Aureus perspective

Surveys, pulse tools, interviews, and workplace analytics can all contribute useful information.

None should be mistaken for the organization itself.

The purpose of assessment is not to collect the largest possible volume of data. It is to understand the conditions affecting leadership alignment, decision-making, transformation readiness, and performance.

When the data says everything is fine but execution feels fragile, leadership should not automatically choose between the two stories.

The disagreement may be the most important information available.

Do not wait for hidden problems to become visible through missed outcomes, failed implementations, or lost talent.

Schedule a Strategic Conversation.

Sources and further reading

Amy Edmondson, “Psychological Safety and Learning Behavior in Work Teams.”

Harvard Business School, “Are Great Teams Less Productive?”

Microsoft Research, “Virtual Voices: Exploring Individual Differences in Written and Verbal Participation in Meetings.”

Microsoft, Viva Pulse Privacy and Data Protection.