Survey feedback method — Organizational Development

The survey feedback method is one of the two founding OD interventions, alongside action-research. Developed at the University of Michigan's Survey Research Center by Rensis Likert and Floyd Mann in the late 1940s.

The finding that created the method: collecting survey data changes nothing; feeding it back and working with it does. Mann's studies showed that units where managers discussed the results with their teams improved, while units that received the same report without discussion did not.

Evidence: Good. One of the few OD interventions with early controlled evidence. The Michigan studies found that units where managers discussed results with their own teams improved while units receiving the same report without discussion did not, and later meta-analyses of OD interventions placed survey feedback among the more effective approaches. The corollary is equally well attested: collecting data without feeding it back produces no improvement and measurably reduces trust.

Use when: you have organizational data and want it to produce change rather than a report.

Do not use when: you cannot commit to feeding data back to the people it came from. Then do not collect it.

The process

Involve leadership in design, so that they are committed to the questions and cannot later disown the answers.
Collect data from all members of the unit, anonymously.
Analyze and prepare feedback at each level.
Feed back down the hierarchy, in "family groups" — each manager with their own team — starting at the top. Each level sees its own data plus the aggregate above.
Each group interprets its own data, decides what it means, and identifies what it will change. The group owns the interpretation, not the analyst.
Each group commits to actions within its own control, and escalates only what genuinely requires a higher level.
Follow up: re-measure, review commitments publicly.

Why the cascade matters

Feeding back top-down through family groups does three things:

Senior leaders confront their own data first, which is the credibility test.
Each manager must discuss uncomfortable results with their own team, which is where behavior actually changes.
Escalation is filtered: teams fix what they can, and only genuine organization-level constraints travel upward.

The common modern deviation — HR publishes a company-wide deck, teams see a number, nothing is discussed locally — reproduces exactly the control condition that Mann showed does not work.

The rules that make it work

Feed back to the people who gave the data, first. Before the board, before the executive summary. This is an OD value and a practical necessity.
Give each group its own data, not just the organizational average.
The group interprets. An analyst who arrives with conclusions has taken the ownership that produces the change.
Actions must be within the group's control. A team asked to action organization-wide pay dissatisfaction will disengage.
Close the loop visibly. "You said, we did" — including "you said, and we are not doing it, and here is why." The honest refusal preserves more trust than silence.

The damage of doing it badly

Surveying without feedback measurably reduces trust below the pre-survey baseline. Participants have been asked to disclose, have taken a small risk, and have received nothing. The second survey then has a lower response rate and more guarded answers, and the organization concludes that people do not engage with surveys.

An organization that has done this three times has taught its workforce that disclosure is pointless. Recovering from that costs more than the original program.

Examples by audience

Health care: national staff surveys produce enormous datasets that frequently stop at board level. The wards that improve are those where a manager sat down with their own ward's results and asked "what do we think this means, and what can we change ourselves?"
Software: developer experience surveys work the same way. Give each team its own results next to the organizational distribution, and let the team choose one thing to fix. See space-framework.
Executive: the executive team should see and discuss its own 360 data first, in the room, before any cascade begins. If the executive team will not do this, the cascade below will be understood as ritual.

Limitations

Slow: a full cascade takes weeks.
Depends on manager capability; managers who cannot facilitate a discussion of poor results about themselves will damage it.
Anonymity is fragile in small teams; set a minimum reporting threshold.
Surveys measure perception, which is real but is not the same as behavior.