Better survey questions don't get you better answers. The quality of the questions is almost irrelevant compared to the quality of the trust environment. And trust doesn't come from survey design.
Psychological safety is the precondition for honest feedback. Not one of several factors. The. Precondition.
When someone doesn't feel safe saying what they think, the channel they use to say it is essentially irrelevant. They'll answer in a way that reflects how safe they feel, not how they actually feel. You get performance, not candour.
What does "safe" mean here? Not safe in a legal sense, or in the sense of a policy on the wall that says we encourage openness. Safe in the sense that the person answering can't imagine the information being used against them. Can't imagine being identified from their response. Can't imagine raising a concern and having it surface in their next performance review. That kind of safety is not declared. It's earned, over time, through consistent behaviour from the people in power.
This creates a real problem for feedback tools. Most of them can't fix the trust environment. They can ask better questions, offer anonymisation, produce cleaner reports. What they can't do is make someone trust that the information they share will be treated honestly. Only the organisation's actual track record with feedback can do that.
There are things a tool can do that help. Genuine anonymity, the architectural kind where responses can't be traced even in principle, removes one class of risk for the person answering. It doesn't create psychological safety on its own, but it eliminates the specific fear of individual identification. That's a meaningful thing to eliminate.
Frequency helps too. A weekly check-in that gets responded to, consistently, over months, builds evidence. Not just that the channel is anonymous, but that it's used. That information comes through and things change. That the organisation's relationship to feedback is genuine. Not performative. This kind of safety is accrued slowly, by accumulation. A single survey that goes nowhere can undo months of it.
The thing that destroys psychological safety fastest is asking for feedback and visibly doing nothing with it. It's not just that nothing improved. It's that the organisation demonstrated, clearly, that the exercise was performative. People update their beliefs about what this channel is for, and they behave accordingly from that point on.
Relay can't guarantee that an organisation will respond honestly to what it hears. What it can do is make the channel itself as safe as possible, give leaders the information they need to respond, and make the cadence regular enough that the response becomes habitual rather than exceptional. The rest is on the organisation. And organisations that make that commitment get genuinely different signal than the ones that don't.
If you're getting flat, neutral, or declining responses to feedback, the question isn't "how do we improve the survey?" It's "do people actually trust what happens to their answers?"
Relay is built to earn that trust, not assume it. Try it for free and see what genuine safety does to the quality of what you hear.