Why Your Client's Employees Don't Trust Your Survey
An outside firm running an anonymous survey faces a distinct trust problem. See what employees worry about and what changes their minds.
Short answer
A client's employees distrust an external consultant's survey when the invitation blurs who is running it or relies on a vague promise that nobody will look at individual rows. Naming both firms plainly and using a tool that encrypts answers in the browser addresses the threat assessment employees actually run.
You have run the survey. Response rate is 38%. The free-text answers are polite. The client says "well, that matches what we thought", and you both quietly know the exercise did not work.
The reason usually is not your questions.
01What an employee is actually doing when your email arrives
They are not thinking about engagement. They are doing a threat assessment, and it takes about four seconds:
- Who is this firm I have never heard of.
- Who hired them, and what were they told to look for.
- Does my manager see this.
- If I say the true thing, and there are nine people in my team, can they work out it was me.
An outside firm makes every one of those worse, not better. The employer at least is a known quantity. You are a name in a footer with a survey link attached.
02The three mistakes that confirm their suspicion
Sending it under the client's logo only. Feels friendlier. Reads as concealment the moment somebody notices the link goes somewhere else, and somebody always notices. Now the question is not "is this anonymous", it is "why did they hide who is collecting it".
Sending it under yours only. Now it looks like a third party is compiling a file on employees on behalf of management, which is the single most damaging reading available and the hardest to walk back.
Promising anonymity in a sentence. "Your responses are completely anonymous" is what every survey says, including the ones that were not. It is a claim without a mechanism, and people who have been burned once do not grade claims, they grade mechanisms.
03What changes it
Name both parties, in order. Your firm, on behalf of the client, at the top of the page before the first question. It is the least defensive thing you can do, and it removes the entire "why is this hidden" branch of the threat assessment.
Say what the tool cannot do, not what you promise not to do. These are different sentences and employees can tell them apart. "We won't look at individual responses" is a policy. "Answers are encrypted in your browser before they are sent, so the survey company stores scrambled text it can't read, and nobody, us included, can tell which answers are yours" is a mechanism.
A policy can change after the survey closes. A mechanism cannot.
Tell them the small-team rule before they ask. The fear is not really "can they read my answer", it is "there are six of us, so it does not matter whether they can read it". Say what happens to a group too small to report: it stays sealed, nobody sees the slice, not you and not the client. If the tool shows small groups with a warning label, that is not a floor, and people who have been through a reorg know the difference.
Give them something to read that is not from the client. A one-page handout at kickoff, addressed to the employees rather than the buyer, does more for participation than a reminder email. We publish one you can print: how this protects you.
04The uncomfortable part
If your survey tool cannot support any of that, the trust problem is not a communications problem you can write your way out of. Most tools cannot say "we are unable to read individual answers", because they can, and their whole product depends on it.
That is why we built ours so that we cannot. Responses are encrypted in each person's browser before upload. What our servers hold is ciphertext.
You and the client see scores as totals, and written answers word for word but shuffled and undated, so there is no version of the admin screen that shows who said what. Where the poll allows it, anyone who'd rather not be read can send just the topics they pick instead. And the plaintext never reaches our servers.
It also means you can hand the client a results document signed by the workspace, with the cohort counts and the anonymity floors that were actually in force, and their own security team can verify it without asking us anything.
For related guidance, see confidential vs anonymous surveys, how to run an anonymous employee survey, and anonymity thresholds for small groups.
FAQQuestions people ask
Both, with the client named as the party it is for. Either brand alone invites a worse reading than the truth.
Explaining the mechanism does not mean explaining the algorithm. One sentence about where the answer is scrambled and who cannot read it does the work; the detail belongs on a page they can click.
It depends on the client and the moment far more than on the tool, but the recoverable gap is almost always trust rather than convenience. Reminder emails do not move a threat assessment.
The opposite, in our experience. The damage comes from ambiguity about who is involved, not from the presence of a consultant.
InviziPoll team
Updated Oct 29, 2026 · 4 min read