How to make a survey actually anonymous (and get people to trust it)
Saying "this survey is anonymous" in the intro doesn't make it anonymous — what makes it anonymous is how it's built and how it's distributed. If either of those leaves a way to identify who answered, the anonymity promise is false, and people who suspect that answer more cautiously (or just stop being honest).
First: don't ask what identifies someone
The most common mistake is leaving "innocent" questions that together identify someone anyway: team, exact tenure, specific role on a small team. On a 4-person team, asking "years at the company" and "role" can identify someone just as well as asking for their name directly. If anonymity matters, it helps to check every question against "does this, combined with the other answers, narrow it down to one possible person?"
Second: how you distribute it matters as much as what you ask
Distributing the survey through a shared general link (the same one for everyone) is anonymous by design: nobody gets tied to a specific answer on the platform side. Distributing it through a contact panel with personalized per-person links isn't, even if the questions themselves don't ask for identifiable data — the system knows which answer came from which link, and therefore which contact.
Don't require a login
If the survey requires someone to log into their account to answer, any anonymity promise stops being credible, no matter what the questions say. Publishing the survey with an open public link, no account required, is part of the same principle: the less identifying infrastructure sits between the person and the answer, the more real the anonymity is.
Communicate it, on top of guaranteeing it technically
Even with all of this set up correctly, if nobody explains how the anonymity works, people assume the worst. A short line in the intro explaining specifically what makes the survey anonymous (not just the word "anonymous" on its own) builds more trust than the promise without context.
Why it's worth the effort
Distrust about anonymity doesn't make people stop answering — it makes them answer less honestly, which is worse than not having the data at all: it creates a false sense that everything is fine, when the most important criticism never actually reached the survey.