Most guides to running an engagement survey start the same way: here are 25 good questions to ask your employees. That’s useful, eventually. It’s also the wrong place to start.
The questions aren’t what determines whether your survey changes anything. Three decisions made before you write a single question do that. Skip them, and you’ll end up with a survey that looks thorough, gets a decent response rate, and then sits in a slide deck that nobody acts on. Here’s the order that actually works.
Decision 1: Pick an owner before you write a single question
Say you’re planning to ask about communication from leadership, workload, and career growth. Good topics. Now ask yourself: if the workload score comes back bad, whose job is it to do something about it?
If you can’t answer that in one sentence, cut the question. Not because it’s a bad question — because an unowned bad score doesn’t get fixed. It gets mentioned once in a meeting and then nobody follows up, and six months later someone asks “didn’t we survey about this already?”
Write your topic list first. Next to each topic, write a name — a real person who will see a bad score and do something with it. If a topic doesn’t have a name next to it, either find one or leave the topic for next time.
This is also where a result needs to turn into an actual task with a deadline, not a line in a report. SurveyRock’s Action feature does this directly from a survey result: you assign an owner and a due date, and a daily email lists anything due that day or overdue. It’s available on every plan — Free gives you three open Actions at a time, which is usually enough to cover your first survey’s biggest three problems. If you’ve got more than three things that need an owner and a deadline, that’s a sign your first survey is trying to cover too much ground.
Once you’ve named your owners, the question-writing part gets much easier — we cover that in detail separately, including the 25 questions that tend to get asked and the 25 that should get cut.
Decision 2: Check what “anonymous” actually covers
Picture a 12-person team. You ask “my manager gives me useful feedback,” and everyone agrees except one person. If the manager can filter the results down to just their own team, it takes about four seconds to figure out who disagreed.
That’s the anonymity problem in one sentence: a survey can say “anonymous” on the label and still let someone narrow the data down to a handful of people. Before you trust a tool’s anonymity claim, ask it three questions:
- Can anyone in the company see which answer came from which person?
- Can a report be filtered down to a small group — a team, a location, a tenure band — small enough that one person’s answer stands out?
- What happens if only two or three people in a filtered group actually respond?
SurveyRock has a per-survey setting that collects responses anonymously. Turn it on before you launch, and the response doesn’t store who answered: no contact record, no IP address, no device details. It also won’t show a filtered view unless enough people match it. By default a filter needs at least five completed responses before anything shows, and once responses start coming in that minimum can be raised but never lowered. So a manager trying to filter down to “just my team of four” gets refused.
For your first survey: turn the setting on before launch, and don’t let anyone filter down to small groups.
Decision 3: Pick a cadence you can actually keep up
A team announces “we’re doing an annual engagement survey.” Year one, good turnout — people are curious, HR promotes it, the novelty helps. Year two, response rates drop, because nothing visibly changed after year one. By year three, the survey itself becomes the thing people are cynical about.
The fix isn’t a better survey. It’s picking a rhythm you can actually sustain, before you promise one.
There are really two patterns, and a hybrid of both:
- An annual deep-dive — a longer survey, once a year, covering most of your engagement topics at once.
- A recurring pulse — a short, frequent check-in, maybe five to eight questions every month or quarter, on one or two topics at a time.
- Both together — the annual survey sets the baseline, the pulse tracks whether anything’s moving between baselines.
If you’re wondering whether once a year is enough on its own, we make the case for the short survey in between in The Annual Engagement Survey Is a Once-a-Year Confession.
Length and frequency trade off against each other directly — a shorter survey can run more often, a longer one needs more runway between sends. We’ve written up the actual decision rule for survey length, with real completion-rate data behind it, which is worth reading before you commit to either pattern.
If you want the pulse to go out on its own without relaunching it each time, recurring email campaigns do that on the Growth plan and up. If you’re running this by hand for now, that’s fine — just pick a cadence you can keep up manually for the next year, not the one that looks most thorough on paper.
Building the survey itself
You don’t need a blank page. Survey templates and folders are available on every SurveyRock plan, including Free, so your starting point is already a reasonable first draft of an engagement survey, not an empty canvas.
Then go through it with your list from Decision 1 and cut any question nobody owns. We’ve written up which 25 questions earn their place and which 25 usually don’t.
If you’re on the Solo plan or above, you can also describe the survey you want and have the questions and answer choices drafted for you, which you can then keep, edit or delete. It’s a faster start than a blank page, especially the first time.
Either way: start from the template, cut anything that doesn’t have an owner, and don’t add a question just because a “best practice” list told you to.
Distributing it so people actually answer
Getting the survey built is the easy part. Getting a decent response rate is where most first-timers lose ground.
A few things matter more than people expect:
- Channel. Company email is the default, and email campaigns with invitations, reminders, and open tracking are available from the Solo plan up. If your company genuinely lives in Slack or Teams, posting the survey link there (SurveyRock writes the message for you to paste, on the Team plan) usually beats email, because people are already looking at it.
- Timing. Don’t send it Monday morning, during a payroll close, or the same week as a reorg announcement. Picture a Tuesday or Wednesday mid-morning, when people have cleared their inbox from the weekend but aren’t yet buried in afternoon meetings.
- Device. A meaningful share of your employees will open this on a phone between meetings, not at a desk. If the survey doesn’t hold up on a small screen, you’ll lose responses to people giving up halfway through — worth checking against a mobile-first design pass before you launch, not after your response rate disappoints you.
Reminders matter more than most people assume. A single send gets you the people who were already going to answer. A reminder two or three days later, sent only to people who haven’t finished, catches a meaningful second wave — which is why tracking who’s responded (without exposing what they said) is part of the point of a real email campaign tool, not just a nice-to-have.
More on building a survey that holds up on a phone screen.
What happens after the responses come in
This is where Decision 1 either pays off or gets exposed. An unowned survey dies right here — results get reviewed once, maybe summarized in a meeting, and then nothing visible happens.
The minimum version of doing this right has three steps:
- Read the results. Use the response dashboard to see where scores are low, and filter by team or location to see whether the problem is company-wide or specific to one group.
- Share what you found — including the uncomfortable parts. People can tell the difference between “we heard you” and “we heard you, selectively.” If workload came back bad, say so, even if there’s no fix ready yet.
- Commit to one visible change. Not ten. One thing, with a name and a date attached, that people can actually see happen. This is the same Action setup from Decision 1 — turn the worst score into an owned, dated task, and tell the team what it is.
The survey itself isn’t the deliverable. The visible change afterward is. If you only do one thing differently on your first engagement survey, make it this: pick fewer topics, name an owner for each one, and make sure at least one of them turns into something people can point to and say “that happened because we said something.”
Ready to build it? Start from a survey template on SurveyRock, with Actions built in for when the results come back. The Free plan allows up to 10 questions per survey, which is enough for a short first survey or a pulse; a fuller engagement survey needs a paid plan.