It’s October. The survey runs in January. You’ve been told you own it, you’ve never run one here before, and “best practices” articles keep telling you what to ask instead of what to do first.
This is the plan. Eight weeks, in order, with the week most people skip — and the one that decides whether anyone trusts the next survey you send.
Before week one: two decisions that come before the survey exists
Don’t open the survey builder yet. Settle two things first, because if either is still open, everything below slips.
Who owns follow-through, by topic. Say you’re planning to ask about manager relationships, workload, and career growth. For each of those, who’s the named person who will actually do something if the score comes back bad? If it’s “leadership will review the results,” that’s not a name — that’s a maybe. A 150-person company we’ll use as an example here: if nobody on the leadership team owns “workload” specifically, don’t ask about workload. Cut the topic, not the trust. Asking a question you have no intention of acting on is the single fastest way to make the next survey’s response rate worse than this one’s.
What you’ll do with a bad number. Not in the abstract — the actual mechanism. If engagement in the 40-person sales team comes back low, is there a budget, a process, a person with the authority to change something? If the honest answer is “we’d just… know,” that’s a gap you need to close now, not in week eight when the data is already sitting in front of you.
These two decisions aren’t advice. They’re gates. Everything from week one onward assumes they’re already closed. Our guide to running your first employee engagement survey goes deeper on the owner decision, along with anonymity and how often to run one.
Weeks 1–2: scope and the anonymity promise
Scope first. Whatever you decided you’ll act on above — that’s your topic list. Nothing else goes in the survey. This is the discipline that keeps a 50-question “let’s just ask everything” survey from happening: every topic has to clear the ownership test from week zero. If you’re building this scope work from scratch, start from a template rather than a blank page — a built set of engagement questions, already organized by topic, is a much faster starting point than writing from nothing, and you’ll spend your time cutting instead of generating.
Then the anonymity promise — and this is the part people skip past. Before you say the word “anonymous” to anyone, decide exactly what it means in your setup, because once responses start coming in, you can’t quietly change it later.
Here’s what that actually looks like on SurveyRock. Turn on anonymous collection for a survey, and the responses that come in afterward don’t carry a name, an email address, an IP address, or device information — just the time, the channel, and whatever you declared up front, like department or location. You can turn this on after a survey is already live, but you can’t turn it back off once even one response exists. Decide before you send, not after someone asks.
The part that actually protects people, though, isn’t the missing name — it’s group size. Say your survey breaks results down by team, and one team has four people on it. Even with no names attached, if you can filter down to “just that team,” you can usually guess who said what. So any filtered view — one team, one department, one date range — is refused until at least five people have answered it. You can raise that minimum later; you can’t lower it once responses exist. Below the minimum, the filtered view just doesn’t load. That’s a real floor, not a policy you’re trusting an admin to follow.
What it doesn’t do: it doesn’t stop anyone from reading the full, unfiltered list of every comment left in the survey. If a comment is in there, someone looking at everything will see it. What the group-size floor does is refuse any filtered view of fewer than five people, so a team of four never shows up as its own slice. Tell people that distinction plainly before they answer. “Nobody can ever see what you wrote” is not true, and it’s the kind of promise that costs you every future survey the first time someone notices it wasn’t.
Weeks 3–4: build
Start from the template from week one and adjust it, rather than opening a blank survey. A blank survey invites you to write forty questions because nothing is telling you to stop; a template starts you at a reasonable count and makes you justify every addition. To decide which questions earn their place, our list of 25 engagement survey questions that work, and the 25 to cut is built for exactly that.
The question budget is a length decision, not a style preference. We pulled the completion numbers from our own survey data: short surveys hold up fine, but completion takes a real hit once you cross roughly 15-16 questions — from 98% to 77% in our data — and it never gets back to where it was. For an engagement survey going to your whole company, you want to be on the right side of that line, not relying on people who’ll push through anything. A useful constraint: for every question you add past 15, name the person who’s going to act on it. If you can’t, you’ve found a question to cut.
One more thing worth checking before you build: the free plan caps how many questions you can put in a single survey. If you’re mapping out a 20-25 question draft, confirm your plan supports that length before you’ve built the whole thing and hit the wall at the end.
Two logic mistakes almost everyone makes here. First: branching a whole section based on tenure (“if hired before 2024, show these 5 extra questions”). It feels efficient and it quietly breaks your ability to compare anyone’s answers to anyone else’s — you’ve just split your company into two surveys that happen to share a URL. Second: asking managers and individual contributors the exact same question about “my manager,” then being surprised the results don’t mean the same thing for both groups. A manager answering about their own manager and an IC answering about theirs are answering different questions that happen to share wording — segment the analysis by role before you’re surprised by it in week eight.
If branching logic is something you want — skipping a question based on an earlier answer — that’s available from the Solo plan up; it isn’t on the free plan. Worth knowing before you design around it.
Weeks 5–6: comms and the pilot
Order matters for the announcement. Managers hear about the survey before their teams do, not at the same time. Nothing erodes trust in a survey faster than an employee asking their manager about it and the manager having no idea it exists.
The invitation should state the real time commitment, and it should be true. If your survey takes eight minutes, say eight minutes — not “a few quick questions.” The moment someone feels misled about length, you’ve spent trust you needed for the close-out two weeks from now.
Reminders, on a plan, not improvised. Two planned reminders work better than three scattered ones. A survey open for two weeks: a reminder at day seven, a final nudge at day twelve, close at day fourteen.
Pilot it on four or five people first — ideally people who weren’t in the room when you wrote it. You will find at least one question that reads clearly to you and confusingly to someone outside your head. This is the cheapest thirty minutes in the whole eight weeks.
Week 7: send, and the week you watch
A normal response curve for a company-wide survey: a spike in the first 24-48 hours, a flat middle, a second smaller spike after the reminder, then a long tail of stragglers. If you don’t see the first spike at all within two days, something’s wrong with distribution — check that the link actually reached people and wasn’t caught by a spam filter — rather than waiting it out.
The one thing not to do: chase a named non-responder. The moment you message someone directly — “hey, noticed you haven’t filled out the survey yet” — you’ve told them (and anyone they mention it to) that you can see who has and hasn’t responded individually. That’s a different promise than the anonymity one you made in week two, and it undoes it in a single message. Send reminders to the whole group, or use the tool’s own reminder feature on an anonymous survey, which sends to whoever hasn’t finished without showing you who — never a one-to-one nudge.
Week 8: read, and the two-week window
Reading open-text comments by hand at volume is a real time cost — budget a half-day for a few hundred responses, not an hour. If your plan includes the theme-grouping and summary tools (that’s the Growth plan and up on SurveyRock), those will group similar comments together and draft a plain-language summary, which cuts that half-day down substantially. If you’re on a lower plan, build the hand-reading time into your calendar now, not when you’re already behind.
Separate the loudest respondent from the real signal. One long, detailed, angry comment feels like the most important thing in the dataset because it’s the most vivid. It might be one person’s bad week. Look for comments that repeat across multiple people before you treat something as a theme — that’s the difference between a pattern and a story that happened to be well-written.
Here’s the part that actually decides whether this whole eight weeks was worth it: say something, in public, within two weeks of closing the survey. Not the full action plan — just “here’s what we heard, and here’s the first thing we’re doing about it.” It doesn’t need to be big. It needs a name attached and a date. This is exactly the kind of commitment that works better with an owner and a due date attached to it than as a line in a slide deck nobody revisits. If two weeks pass and nobody’s heard anything, the silence is itself the message, and it’s the one people remember when you ask them to fill out the next survey. For what to say and when after that first commitment, the 30/60/90 plan in our complete guide to employee engagement surveys lays it out.
The one-screen version
Copy this into your own project plan:
- Week 0: Name an owner for each topic you’re considering. Cut any topic with no owner.
- Weeks 1–2: Finalize scope from your template. Decide exactly what “anonymous” means in your setup and what the real floor is — and what to tell people about it.
- Weeks 3–4: Build from the template. Hold the question count against your incentive and channel. Check logic choices against tenure-branching and manager/IC wording traps.
- Weeks 5–6: Brief managers before employees. Write the invitation with an honest time estimate. Plan two reminders: one mid-window, one final nudge. Pilot on 4-5 people.
- Week 7: Send. Watch for the first 48-hour spike. Never chase a named non-responder directly.
- Week 8: Read for repeated themes, not the loudest comment. Publish a named, dated first commitment within two weeks of close.
If you’re starting from scratch, don’t start from a blank page — start from the Annual Employee Engagement Survey template and adjust it for your org. It’s free with SurveyRock.