Every survey tool brags about completion rate.
Higher completion, better tool. It is the number on every landing page, including ours. Conversational forms get higher completion than static ones, the data is real, and we have written about it.
But I want to push on it, because completion rate has a problem hiding in plain sight.
It is trivial to game, and gaming it makes your survey worse.
The easiest way to get a perfect completion rate
Want a 100% completion rate? Ask one question. Make it a single tap. "How was your experience?" with three emoji.
Everyone finishes. Your dashboard glows green. And you have learned almost nothing.
You know a vague direction of sentiment and nothing about why. You cannot act on it. You cannot tell the product team what to fix. You optimized the metric and lost the point.
This is the tell of a vanity metric. You can move it without creating any of the value it was supposed to represent.
Completion rate measures whether people got to the end. It says nothing about whether getting to the end was worth anything to you.
What we actually want from a survey
Step back and ask what a survey is for.
It is not for completions. It is for understanding. You are spending something precious, a slice of another person's attention, and you want to convert that attention into the most useful knowledge you can.
So the real question is not "what fraction finished?"
It is "how much did we learn per minute of attention we spent?"
I think of this as insight density. Useful, specific, actionable understanding divided by the time and effort you asked the respondent to give.
A survey can have high completion and terrible insight density. Twenty lazy multiple-choice questions everyone clicks through on autopilot. High completion, low truth.
A survey can have lower completion and excellent insight density. Three sharp questions with real follow-ups, where the people who finish tell you something you can actually use.
The second survey is better. Completion rate ranks it lower.
Why high completion can mean low signal
Here is the uncomfortable part. Sometimes high completion is a symptom of a survey that asks for nothing.
Effort and honesty are linked. The questions that produce the best answers are often the ones that ask the respondent to think. "What specifically made setup confusing?" is harder than "Rate setup 1 to 5." The hard question has a higher chance of being skipped, and a far higher chance of teaching you something.
If you optimize purely for completion, you systematically remove the questions that require thought. You sand the survey down until it is frictionless and empty. Everyone finishes because there was nothing to finish.
A perfectly smooth survey and a perfectly useless one can look identical on a completion dashboard.
Measuring insight density instead
You cannot put insight density on a dashboard as easily as completion, but you can get close, and you can change what you pay attention to.
A few things I would watch instead of, or alongside, completion:
How many open-ended answers were substantive rather than one word. Three words tells you the respondent disengaged even if they technically answered.
How often a follow-up actually changed your understanding. If asking "why" never surfaces anything new, your survey is decorative.
How much of the result a team could act on. Read the responses and ask: did this change a decision? A survey that changed nothing was a cost, no matter how many people completed it.
Whether the people who dropped off were the ones you needed. Losing bored respondents on question two is fine. Losing your most engaged power users on the question that mattered is not. Completion rate treats both the same. It should not.
How much attention you spent to get there. A 90% completion on a survey that took eight minutes is not obviously better than 70% on one that took ninety seconds. Per minute of respondent time, the short one may have taught you more.
None of these is a single clean percentage. That is exactly why they are harder to fool.
Completion still matters, in its place
I am not saying ignore completion. A survey people abandon halfway is a real failure, and a sudden drop at one question is one of the most useful signals you have. It usually means that question is confusing, repetitive, or asks too much too soon.
So completion is a great diagnostic. It tells you where the survey hurts.
It is a terrible goal. The moment you optimize for it directly, you start removing the friction that was doing the actual work.
Use it like a thermometer, not like a scoreboard.
What this means for Formaly
We will keep reporting completion, because it is genuinely useful for finding where a survey breaks.
But the features I care most about are the ones that push insight density up, not just completion. Follow-up questions that turn a thin answer into a real one. Logic that skips what does not apply so the questions that remain are worth asking. Analysis that pulls the actual themes out of open text instead of leaving them to rot. Drop-off views that tell you which question lost people, so you can fix the question instead of deleting it.
The goal was never to get more people to the end of a survey that asks nothing.
The goal is to learn the most you possibly can from the attention people were generous enough to give you. Completion rate is one clue toward that. It was never the prize.