Engagement programmes fail in a small number of predictable ways, and almost none of them are about the survey. They fail because the loop is never closed, because the results are reported at a level nobody can act on, because the causes sit with line managers while the programme sits with HR, or because the questions promised action on things nobody had authority to change. This page names those four, and is honest about the awkward structural fact underneath them: the people function usually owns the measurement and not the causes.
The loop is open, and everybody can tell
The commonest issue by a distance: a survey runs, a deck is presented, and nothing visible happens before the next one. People do not conclude that the company is bad at follow-through; they conclude that asking was a formality, and they answer the next round accordingly or not at all. The fix is unglamorous and cheap: commit to fewer things, do them, and say what was done in the same channel the survey arrived in.
The result is reported where nobody can act
A company-level score is a headline. Action happens in teams, so a result that stops at the company average or at a division of four hundred people leaves managers with nothing to do and HR with a number to defend. The counter-constraint is real: teams small enough to identify individuals must not be reported separately. Between those two sits the reporting level that is actually useful, and finding it is a design decision to make before the survey, not after.
HR owns the measurement and not the causes
Workload, headcount, pay bands, the quality of a particular manager, whether a promised system ever got bought: these are what the survey surfaces and almost none of them are the people function's to fix alone. A programme designed as if HR can act on its own findings produces initiatives in the space HR controls, which is communications and events, and those are not where the score is. Naming this at the start, and getting each finding an owner outside HR where it belongs, is what separates programmes that move from programmes that repeat.
You asked about something you cannot change
Every question implies willingness to act. Asking about pay with no budget, or about hybrid policy that is set elsewhere, produces a predictable low score, a comment set full of the same thing, and no action, and it costs credibility on everything you could have acted on. Cut the question set to what is genuinely in play, say that you have done so and why, and you will get a shorter list of things you can actually finish.
Questions people ask about employee engagement issues
Our score dropped after we started acting. What happened?
Usually attention: asking people to think about something raises their standard for it, and a company that starts taking engagement seriously often surfaces problems that were always there. Read the comments and the per team movement rather than the headline, and give it a second round before concluding anything.
Response rate is falling. Is the survey the problem?
Almost never. Falling participation is the most reliable available measure of whether people believe anything follows, and the fix is on the action side rather than in the instrument or the reminder schedule.
Should HR run the programme at all?
HR should run the instrument, because it needs consistency and confidentiality. Ownership of the findings has to sit with whoever can change the thing found, and making that explicit at the start prevents the whole programme being judged on results HR could never have moved.