top of page

Where employee surveys still go wrong

Employee surveys have been with us for a long time and are simultaneously loved and hated by HR. Even after all this time, there are some mistakes that I feel undermine the usefulness of these surveys. Let me take the opportunity to share my thoughts, and I’d be interested in hearing yours.



The typical employee survey project has a plan. It starts with a process to select the questions, then there is communication to employees about the survey, the administration of the survey, analysis of the results, a preliminary presentation to senior management, and so on right through to getting managers to produce individual action plans. The whole process takes months.

What does that sound like? It sounds like the old-style programming process called the waterfall approach where everything is planned far in advance. The newer style of programming is called agile and it says, “Let’s not plan too much in advance, let’s take a step, get some feedback, and move forward from there.”

The danger of the waterfall approach is that you can get locked into an incredibly time-consuming round of activities that need to be ticked off while moving further and further away from what is relevant and urgent for the business.

An agile approach to employee surveys would be to ask a few questions, see what that suggests and then move forward with more questions or other actions from there.

Data first mindset

Problem number two is the data first mindset. We collect a whole lot of data first and then struggle to pull some insights out of it. Anyone who has done these surveys knows the pain of looking at reams of data and wondering what insights they are meant to draw from it. It’s an exhausting process.

Sometimes HR thinks the solution is just to create a bunch of charts and show them to management. We hope managers can find insights that were not apparent to us. Alas, managers are rarely any better than HR at wading through the mass of data. It’s time-consuming, tiring, and often results in limited value.

My approach to analytics is what I call “decision first”; that doesn’t mean you make a decision before you see the data, simply that you are very clear about what decision you are trying to make or what question you are trying to answer before you go out to get data.

Saying that we are running the survey to determine whether employees are engaged is not precise enough. Imagine instead if the decision you were trying to make is whether to hire extra technicians to create a buffer in case there is excessive turnover among technicians in the next 18 months. A survey to gather data to inform this decision would be a much smaller and faster effort than the full-blown “collect data on everything” survey. Furthermore, because we know what decision we are trying to make (that is, do we hire extra technicians), deriving insight from the data is straightforward. If the number of technicians planning to leave is high, then we make those extra hires; if not then we don’t need to do anything special in recruiting.

Questions we can’t act on

One painful aspect of employee surveys is when we get answers that we can’t act on. Perhaps you ask if people are happy with their compensation. Everyone says no. However, if you are in tough economic times then all you can say is ‘Yes, we know this is your number one concern; and no, we are not going to act on it.”

This is an example of asking a question without thinking first about what decision it is going to drive. If we can’t act on the answer to a question, then there is usually no point asking it.

No action on the results

This is the problem everyone is aware of: we do surveys every year, analyse data, and write reports but nothing happens. The usual antidote to this is insisting managers write action plans, but it can be hard to get them to do so, hard to check if they’ve executed on them, and commonly, even when they have executed on them, they don’t have any impact. I think this last well-known problem is in part a consequence of the earlier three problems. Address those problems and you are less likely to end up in a situation where you’ve run a survey and get no useful action.


The idea of the traditional employee survey is very attractive, so attractive that we are loath to change it. But perhaps it’s time to question whether the time and effort have been justified by the results. Maybe we could do better with an agile approach that is aimed at delivering data that will inform specific questions that we intend to act on.


bottom of page