You have employee survey results and other workplace data—and you still can’t explain what’s happening.
You want to make good decisions about your organization, and you’ve made a real effort to listen to employees and try to understand their experience. You have an engagement dashboard, three years of employee survey results, open-ended comments, turnover numbers, exit interview summaries, pulse surveys, HRIS data, program participation data, and even a consultant’s report. Basically, you’re not short on information.
But then someone in a leadership meeting asks, “Okay, but why is this happening?” And nobody really knows. You’re left thinking, We have all this information. So why don’t I feel confident that I understand what it means?
The problem isn’t that your organization hasn’t been listening or hasn’t collected enough information. In fact, you may have done exactly what a thoughtful leader is supposed to do: gathered evidence, asked employees for their perspectives, and tried to make decisions based on what you learned.
But having lots of employee data doesn’t necessarily mean you understand the problem any better. Your data can tell you a great deal about what is happening in your organization without necessarily explaining why it’s happening. A data point can reveal a pattern, a difference, or a change over time. It can tell you where to pay attention. But it can’t always explain what produced the pattern you’re seeing.
That can put you in an uncomfortable position. You have enough information to know that something deserves your attention, and people are looking to you to decide what to do about it. You might feel pressured to turn what the data shows into an explanation, even when the evidence doesn’t support one yet.
You don’t have to make your data answer a question it’s not capable of answering.
Saying “we don’t know why yet” isn’t ignoring the evidence. In fact, “we don’t know yet” can be a rigorous conclusion because you’re being precise about what the evidence does—and does not—allow you to infer. That distinction matters, because an interpretation that gets treated as a finding can quickly become the basis for what you do next.
That doesn’t mean your data is bad or your survey was a waste; quite the opposite, actually. The information you already have may be extremely useful. The important thing is to get clear on what the evidence allows you to conclude and what it really doesn’t allow you to conclude.
This is where I usually start when I’m trying to understand an organizational problem. Organizations have lots of data, but the evidence you can glean from that data is rarely neat. Different sources can point in different directions. What employees say in a survey might not line up with what managers think is happening. Turnover data may reveal a pattern that exit interviews don’t fully explain. And sometimes, the information you have simply wasn’t ever going to answer the question you’re asking now.
My job as a researcher isn’t to dismiss those different forms of evidence or force them into a single explanation. It’s to look carefully at what each source can tell us, distinguish what the evidence shows from what we’re interpreting, and identify the questions that remain unanswered. You can start doing the same thing with the information you already have.
A Pattern Is Not the Same as an Explanation
Suppose your employee survey shows that only 42% of employees agree that communication from leadership is effective. Now, you know something about employees' experience of leadership communication. Maybe the score has fallen since last year. Maybe it's especially low in one department. Maybe communication also comes up repeatedly in open-ended comments. It can be tempting to look at that evidence and conclude: We need to communicate more often.
Maybe you do. But the data hasn't established that yet. Employees might be frustrated by how often leaders communicate. Or they might be telling you that important decisions aren't explained, messages from senior leaders and managers don't line up, they hear about changes too late, or that they’re being given information without meaningful opportunities to ask questions or respond. What employees call a “communication problem” might even involve something less obvious, like trust, transparency, fairness, or whether they feel they have a voice.
The survey has identified a pattern worth paying attention to. It hasn't necessarily explained what produced that pattern. That's where it's useful to separate what the evidence shows from what you're interpreting.
The evidence: Employees rate leadership communication poorly.
The interpretation: Employees want leadership to communicate more frequently.
The interpretation isn't necessarily wrong. In fact, it might be a very reasonable hypothesis based on what you already know about the organization. But until you have evidence that supports it, it's still a hypothesis.
And that distinction matters because interpretations have consequences. If “employees want more communication” becomes accepted as the explanation, the next step seems obvious: send more updates, hold more town halls, ask managers to communicate more frequently. In this way, you could end up doing a very good job of solving a problem employees weren’t actually describing, while the problem they were describing continues.
This is where it helps to get precise about what the evidence actually tells you and what you’re interpreting from it.
Start With What Your Employee Data Actually Tells You
When you're trying to make sense of employee data, it helps to separate what the evidence actually establishes from the meaning you're attaching to it.
Start with what you can confidently say:
We know: Our engagement score fell nine points.
We know: Trust in leadership is lower in Department X.
We know: Comments about workload increased.
We know: First-year turnover is higher than turnover among other employees.
Then look at what you've inferred:
We think: Managers aren't communicating enough.
We think: Employees don't understand the strategy.
We think: People are burned out.
We think: The reorganization damaged trust in leadership.
Those interpretations may be thoughtful and well-informed. They just aren't necessarily what the evidence established.
Your interpretation isn't the problem. The problem begins when you treat your interpretation as though the evidence established it as true. Leaders, HR professionals and managers bring context, experience and organizational knowledge that matters. “We think” is a legitimate and useful place to begin. It can generate hypotheses and point you toward what to prioritize in your investigation. But just be careful that “we think” doesn’t slip into “we know.”
Look Across the Employee Evidence You Already Have
Once you've separated what the evidence shows from what you're interpreting, you might be tempted to collect more data to fill in the gaps. Before you do, take a look at what your organization already knows.
Your employee survey is only one source of information. You may also have open-ended comments, turnover data, employee feedback from other listening efforts, exit and stay interview data, employee relations information, absence patterns, program participation data, findings from previous research or consulting work, manager observations, and informal feedback employees have shared along the way.
Looking across all of that evidence doesn't just mean gathering it in one place. You're trying to understand what the different sources can tell you when you consider them together. Where do they reinforce one another? Where do they seem to tell different stories? And what can none of them explain?
Suppose your engagement survey shows that trust in leadership is especially low in one department. When you look at your turnover data, you see that voluntary turnover is also higher there. Open-ended survey comments mention communication and inconsistent decision-making. Exit interviews include concerns about management.
Now you know more than you did from the engagement score alone. Several sources are pointing your attention toward the same part of the organization, and they're beginning to give you a clearer picture of what employees are experiencing. But be careful not to make the same leap from pattern to explanation.
Taken together, the evidence still may not tell you what employees experienced that led them to distrust leadership. You may not know what they mean when they talk about inconsistent decisions or communication. You may not know whether employees who left experienced the department differently from those who stayed. And you may not know whether the same conditions are affecting everyone or whether different groups of employees are having very different experiences.
That's why looking across your evidence can be useful even when it doesn't give you a complete answer. It can help you see where the evidence converges, where it complicates the story you thought you understood, and where important gaps remain. Those gaps aren't a reason to throw out the data you have. They're clues about what you need to understand next. And sometimes that's the most useful thing your existing data can tell you.
Turn What Your Employee Data Doesn’t Explain Into a Question You Can Investigate
Once you've looked across the evidence, the unanswered questions become easier to see. That's not a failure of the analysis; instead, those unanswered questions tell you what you need to understand next.
Maybe you know that trust in leadership has declined. What you don't yet understand is what experiences are shaping employees' trust in leadership. Maybe you know that participation in a new program is low. What you don't yet understand is how employees are experiencing the program and what exactly influences whether they participate. Or maybe you know that one department's engagement score is considerably lower than the rest of the organization. What you don't yet understand is how the employee experience in that department differs from other areas of the organization and what conditions might be shaping that difference.
This is especially important when you're dealing with a complex organizational problem, because there may not be one explanation. Low trust, for example, might be connected to a recent reorganization, inconsistent decisions across managers, workload pressures, past experiences with leadership, or differences in how employees experience the organization depending on their role or location. Several of those things might be happening at the same time. And they may affect different employees differently.
So the goal isn't necessarily to find the reason. It's to develop a question that helps you investigate what's shaping the pattern you're seeing without deciding on the explanation before you've done the work. That might mean moving from: why don't employees trust leadership? to: What experiences and organizational conditions are shaping employees' trust in leadership, and how do those experiences differ across the organization?
The second question leaves room for complexity. It doesn't assume there's one cause, that every employee is having the same experience, or that you already know where the problem is. That's what makes a good question useful. It gives you something you can investigate without narrowing down the answer too soon.
When Your Employee Data Tells You Different Stories
Complex organizational problems don’t often produce perfectly consistent evidence. Your employee survey results might show that employees generally understand the organization's strategy, while interviews suggest people aren't sure what the strategy means for their day-to-day work. Managers may believe a new initiative is going well while participation data shows employees aren't using it. Exit interviews might point toward compensation while current employees talk more about workload or management.
The temptation is to decide which source is right. But disagreement between sources can itself tell you something important. Different groups may be having different experiences. Employees may answer differently depending on how a question is asked or when they're asked. Managers and employees may be seeing different parts of the same problem. Or two findings that initially seem contradictory may both be true.
Instead of immediately resolving the contradiction, ask what might explain it: Who is represented in each source? What part of the employee experience does it capture? When was the information collected? What might one source be showing you that another can't? Sometimes the places where your employee data doesn't line up are exactly where you need to look more closely.
Before You Collect More Employee Data, Decide What You Need to Understand
Once you have a clearer question, you can decide what kind of evidence would help you answer it. That doesn't always mean collecting new data. The information you need may already be in open-ended survey responses that haven't been examined closely. You may need to look at turnover or participation data by role, department, tenure, or another meaningful group. You might need to compare employee feedback before and after a change. Or you may need to revisit previous research with a different question in mind.
You may also discover that you really do need more information. Another employee survey might make sense. So might interviews, focus groups, document review, additional quantitative analysis, or some combination of these.
The question you're asking should determine what evidence you look for next.
The point isn't that surveys can't tell you why something is happening because sometimes they can. And interviews aren't automatically the answer simply because you're trying to understand employee experience. What matters is whether the information you have (or the information you decide to gather) can actually help you answer the question you're asking.
Otherwise, it's easy to keep adding information without getting any closer to understanding the problem. Another dashboard. Another survey. Another round of employee feedback. You have more data than you did before, but no better explanation of what's happening.
The better starting point isn't what else should we collect? It's what are we trying to understand? Once you can answer that, you can make a much more deliberate decision about what evidence you need next or whether you already have it.
Let the Question Lead You Forward
You may already have a great deal of useful information about your employees. The goal isn't to discount it or immediately go looking for more. It's to understand what that information allows you to say with confidence, where different pieces of evidence complicate the picture, and what remains unexplained.
With complex organizational problems, that may not lead you to one clean answer. You might find several conditions contributing to the problem. You might find meaningful differences across groups. And you might discover that an explanation you started with is only part of the story. That's still progress.
The goal isn’t certainty for its own sake. It's knowing what your organization reasonably understands, what it doesn't understand yet, and what it needs to learn before deciding what to do next.
That's the process I think about as: Listen → Investigate → Understand → Move Forward.
Listen to what the evidence is telling you. Investigate the questions it raises. Build a deeper understanding of what's happening. Then use that understanding to decide what to do next.
The goal isn't to have all the data. It's to understand enough to make a defensible decision about what comes next—whether that's learning more, changing course, or taking action based on what the evidence actually supports.
Want to get clearer about what your employee data supports and what you still need to understand?
What Are You Trying to Understand? is a free practical guide designed to help you separate what you know from what you're assuming, take stock of the evidence you already have, identify what it can't explain, and turn those gaps into questions you can investigate.