How to Work With “Imperfect” Data—Not Around It
People data is never static, never flawless, and never finished. And that’s exactly why it’s so valuable.
Unlike financial data, which gets reconciled to the cent, people data is shaped by real human behavior. It evolves in real time—because people evolve in real time. And yet, HR and business leaders often expect clean, complete datasets before making a move. That mindset can stall progress.
My approach?
“Get to 90% confidence, and clearly flag the 10% that’s in flux.”
If you wait for perfection, you’ll wait forever.
Here’s how I’ve learned to work with imperfect data while still driving business impact—and how you can too.
1. Progress Starts at 90%
In one role, I was asked to analyze attrition trends across a global org. The catch? Job titles and levels were inconsistent across regions. We could’ve spent months normalizing every variation—but we didn’t have that luxury.
Instead, we defined what 90% accuracy looked like. We standardized the top-level categories, flagged edge cases, and pushed the analysis forward. Leadership got the insight they needed to act, and we followed up with a deeper cleanup later.
If we had waited for perfection, we’d still be cleaning spreadsheets.
“We don’t need perfect. We need useful.” — a Finance leader once told me after we shared preliminary headcount insights.
2. People Data Is Living Data
You’ll never have a fixed snapshot of people data. Promotions happen mid-analysis. Contractors shift to full-time. Someone switches teams between the download and the presentation.
When I led a compensation analysis some years ago, one key challenge was that managers had already made changes while we were building the dashboards. The data had shifted before we even hit "refresh." So we had to reframe the project:
“This isn’t a static report—it’s a living tool.”
We aligned on cadence and cutoff dates, and made those rules transparent to everyone. Instead of frustration, we got buy-in.
3. Flag the Gaps, Don’t Hide Them
You build more credibility by saying, “This is 90% accurate—here’s where the 10% might skew the picture.”
When we launched an internal mobility dashboard, I noticed that the historical location data had holes. Instead of trying to patch it silently, I added a clear note on the first slide: “Location history may be incomplete before 2022 due to system changes.”
That small act of transparency did more for stakeholder trust than any extra chart.
“The note at the top actually made me trust the data more.” — Director, Talent Development
4. Imperfect Data Often Reveals Process Flaws
Here’s the thing about messy data—it usually points to a deeper issue. A missing job level might mean someone wasn’t onboarded properly. A misaligned org chart might reflect a gap in manager training.
Instead of seeing it as a barrier, I’ve learned to ask: What does this inconsistency tell us about our people process?
For example, when we ran a DEI analysis and saw inconsistent gender data in LATAM, it led to a broader conversation about self-identification practices across regions. That insight shaped a new HR policy—not just a dashboard fix.
5. Move Fast, Fix Along the Way
One of the myths in People Analytics is that you need a massive data cleanup project to “get ready.” But real progress happens in smaller, ongoing actions.
Every time I create a new dashboard, I take 10 minutes to correct field names, tag anomalies, or reformat tables. That work builds up over time—and reduces the backlog without slowing down delivery.
“You don’t need a data lake. You need a culture of responsibility.” — advice from a colleague during a system migration.
6. Bring Stakeholders Into the Mess
People are more forgiving of messy data when they feel like they’re part of the process. One of my best moments wasn’t delivering a perfect report—it was running a working session with our People Partners using a live dataset.
We pulled up raw attrition data and cleaned it together, discussing where discrepancies came from and what they meant. It wasn’t about the numbers—it was about co-ownership.
“This made the data feel like ours—not just something that gets emailed to us.” — HRBP, US team
Conclusion
Working with people data means working with people. And people change. Systems break. Labels shift. Data gets messy. That’s not a failure—it’s the nature of the work.
What matters is your mindset.
Start with 90%. Be honest about what you don’t know. Invite others into the process. And use those imperfections to spark deeper questions.
Because in People Analytics, progress isn’t about getting everything clean—it’s about making meaning from what’s real.