Real variance or bad data? 4 causes of false cut/fill results to rule out first

The survey says you’re off grade. But what if the survey is wrong?

Your cut/fill comparison came back with a variance that doesn’t make sense. The site is showing high in places the crew reports building to grade. The numbers don’t match the conditions on the ground. Before you start a difficult rework conversation, before you reschedule a crew or call a second survey, there’s one question worth asking first.

Is the variance real?

False variance means the comparison shows a problem that doesn’t exist on the site. Start by checking the coordinate system, survey references, and design version before deciding what needs fixing. Acting on false variance is expensive in exactly the same ways as acting on real variance: crew time, mobilization, delay, and the loss of confidence that comes with sending people to fix something that didn’t need fixing.

The webinar this blog series follows covers how experienced teams validate a variance before acting on it. Here are four checks to help you validate the result.

Cause 2: The surveys have drifted vertically over time

Ground control isn’t a set-and-forget thing. If your ground control points (GCPs) shift, settle, or get disturbed between surveys, the vertical relationship between your earlier and later captures can drift, even if the drone, the processing, and the platform are all doing exactly what they should.

This shows up as apparent variance that’s relatively consistent across the site: everything running slightly high or slightly low. It can look like a real grade issue when the actual problem is that your reference elevation has moved.

What to check: Compare a stable, well-understood reference point (a check shot, a known benchmark, or a control point from the original survey setup) across your recent captures. If that reference point is showing a shift, your vertical relationship between surveys has drifted and the comparison needs to be re-anchored, not acted on as-is. Re-survey the GCPs, confirm your check shots, and reprocess before making any field decisions from the comparison.

This is also a good reason to include check shots at known elevations in every survey as a matter of routine. They don’t take long to capture and they make this kind of validation trivial. Without them, you’re relying on the comparison itself to detect a problem that’s actually in the reference data.

Cause 4: You’re checking so late you can’t explain the variance even if it’s real

This one is a little different. The variance might be real. But if you’re running your first design conformance check at phase closeout, you’ve lost the ability to trace when the issue appeared, what was happening on site at the time, and whether it was caused by a design revision, a coordinate issue, or actual construction error.

Without a survey history, it’s harder to establish when the variance appeared. Earlier captures help you check whether it developed during this phase, was present before work started, or followed a design revision.

That’s why the teams who catch variance early, who run regular checks throughout the project, are also the teams who handle late discoveries with far less friction. They have the timeline. They have the data. They can show exactly when the variance appeared, what the site looked like before and after, and what was happening during that window.

For more on what a regular conformance check workflow actually looks like, the step-by-step guide covers it from calibration through to exported report. And why late discovery costs more gets into the cost side of that same issue.

Watch the full walkthrough on demand

Not sure where to start with your own conformance setup? Take the readiness quiz. Two minutes, clear output.