How Do You Prove the Survey Is Right?

“How do you know it’s right?”

Every survey you send out the door is a claim. A boundary is here, this corner is that, the grade falls this way. Someone – a client, a contractor, a reviewing surveyor, a court two years from now – may rely on that claim for a very long time. And every one of them is entitled to ask the same question: how do you know?

At most firms, the honest answer is a chain of reputation. An experienced surveyor looked at it. We have done this a thousand times. Our name is on it. None of that is nothing – it is the accumulated judgment that makes a firm worth hiring. But it is not evidence. It does not scale past the one or two people whose judgment it rests on, it does not travel with the deliverable, and it does not survive the moment the work is actually challenged. “We are careful” is a statement about the firm. It is not a statement about the dataset in question.

The firms that will be hard to compete with over the next decade are the ones that can answer the question a different way: with a record. Not “trust us” – here is the check we ran, here is what it found, here is the grade.

Three things go wrong, and they don’t look alike

The reason “just double-check it” is not a QC process is that survey error comes in three flavors that fail differently and hide differently. A process that catches one and misses the others produces false confidence, which is worse than no confidence.

Blunders. A rod height keyed as 6.00 instead of 5.00, a bad instrument setup, a point shot to the wrong target. These are large, discrete mistakes. They are usually obvious once you look at the right number – but they are individual needles in a haystack of thousands of shots, and finding them by eye is exactly the task human attention is worst at.

Systematic error. A datum mismatch, an unexpected scale factor, a misconfigured coordinate system that shifts every point by a consistent amount. Point by point, nothing looks wrong – each shot is internally plausible. The error only appears when you look at the dataset statistically, as a distribution, rather than one point at a time.

Bad codes and geometry. A feature miscoded, a linework string that crosses itself, a code the office GIS does not recognize. This is not a precision problem at all – it is a deliverability problem. It does not make the survey wrong; it makes it stall on the CAD team’s desk while someone untangles it by hand.

Catch only the blunders and a scale factor ships. Catch only the statistics and a self-crossing boundary reaches the client. Real QC has to run all three passes, every time, on the whole dataset – which is precisely the kind of exhaustive, repetitive checking that does not survive a deadline when it depends on someone remembering to do it.

What a defensible QC record actually looks like

This is the work PointScout Desktop was built to make automatic. When field data comes in, it runs the deliverable through the checks and produces a single audit report – the artifact that answers “how do you know.”

Concretely, that report does four things:

• A statistical check for outliers. Elevations and residuals are evaluated as a distribution, and anything beyond three standard deviations is flagged. This is the pass that catches systematic drift and the subtle blunders that look fine individually. “Zero elevation outliers beyond 3 sigma” is a sentence with evidence behind it.
• Blunder and setup detection. Targeted checks for the common field mistakes – rod-height jumps, Z-spikes, suspect setups – surfaced as specific warnings tied to specific points, not a vague “looks off.”
• Feature-code and geometry validation. Every code checked against the firm’s library, every geometry checked for validity. Unrecognized codes and invalid linework are caught here, before they reach CAD, not after.
• A grade. The whole thing resolves to a summary – a score, a count of criticals and warnings, a letter grade – so a reviewer can see the headline in one line and drill into the detail if they want it.

The point is not any single check. It is that the checks run the same way every time, on every dataset, and leave behind a record – so the answer to “how do you know it’s right” is a document, not a memory.

Proof is a deliverable, not a favor

Here is what changes when QC produces an artifact instead of a feeling.

It wins work. “Here are your points” and “here are your points, with a QC report that grades the dataset A and shows zero outliers beyond three sigma” are two different sales conversations. One asks the client to trust you. The other shows them why they can. In a market where every firm claims to be careful, the one that proves it stands out – and gets asked back.

It defends work. When a boundary is contested or a deliverable is reviewed, the firm that can produce a dated QC record of exactly what was checked and what it found is arguing from evidence. The firm reconstructing its diligence from memory is arguing from a position of weakness. The audit report is not just a sales tool; it is the record you want to have when the work is questioned.

It protects you from yourself. Most of the errors a QC pass catches are never seen by anyone outside the firm, because they are caught before delivery. The report you run to reassure the client is the same one that stops your own blunder from becoming your own liability. The cheapest error to fix is the one you find on your own screen.

Where the proof comes from

None of this replaces the surveyor’s judgment – it arms it. The report does not decide whether the survey is acceptable; it hands the person who makes that call the evidence to make it well, and a record of having made it.

It also does not start in the office. The cleanest audit report is the one with the fewest surprises in it, and the way you get there is to catch the gross errors in the field, on-site, while the crew is still standing on the point – which is exactly what PointScout Mobile does, free, on every crew’s phone. Field data flows into PointScout Desktop, the report runs, and clean LandXML flows out to CAD. Field-to-finish, with the proof generated along the way instead of reconstructed at the end.

The question is not going away. Clients will keep asking how you know, reviewers will keep asking what you checked, and the occasional dispute will keep asking you to prove it. You can keep answering with your reputation. Or you can answer with a record – and let the work speak for itself.