Accuracy 101: relative vs. absolute, and the crawl-walk-run path to maps you can trust

"How accurate is it?" is the first question every project manager asks about a drone map, and it has no single answer, because there are two different things the word can mean. Getting them straight tells you exactly how much equipment and process a job actually needs — and where you're paying for precision you won't use.

Two kinds of accuracy

Relative accuracy is internal consistency: do the measurements inside the map agree with each other? If the pad is 120 ft wide in the field, is it 120 ft on the map? Conner Jones of DroneDeploy describes this as the level that's sufficient for single-date measurements and visual inspection — you're comparing things to other things in the same flight.

Absolute accuracy is alignment with the real world: is that pad corner at the coordinates the survey says it's at? Jones flags this as the requirement for earthworks analysis and for comparing datasets over time — and that second part is the one PMs underestimate. If two flights a week apart are each a foot off in different directions, the "change" between them is noise. Cut/fill against a design surface, progress volumes, and design overlays all live or die on absolute accuracy.

Crawl, walk, run

Jones frames the path to accurate mapping as three stages, and it maps neatly onto how most contractors actually adopt the technology.

• Crawl. An off-the-shelf drone, no ground control, standard onboard GPS. You get a good-looking ortho with solid relative accuracy; absolute position is typically off by a metre or more. That's fine for weekly progress photos, marketing, and measuring a stockpile's footprint on a given day.
• Walk. Add ground control points, RTK positioning, or both. Absolute accuracy drops to the centimetre range, which is what unlocks volumes, multi-date comparison, and design conformance checks. This is the stage where Jones suggests identifying which workflows gain real business value from high accuracy — not every flight needs it.
• Run. Make it repeatable: standard flight plans, consistent control, independent checkpoints, and QA on every dataset, then roll the same recipe across sites. The win here is less about any single map and more about being able to trust a year of them.

What it's worth

The business case Jones lays out is the one we see on every site: measured once correctly means fewer surprises and less rework, fewer people walking active work areas with a rover or a tape, and decisions made from a faithful digital copy of the site rather than from memory and a few photos. Accuracy is what turns a drone from a camera into an instrument.