Point-cloud classification looks like a commodity until the deliverables arrive. These ten questions separate production-grade vendors from resume-grade ones — before the contract is signed.
Why LiDAR procurement goes wrong
LiDAR classification bids often differ by 3× on price for what reads like the same scope. The difference hides in the things the cheap bid does not say: how much manual editing follows the automated pass, who checks the work, and what happens when tiles fail your acceptance criteria.
The buyer’s real risk is not price — it is a season lost to redelivery cycles. Every question below is aimed at that risk.
Questions one to five: capability
Ask for per-class accuracy on a comparable past project, not overall accuracy — ground and building classes can look excellent while conductor or low-vegetation classes fail. Ask what share of effort is manual refinement; honest answers run 30–60% for demanding specs. Ask which software stacks they run and whether your delivery spec (LAS 1.4 point source, classification flags, tiling) is native to their pipeline or a conversion afterthought.
Ask how they handle your projection and geoid — vertical datum mistakes are the most expensive class of LiDAR error. And ask for their tile-level QC dashboard: a vendor who cannot show you live per-tile status is managing by spreadsheet and hope.
Questions six to ten: production reality
Ask about surge capacity: what happens to your schedule when a bigger client’s emergency lands. Ask who performs QA and whether they report to production management or independently — the org chart predicts the accuracy report. Ask for the redelivery policy in writing: what error rate triggers free rework, measured how, on what sample.
Ask about data security concretely — where the point cloud physically sits, who can access it, what deletion certification looks like. Finally, ask for a paid pilot on your own data with your acceptance criteria. A vendor confident in their pipeline will price a pilot cheaply; a vendor who resists a pilot is telling you something.
The pilot is the interview
Structure the pilot as a rehearsal of the real contract: same spec, same delivery format, same acceptance test, on tiles you select for difficulty rather than convenience. Score the communication as much as the data — clarification questions asked early are the best predictor of production quality.
GlobeIR answers all ten questions in writing as a standard part of any LiDAR proposal, and prices pilots against your hardest tiles deliberately. Terabyte-scale classification is our production floor, not our stretch goal.