You have two quotes on your desk for the same site. One is $900. One is $2,800. Both say “aerial survey,” both promise an orthomosaic and a report, and both vendors sound confident on the phone.
The proposals do not tell you which one is which. The questions below do. None of them require you to know photogrammetry — they only require you to notice whether the answer contains a number, a standard, or neither.
1. “What accuracy class will the deliverable cite?”
This is the one question that sorts the field fastest.
The answer you want names ASPRS 2024 (Edition 2, v2.0) and states a class — a horizontal and vertical RMSE the product actually meets. The answer you do not want is “survey-grade,” “centimeter-level,” or “we use RTK.” Those are equipment descriptions, not accuracy statements. RTK on the aircraft tells you how the camera positions were solved. It tells you nothing about the error in the finished ortho.
A vendor who cites a class is making a claim you can test. A vendor who says “highly accurate” is making a claim that cannot fail, because it means nothing. If the deliverable is going under a CAD drawing or into a quantity takeoff, you need the first kind.
2. “How is that class verified?”
The follow-up matters as much as the answer. Ask what checkpoints were shot and whether they were held out of the bundle adjustment.
Checkpoints fed into the adjustment prove nothing — they tell you the math agrees with itself. Held out, they measure real error against independent ground truth. That is the difference between a stated accuracy and a verified one, and it is a distinction a competent operator will be glad you asked about.
If the answer is that the software reported the accuracy, you are buying the processor’s self-assessment.
3. “Which standard governs the inspection itself?”
Positional accuracy is one axis. If you are buying a condition assessment rather than a map, there is a second one, and it is standard-specific:
- Infrared moisture surveys on roofing systems — ASTM C1153-23
- Building envelope thermographic inspection — ASTM C1060-23
- Photovoltaic array thermography — IEC 62446-3
Each of these specifies conditions the imagery has to be captured under for the results to mean anything: thermal loading, time window, wind, moisture, equipment class, and what the write-up owes you. A flight that ignores those conditions can still produce orange-and-purple pictures. It cannot produce a finding you would want to hand a roofing consultant.
Ask which standard applies to your scope, and ask the vendor to name it back to you unprompted.
4. “Who is flying, and under what authority?”
The floor is FAA Part 107 — a certificated remote pilot in command, and airspace authorization where the site requires it. Much of Northeast Florida sits under controlled airspace shelves; if your site does, the operator needs LAANC authorization or a waiver in hand before the day of the flight, not a plan to figure it out on arrival.
Then ask about insurance limits and whether the certificate names your project. It is a routine request. A vendor who treats it as an imposition is telling you something.
5. “What files do I get, and in what coordinate system?”
“A report and some photos” is not a deliverable spec. Ask for the list, by format:
- Orthomosaic as GeoTIFF, with the coordinate reference system stated
- Point cloud as LAS/LAZ, if you are doing volumes or grading review
- Mesh as OBJ or equivalent, if massing or sections matter
- A written report identifying findings by location, not by “northeast corner”
The coordinate reference system is the part people forget. A GeoTIFF that does not say what frame it is in cannot be dropped under your civil drawing without someone guessing, and guessing is how a 2-foot shift enters your as-built. State plane, datum, units — get it in writing before the flight, not after.
Also ask who owns the data and whether you get the raw imagery. On a progress-monitoring program, that answer determines whether you can switch vendors in month eight without losing your baseline.
6. “If this ends up in a claim, does the file survive?”
Most aerial work is bought for planning and ends up in a dispute. Delay claims, damage claims, scope arguments, and insurance filings all eventually ask the same question of your exhibit: what is this file, and where did it come from?
Under FRE 901/902, evidence is admissible when it can be authenticated — when someone can establish the record is what it purports to be. For aerial data that means the chain exists: flight logs tying imagery to a specific sortie and date, the correction source and control layout, the held-out checkpoint results behind the stated accuracy, retained original frames with intact metadata, and processing lineage from raw capture to finished product.
That chain has to be built at capture. It cannot be reconstructed eighteen months later when the claim is filed, which is the moment everyone discovers they needed it.
What the price gap usually is
Back to the two quotes.
The $900 flight is often a real flight. The gap is rarely the aircraft — it is the ground control, the held-out checkpoints, the standard-specific capture conditions, the processing time, the report, and the records retention behind all of it. Those are hours, and hours are what gets cut to reach a number.
That is not automatically the wrong purchase. If you want a marketing flyover or a look at where the crew parked the excavator, buy the $900 flight and be happy. The failure mode is buying the $900 flight for a job that will later be measured off, stamped, or filed — and discovering the deliverable cannot carry the weight you put on it.
Match the rigor to the use. The questions above are how you tell which one you are being sold.
Takeaway
Ask for a number, a standard, and a chain of custody. A vendor who can give you all three in one phone call is selling a record. A vendor who answers all three with adjectives is selling pictures — and pictures are fine, right up until the day they aren’t.