We get it. You are four states away, you have a site to make a decision about, and you have Street View, satellite imagery and an AI that will describe the property back to you in confident paragraphs. It is free, it is instant, and most of the time it is close enough. Until the part that matters is underground, at which point every one of those tools is guessing and none of them will tell you so.
This is not an argument that the tools are useless. They are genuinely good at several things and we use them too. It is an argument about which decisions they can carry and which they cannot.
What remote tools are actually good at
Worth saying first, because a piece that pretends otherwise is not worth reading.
They are good at narrowing. Ruling out sites that are obviously wrong, checking whether access physically exists, measuring a parking lot or a roof, seeing how a property sits relative to a road, getting a rough sense of scale before anybody spends money. For that work they are faster and cheaper than a person and it is not close. A satellite takeoff will measure a lot better than somebody with a wheel.
They are also good at preparing a visit. Turning up already knowing the layout is better than turning up cold.
Where they stop, and why it is always the expensive part
They cannot see under anything. Not under pavement, not under grass, not under snow. Which is unfortunate, because the underground is where the money is on almost every site problem worth asking about.
The image has a date on it, and it is usually not recent. Go and look at the date on the Street View image of a property you know well. That is what an assessment based on it is describing: the site as it was on that day. Curbs get broken, drainage gets regraded, buildings get extended, and a photograph from three years ago describes none of it.
They cannot tell failing from ugly. A patched lot and a failing lot look identical from above. So do a lawn that is soft because it rained and a lawn that is soft because a water service has been leaking into it for five months.
They flatten grade. Slope is most of what decides a drainage problem and it is the thing overhead imagery represents worst.
And AI will answer anyway. This is the part worth being blunt about. Ask a model what is under a driveway and it will not say "I cannot see that." It will produce a fluent, plausible, well-organised answer built from what is usually true. Usually true is a good bet and a poor foundation. The failure mode is not silence, it is confidence.
Two jobs where the ground disagreed with the picture
Bridgewater: a driveway that looked like a routine bore
Every remote signal said this was straightforward. A driveway, a house, a run to get under it. Any assessment made from imagery would have called it a standard bore, and so did we, which is why we tried.
Five attempts. A shale ledge in the way each time. The last attempt started at five and a half feet down and the head still surfaced at sixteen inches, because rock does not care what the plan said. The driveway got excavated and open cut in the end.
Nothing available from a screen would have found that ledge. Not one photograph, not one measurement, not one model. It took a drill hitting it, five times. The full account is in why we could not bore this driveway.
Andover: the difference between somewhere and here
A five hundred foot water service that had been leaking for five months. What made it visible at all was weather: a drought came, everything else browned off, and one strip of lawn stayed soft.
But a wet patch tells you there is a leak somewhere in five hundred feet. It does not tell you where, and it frequently is not above the break, because water follows the trench or a stone seam and surfaces wherever it can. We put leak detection equipment on it and located the actual area, dug one hole, and found a one inch poly line four feet down, cracked on the underside, in a service the house believes is copper.
The spot repair was $4,750. Trenching five hundred feet to look for it would not have been. That is the entire economic argument for having somebody competent on site with equipment, in one number, and the write-up is in a 500 foot water line in Andover.
The asymmetry nobody does the arithmetic on
| Decision | What guessing costs when it is wrong |
|---|---|
| Is this a bore or an open cut? | Mobilising a drill for attempts that cannot work |
| Did the sub actually do the work? | Paying an invoice for something that is not there |
| Where is the leak? | The difference between one hole and five hundred feet of trench |
| Is the restoration going to hold? | Finding out next spring, after the retainage is gone |
| Is that change order fair? | Approving it, or fighting it, both without facts |
A visit starts at $150. Put that against any line in the right-hand column and the arithmetic is not difficult. The reason it does not get done is not cost, it is that sending somebody feels like a hassle and looking at a screen does not.
The honest rule
Use the remote tools for what they are good at. Narrow the list, check access, measure the lot, arrive prepared. Then, before money moves on the strength of it, have somebody stand on the ground.
Not somebody with a phone. Somebody who has done the work and can tell you the trench was not bedded, the ledge is at four feet, the restoration will sink over a winter, and the wet patch is not above the break. That is a different service and it is the one we sell: third-party site verification in NJ, PA and NY.
Common questions
Can I use Google Street View to assess a construction site? For some things, yes, and it is a sensible first step. It will tell you whether access exists, roughly how a site sits, and what the frontage looks like. What it will not tell you is anything below the surface, anything about grade with much accuracy, or anything that has changed since the image was captured. Check the date on the image before you rely on it; it is displayed, and it is often years old.
Can AI tell me what is under a site? No, and the risk is that it will answer confidently anyway. A model has no access to what is beneath a specific property and will assemble a plausible answer from what is generally true. That is useful for framing a question and dangerous for settling one. It will not tell you it is guessing.
Is a site visit worth it for a small job? Compare it to the cost of being wrong rather than to the cost of the visit. A visit starting at $150 against a mobilised drill, a wrongly approved payment, or a trench dug in the wrong place is not a close call. Where the visit is not worth it is when nothing expensive turns on the answer, and that is a real category.
What can somebody on site tell me that a photograph cannot? Whether something is failing or merely looks bad. What the grade actually does with water. What is under the surface where it can be seen at an edge, a joint or an open section. Whether the work matches the plan. And what has changed since whatever imagery you were looking at, which on any active site is most of the point.
Still not sure how this applies to your property? That is what a free on-site estimate is for. Call 862-268-2510 or tell us what is going on, and we will give you a straight answer, including when the answer is that you do not need us.
Sources and further reading
- Why we could not bore this driveway, the Bridgewater job in full
- A 500 foot water line in Andover, including the numbers
- Third-party construction inspection and site verification, what a visit involves and what comes back
Written by an excavation contractor who uses satellite imagery most weeks and would not price a job from it.