Every field video product has the same moment sooner or later. You are on the third slope, mid-sentence, and the connection goes. What happens in the next ninety seconds decides whether that inspection was interrupted or ruined.
Most of the damage is not technical. It is that somebody in a driveway, or at a desk forty miles away, just watched their window into your work go black and does not know whether to wait, call, or give up.
What a drop actually is
Cellular coverage is not a wall you step past. It is a gradient with holes in it, and the holes move. A chimney between you and the tower, your own body between the phone and the sky, a neighbor's roofline, an air handler kicking on and pulling the local network around — any of it can take upload capacity below what live video needs while your phone still shows bars.
Bars measure the downlink. Video needs the uplink, and the uplink is the only number that matters on a roof. That mismatch is why a drop so often feels like it came out of nowhere.
Well-built streaming does not fail all at once. It steps down first — resolution drops, the picture softens, motion gets mushy — and it holds a degraded connection for as long as there is anything to hold. A hard drop means the ladder ran out of rungs.
The recording keeps going
The important thing to know, and to tell people, is that the recording and the live view are not the same pipe.
Recording happens on the session, not on the viewer's connection. A viewer losing their feed does not stop the record. Your own connection dropping interrupts the capture of that stretch — you cannot record what never left the phone — but everything up to that point is already banked, and it does not evaporate because the last thirty seconds were lost.
So the practical rule is: the walk is not wasted. Whatever you had established before the drop is established. What you need to decide is only what to do about the part you were in the middle of.
The ninety seconds after
What I do, in order, and it is worth having as a habit rather than a reaction:
- Stop walking. Do not keep narrating into a dead connection. You will have to redo it and you will be less careful the second time.
- Move, do not fiddle. Ten feet back toward where it last worked beats two minutes in settings. Get off the far side of the chimney, stand up, get the phone away from your body.
- Text, do not call. "Lost signal for a second, still on the roof, coming back." Fifteen seconds of your time, and it converts a black screen from abandonment into a pause.
- Resume where you left the surface, not where you left off talking. Back up one landmark so the viewer can re-anchor.
- Say what happened, on camera, once you are live again. "We dropped for about a minute on the north slope. Nothing changed on the roof. Picking back up at the vent."
That last one matters more than it sounds. A gap in a record with no explanation is the kind of thing that gets asked about later. A gap with a sentence attached is just a gap.
When to stop instead of fight it
Two drops in the same spot is information, not bad luck. Three is a decision.
If the connection will not hold, switching the job is better than half-documenting it. Record locally and walk the homeowner through it afterward, or reschedule the live portion and complete the physical inspection now. Both are honest outcomes. What is not honest is a record that limps through with two-thirds of the slopes covered in a picture nobody could evaluate, because the alternative felt like an admission.
The judgment about when not to go live at all is the same judgment, made earlier and more cheaply.
None of this is exotic. Field connections fail; the difference between operations is entirely in whether the failure was handled as a procedure or as a surprise. If you want to see how the session and the recording behave when the network gets ugly, how it works covers the mechanics, and running a real session on a real job is the only test that means anything — every plan starts with a free week for exactly that reason.
InspectStream records what an inspector observed. It is not a public adjusting service and does not prepare, negotiate, or advise on insurance claims.