I have spent a lot of time on docks and in yards with teams who know detention is eating them and still cannot say, from the systems alone, when work on a trailer actually started and finished.
What I find interesting is how often the conversation gets framed as a safety or yard-camera problem first. Detention is not a safety metric. It is an ops P&L line. Trailer dwell, empty bays while appointments stack, freight that clears the dock and then waits in staging. Those burn money and labor minutes whether or not anyone was wearing the right PPE.
What detention looks like on a shift
In plain terms, detention shows up when a trailer is at your site longer than the agreement allows, and you cannot prove the work happened in the window you thought it did.
The appointment said the bay was ready. The WMS said the load was available. The yard board said the trailer was spotted. None of that tells you whether a crew was on that trailer, when the first pallet moved, or how long the bay sat empty while the next appointment waited outside.
So the carrier invoices. Ops argues from incomplete timestamps. The site GM gets the bill and a story that does not quite hold.
That is not a camera-company problem. That is a floor-truth problem on a workflow that already has an owner.
The questions that actually matter
If you run a DC, a plant shipping dock, or a network with carrier agreements, the useful questions are operational:
- When did dock work on this trailer actually start and finish?
- How long was the bay empty while the next appointment waited?
- Where did the trailer sit in the yard before it was spotted to a door?
- Where did freight sit between dock and putaway?
- Which handoff lost an hour between shifts?
Those are the questions that decide whether detention is a mystery or a managed workflow.
Why systems of record are not enough
Appointments and WMS timestamps are necessary. They are not floor truth. The gap between the planned dock event and what actually happened on the concrete is where detention hides.
Vision is often how the first signal arrives for this workflow, because docks and yards already have cameras. The wedge is mid-piece: turn existing video into start/finish, dwell, and queue signal for the person who owns detention on that site. Not a camera pitch. An ops workflow.
What done looks like
Done is not more footage in a library. Done is a named owner for detention on that site, a definition of done the site GM would recognize (shorter dwell, fewer empty-bay minutes, fewer surprise invoices), and a signal that reaches the shift while there is still time to act.
If you cannot name the owner and the done condition, you do not have a production target yet.
Detention is plain when you can see the work. Start with one dock or yard workflow, put an owner on it, and expand only after the first one holds.