14:10, the complaint
The client stood three metres back and pointed at the vertical seam between columns two and three. It was visibly wider at the top than at the bottom. On a 3.5 mm bezel wall, a 1.8 mm variation is enough to read as a mistake.
14:25, measure first
We measured the gap at nine heights along that seam. The gap was 5.3 mm at the bottom and 7.1 mm at the top. The other seams were 5.3 mm throughout. So the panel on the right was leaning, not the frame.
14:40, check the obvious
Panel seated on its brackets? Yes. Bracket screws tight? Yes. Then we put a straight edge across the face of the three panels in the bottom row. They were coplanar within 0.3 mm. The problem was vertical, not depth.
15:05, the actual cause
The mounting bracket on the top-right position had a burr on the locating pin, about 0.6 mm high. The panel sat on the burr instead of the shoulder. The error was multiplied by the distance from the pin to the edge, which is why the gap opened by 1.8 mm at the top.
15:40, the fix
File the burr, reseat the panel, re-measure. The seam closed to 5.3 mm at all nine heights. Total time on the burr: four minutes. Total time finding it: an hour and a half.
What we changed afterwards
Every bracket now gets a pin check with a straight edge before it goes into the crate. It added twelve seconds per bracket and removed a whole class of site problems.
For the client's warranty file
We photographed the seam at nine heights before and after. If the seam drifts in six months, that record tells us whether it moved or was never right.