Trace the problem. Repair the right detail.

Many problems confined to one area can be repaired, but widespread wear, repeated failures, unavailable matching materials, or hidden substrate issues can change what is technically sensible.

Only far enough to tell you whether it is worth a visit. A photograph shows one plane from one angle, in one light, and it cannot show the deck underneath, the condition of the flashings, or how far a problem runs past the edge of the frame. Send them by all means — they are useful for working out how quickly somebody needs to come. A price that survives contact with the roof still needs somebody on it.

Almost always something that was not visible until the covering came off: soft decking, a flashing that turns out to have been sealed rather than built, an extra layer nobody knew was there, or a second entry point that only appears once the first is opened. The honest way to handle that is to name it in advance — what triggers a change, who authorises it, how it is priced by unit and how it is documented. A repair scope with no change rule is not cheaper, it is undecided.

You do, from a scope that actually says where. A repair has a boundary, and the useful ones state it: these planes, this detail, these components, and — explicitly — what is not included. Where that boundary is left vague the argument arrives later, because everything outside it looks in hindsight as though it should have been covered. Ask for the exclusions in writing alongside the inclusions. They are the more informative half of the document.