Once you have established that details are desired or required (see the page on when to submit details), consider all these methods of documentation:
Objective forms of evidence, such as photographs or videos, are superior to sketches or descriptions. However, we realize that few people like to drag the extra equipment along, so in practice, written descriptions are used most often.
Writing a good description is one of the most important skills of a competent birder. At the very least you should know which field marks to look for, but don't stop there. The rarer the bird, and the more prone it is to misidentification, the more you will need to describe everything you can---not just plumage details, but also soft part colors, shape, posture, behavior, locality, microhabitat, and vocalizations. Describe also the circumstances of the observation---duration, distance, optical aids, amount and angle of available light. Discuss what other species should be considered as candidates for the ID, and which field marks you used to eliminate them.
The length and amount of detail should fit the rarity of the species and the difficulty of identifying it in the field. For example, if you saw a female Red-breasted Merganser on a lake where they have often occurred in the past, it is probably sufficient to mention the lack of contrast between throat and breast colors, although it wouldn't hurt to discuss bill size and shape as well. For a first state record or something almost that rare, though, several pages of details are not excessive. For example, John Trochet's description of a female Rose-breasted Grosbeak on the 99th Zuñi CBC was three pages long---quite appropriate considering that it was a first CBC record for the state, and a rather subtle ID problem as well.
Here are some things that I see a lot that do not constitute good descriptions:
If we ask for more details, please don't take it personally. Our motivation is to make life easier for researchers in the future. They won't know any of the people who made the records, so all they have to go by is the actual details.
The basic rule here is to be as objective as possible. Describe exactly what the observer saw. It is better to say ``the bird had a white throat patch bordered by thin black lines'' than to say ``the bird had the usual throat pattern of the White-throated Sparrow.''