I agree with most of the great comments above. These are a few I want to add to, and some that I didn't see brought up that I think are important:
- If the module is going to be large (for example, a long event based or large dungeon crawl) please provide easy methods to take breaks in the action. Some play groups like my own may only be able to get in 2 to 3 hours in a single session, and it's much easier to handle this if such considerations are written into the module. Most large dungeon crawls tend to take far longer than this to complete. Related to this, it may be that the same cast of characters can't arrive for each session, so having a method to explain characters sudden absense or arrival is incredibly useful as well.
- Give the party a very good reason for tackling this adventure. Involve them somehow or offer a decent enough reward. Don't just assume that a party will simply be motivated to help those in need and jump right into your adventure. Not everyone plays good characters, and in many cases they may not care about the events unless they are effected directly by them.
- Truly flesh out the NPC's. IMHO, the strength of an adventure is most limited by 2 factors - the plot and its NPC's. Strongly developed NPC's can make up for some weakness in the plot, and the converse is also true to some degree. However, I find it much easier to run well-developed NPC's as the DM, rather than try to keep a story interesting when the NPC's are bland or not even developed at all. If the NPC is going to talk to the party, then provide a list of suggested responses for various questions. This can greatly help to lessen the burden on the DM, as well as provide more clues as to their personality.
- Include commonly used skill check info results related to the adventure. Examples would include Gather Information rumors that can be found (have a list of infos with a DC and the appropriate fact or rumor).
- Keep the adventure as modular as possible. This often requires you to keep certain pieces of information rather simple or basic. For example, which god you choose should likely be listed as generically as possible. Any location history should be written to make it easy to drop into any world. If you keep a history to a much more local basis and avoid grand events such as wide-sweeping events it helps to allow the module to be dropped in anywhere. For example, a history that mentions a local baron's past is ok, as is a localized disaster such as a forest fire or a drought - each of those can usually be worked in fairly easy to any campaign setting. But histories that require past or present events like continental wars may not be very easy to shoehorn into a campaign.
- Keep the EL's reasonable. Too many of the modules I've seen have unrealistic EL's that are often way too high for the suggested party level. So don't go throwing an EL12 finale into your module suggested for characters level 5.
- Along that same line, don't classify the suggested character level with a broad range. There's a major difference between even a couple levels of characters, especially at lower levels. So instead of classifying an adventure as being suggested for levels 3 to 6, be more specific and say the level you really intend it for (for, example level 5). Obviously, different parties will handle it differently, but if you are consistent in your suggested levels, DM's will learn to adjust this. The broad range method makes this much harder to diagnose.
- Assume that characters are going to do unexpected things, and take unexpected paths in the environment. If your map has 1 primary and intended way in, but also has a secret back way in that is only intended for the villains, assume that a smart party may uncover this and use this path instead. Also, assume that a party may find yet another alternate path using magical spells such as stoneshape to entirely bypass the expected route.
- Remember there are 3-dimensions and many spells allow you to act in them. Spells like Spiderclimb, Levitate or Flying, or druid shapechange abilities may totally foil a well-planned 2-dimensional trap or pursuit.
- Avoid "Save the World" styles of adventures. It's fun to run these once in a while, but after you've done it once, they tend to lose their appeal. A much better style is just to have the group "Save their own hides", which tends to be a lot more interesting.
- Finally, balance the risk with the rewards. Try to follow the suggested tables in the DMG. If the adventure is rather easy, don't give out an over-abundance of treasure. On the other hand, if the party will need to overcome tremendous odds to survive, then by all means, award them well. Such rewards can include magic, but even large sums of gold or other mundane items (property deeds, a few crates full of masterwork tools, etc) can be just as rewarding, and often more believeable.