We were recently asked to review a university web site that uses a CMS system. Like many large complex sites it was really a collection of separate sites and content sections. Looking closely you could almost see the evolution of the site from a manageable and unified tree of usable information to an unwieldy road map of confusion.
Did the Site outgrow it's CMS? It's possible, but its more likely that the site outgrew the process that was used to manage the content. The process used to update a large, content-driven web site is as important as the chosen tools or system.
While reviewing we identified 3 problem areas on a typical "Bulked up" content driven site. Each one of them can be at least partially alleviated by implementing standards, a work-flow and some general rules regarding publishing - in other words, a better and more productive process for managing content. The areas are:
A CMS system usually has a Tree as a heart beat. The tree can be described as levels or sections of the site. The tree is expressed as the menu system. A user drills down into the tree by clicking on a link which takes him or her to a page with other links and so on. Take, for example a real estate site. To get to a listing a user might have to do this:
On a site with hundreds and possibly thousands of pages (i.e. a university) this can result in so many levels that the user quickly loses track of where they are on the site. If no one is watching the ship it will quickly evolve into a rats nest of links and menu systems with unnecessary duplication, loop backs, orphaned pages and my favorite - the roach motel page. The roach motel page is the one you can get into but for some inexplicable reason you can't seem to get out. Usually because of redirection instructions issued at the browser - meaning the back button doesn't work as expected - and no bread crumbs or links back to the previous page. What's the solution to navigation disfunction? Is there a Navagra or Cialink for a rats nest of links and levels?
First, have a plan for navigation. Deciding on a new level or link should require forethought and approval from someone who has a grasp of the scope of the whole web site. It should not be left to individual content creators. An approval process will help. Content creators submit new content to a queue for approval where someone with a broader view can edit and publish it.
Secondly, once you have a logical navigation tree, use bread crumbs to help the user get a sense of "where" they are at. Probably no site on the web has quite as much content as eBay - right? Yet, when you are searching for something on eBay you always have access to categories and sub categories - both forward and backward. You always have a sense that you can broaden or narrow your search for information. That's good planning, but it takes forethought and a pre-conceived structure that anticipates content growth.
Finally, have a consistent and intuitive menu scheme. When a user first examines your site the first thing he or she will try to determine is Where are the main sections. I always look for a "menu bar" or "link list" somewhere that (theoretically) shows me where I can go on the site. If that main menu appears at the top and I click on an item, I do not expect it to suddenly shift to the left sidebar. If the menu is on the left hand sidebar and I click on something and new items appear, I do not expect the main links to disappear. I expect to be able to get to the main sections of the site from wherever I am at in the site. I also do not expect "hidden" sections to suddenly appear from clicking on an obscure link. This is a major problem with large content sites who do not have adequate controls on content publishing. In the case of university there were at least 5 whole "sections" that should have had a main link - or possibly be rolled into another section. Someone with a great idea for a university program and access to the publishing tools simply created what amounts to a separate "site" within the site and put links to it inside of other content - rather than a process determining the best way to handle it.
Here's a question for your content creators. What fonts should they use. If they have access to a WYSIWYG editor chances are it's up to them. The engineering department uses fixedsys while the art department uses wingdings. How about images? Can content creators upload them and link to them in a page? If so, are there any controls on size? Is the photography department loading pages filled with 500 k photos?
Without a defined policy in place for the "look" of the site the content will become a panorama of styles. The solution here is straightforward but not simple. Create and enforce a set of styles. Make style and standard checking part of the process of approval for getting content published. If you don't you will have the pleasure of seeing the personal tastes of everyone's favorite document style.
In our recent review I went into the faculty directory. While there I saw a link to "current news". I clicked on it and found a fairly decent list of stories with teasers and captions nicely formatted - and 3 years old. The youngest story was from June 2002. Again, here's a case of someone not watching the ship. Content should be allowed to expire. On a typical content-driven site about 60% of the stuff you find is usable and the rest needs to be retired. Old content results in bad search results, missing links to dead sites, mis-information or false information - a whole host of problems.
Implement a review policy. Make sure someone is watching out for content that is old, out of date or in need of correction. Sometimes the best person for this is not a webmaster. Often it is someone for whom the actual information is the most important and the web site itself is merely a vehicle for disseminating that information.
Use dates and have an expiration policy. If you are using a database driven CMS tool this is easier to implement than in a document driven CMS tool. Not all content should be time sensitive - but at least some things should be marked for expiration. It bugs me to go to a site and see press releases from 2001 in the media kit. Still, when I go to a support site I appreciate being able to find obscure manuals for a motherboard or disk drive. The point is that in regard to content relevance there should be a standard that makes sense.
Update your search collections. It may seem like something that should be so obvious as to be left unsaid, but many developers know how to add new documents or content to the search, but fewer of them seem to know how to remove expired documents from a search index or collection. If you are using verity, remember to appropriately update the collection index to remove expired content. It's another pet peeve of mine to search for something on a web site and be presented with old or broken links.