Top 3 common rollout mistakes that spell "failure" in any UCM project

These are more fundamental mistakes than security flaws and bad hardware choices. I'm talking project killers, things that are sure to take your project under. And the worst thing of all is when your new Oracle UCM is not accepted by your organization. If that happens — your project is finished! When people don't like your system, they won't be using it. When they are not using the system, no one can benefit from it. End of story.

Now, let's look a little deeper. What does it take to get people to use it? Let's consider the requirements.

Integration into organization culture

Oracle UCM just like SharePoint, Vignette, BroadVision, you name it, is just a framework. All it does is allow you, the designer, to quickly and easily automate business processes. Eliminate paper. Help them find the information faster. Destroy when it has to be destroyed.

You need to adjust your content collection, management, and presentation systems to the processes and demands of your organization. That's the real goal, not the software installation!

So guess what comes first?

If you misunderstand your organization's culture, processes, and business requirements, no system out there will save your project.

Take the time to understand the real needs of business people. Understand the goals of your enterprise. Then go ahead and automate them. You won't be stuck trying to automate a bad process! (Be sure to read about Iterative Development later in this chapter.)

Another fatal mistake commonly found in failed UCM rollouts is bad Metadata design.

Metadata design

Once you've identified your business clients and spent time understanding their needs, you're ready to reflect your knowledge in a content management system. And you do this through metadata design and workflows.

As you've seen earlier in this chapter, metadata design is all about creating custom fields. Fields that describe various types of content you'll be storing in the system.

For example, if you store Project Plans, you might like to add a project manager's name, completion date, and expense account number.

And remember, if users don't understand the purpose of a metadata field, they'll skip it, put wrong stuff in there, or just type up some garbage, so the system lets them submit the form.

Here are some tips for successful metadata design, expensive to ignore.

Don't ask for all of your fields every time

If they check in an employee record and you're asking them for PM name, guess what kind of name you'll get. It must make sense. And it also must be as quick and painless as possible. Business people are busy and many perceive content management as a "necessary evil", even "pain in the neck", so if you don't have to ask them for a specific field value don't.

Now if you don't need a user's input on a specific field, consider hiding it. On the other hand, if you do need their input, consider using validation.

Another great thing you can do is take advantage of features that let you pre-populate metadata values, features such as Content Profiles (covered in detail in Chapter 4), Content categorizer (see Appendix A), or the Multi-Field Configuration Utility from Fishbowl Solutions (www.FishbowlSolutions.com/StellentSolutions).

Beware of the snowball effect

This is what happens when you ignore the previous rule.

Picture yourself having a nicely configured system with clean data. Every document has correct meta values. Now you begin asking for a field that doesn't make sense.

Your contributors will begin putting garbage there and they'll be putting garbage in other fields as well! How?

Once you have put garbage in one field, say 'asdf', it becomes very easy to simply put that in the rest of the fields all the way down the form. The system will soon become polluted and you might end up losing the meta values for your entire repository. Why?

You won't know which records have the real values and which just garbage.

Soon enough you won't be able to trust the data anymore!

Keep an eye on what content is getting entered and always keep communicating with your business users.

Know this single, most expensive type of UCM activity

The single most expensive UCM activity is manual Cleansing and Matching. Unless you have an army of data analysts looking for stuff to do, I don't recommend letting your project get to this point. You must prevent it almost at any cost.

And here is one more common way organizations kill their content management projects. It's by ignoring the principles of iterative development.

Iterative development

You can still afford to delay the rollout until you have all the knowledge you need to design your new system. It doesn't have to be chunked into large and expensive phases — expensive and vulnerable to scope creep.

You can easily implement a useful system while you're still acquiring knowledge. The key here is to have a live working system in the hands of your business people as soon as you possibly can.

All you need to do is make small incremental changes such as adding different types of documents, new modules, new workflows, and so on.

And listen to feedback. If business is happy you did great. If not all you have to undo is just a small change! So don't try to bite off more than you can chew. (If you need more info on agile and iterative development processes, Wikipedia (www.wikipedia.org) is a good place to start. Just search for "Agile software development". Congratulations!