authors to locate, create and manage associations and links among parts of
content. On the web, links,
tags that point elsewhere in the content, are everything. Rich content must contain rich links if
it is to fulfill its intended delivery mission.Without support, links can be a nightmare to author and are
notorious for breaking somewhere between creation and delivery. A major publisher found that when
authors are forced to break their train of thought to create a link, fewer
links result despite their best efforts to find and create them.
authors to create content suitable for multiple audiences. In a multimedia world, content must
be delivered in slightly different form to each segment of the total
audience. Whether different
delivery media, user interest or skill levels, product variations or what
have you, every content provider faces a layered audience that wants even
common content tuned to its unique needs.
Your ability to support this kind of targeting is closely linked to
the way you decide to manage your content and must be a consideration in
selecting the CM software you will use.
collaboration and communication among knowledge providers. Rich content passes through many
hands on its way to delivery. Add
the pressure of deadlines and the situation can degenerate into
chaos. Any CM system must support
collaboration among actors within and outside the primary content
site.Many vendors address this
challenge with classical workflow software designed to support
claims processing or other highly structured applications.This is usually too restrictive for
professional knowledge workers and can be expensive out of proportion to
its value. The answer is a CM
environment based on individual decisions by each actor as to what the
next step in content preparation should be. While a system might set certain limits on whom could do
what under what circumstances, laying out rigid workflow paths ahead of
time almost always degrades the productivity of the intellectual process.
and track revisions to content at a granular level. As information moves ever faster,
knowing how it got to its current state becomes a growing factor in
satisfying your audience. One
approach to this has been to keep each version of a document in work, comparing
each version to the previous to determine what has changed. This approach leaves much to be
desired, like not being able to find the changes without running the
compare program. Moreover, the
comparison method doesn’t collect information about who made changes and
for what reason. A better approach
is to make content revision a specific transaction and support it with
tools and tagging. Revision
tracking can add much to the content management process. Some publishers, for example, tie
revisions to specific projects, turning them on selectively as events
possible, keep content in its original state. If you’re creating XML content, the best approach is to
keep the XML in its originally authored format until time for
delivery. Some CM approaches
translate the content into a proprietary (often database) format, and then
retrace the process to extract the XML.
This can be made to work somewhat but suffers from the fact that
rich content structures tend to have problems when theyre mangled and
glued back together again.
Moreover, the transformation is based on complex computer
processing so these systems become resource hogs as soon as the volume and
complexity of the content grows.
portions of content in multiple places but keep the ownership
centralized. Frankly, beyond
sharing inherently common content like warnings and boilerplate, the
entire concept of “reuse” is somewhat overblown. You may find that content authored for one place in your
collection is difficult to lift and use elsewhere without at least some
modification. Change even one
character in order to reuse content and you aren’t really reusing but
paraphrasing. Some vendors suggest
that you deal with this by breaking your content down into ever-smaller
pieces so you can collect them in different ways for reuse. Beware; the
complexity and risk grows in direct proportion to your usage.
the content safe and under control.
I list this last because everyone knows that controlling access
and data integrity is part of CM.
Vendors, especially those from the database world, often lead with
this function and construct their demonstrations around it (given that
authors spend, on average, only 5 percent of their time in the CM system,
one might wonder what this adds.)
In truth, most do the basics well, making it less of a
differentiator than a ticket punch.
The important variable in this area is the extent to which a system
must “decompose” rich content in order to manage it. Content richly tagged in XML (or SGML)
often contains structures simply not capable of being snipped and laid end
to end. Systems that require this
kind of content fragmentation limit the richness of content they can fully
support and are likely to encounter technical problems.
Dealing with the Vendors:
Asking a software vendor if he supports CM is like asking a
user car salesman if he has any cars for $5,000. Of course he does.
Instead, send your function list and a description of your overall
environment to candidate vendors, with the question, “Can your product/s
properly support these functions and, if so, tell me how in detail.” Because reputable vendors will opt out if
they can’t comply, this will help to weed out the fellows who might sell you
“CM” but can’t solve your problems. Any
who try to sneak through will stand out like a sore thumb.
Issue your document as a “request for information”
rather than a “request for proposal.”
This will allow vendors to respond in a flexible way, telling their
stories without the pressure of submitting a formal proposal with final
pricing, etc. (although you should ask for budgetary costs in your RFI.) It will also allow them to submit responses
to only those portions of your needs they can address directly without the fear
of being disqualified. You should
encourage partial responses in your solicitation because you may end up
integrating pieces from different vendors.
When you receive responses from vendors who believe they can meet your
needs, cull them and schedule the survivors for sessions that include two
1. A detailed and frank discussion of the vendor’s approach to
each of your functional requirements. This will probably require the vendor’s
technical staff. Salesmen won’t like it
but your future is at stake so persevere.
2. A demonstration of at least some of the functions, with your
data if possible. If a vendor can’t
tune his demo to your needs given reasonable time, he probably can’t tune his
product to them in operation. This is
listed second because the demo should never be the major criterion for a
selection. Instead, it should merely
validate the vendor’s descriptions of how he plans to address your needs. Demos tend to be like carnival side shows;
what you think you see isn’t always what’s really there.
At this point, you will be in a position to go ahead with a
formal RFP or just select a vendor and start negotiations. You will also be in a position to ask the
winning vendor to write a contract that commits him to meeting your needs as
described in your function list, giving you some recourse if things don’t work
out. It will also let the vendor know
the bar he must reach. If you don’t
feel comfortable with the processes described above, don’t be afraid to hire a
consultant; someone with the experience to help you ask the right questions and
fully profit from the answers.
While there’s no silver bullet in matters like these, if you
follow the suggestions above, take your time and keep your wits about you, you
have a significantly increased chance of success in your quest for effective
Barry Schaeffer is President of X.Systems, Inc., a consulting and system development firm specializing in the conception and design of text-based information systems, with industrial, legal/judicial and publishing clients among the Fortune 500, non-profit organizations and government agencies. E-mail him at [email protected]
Editor’s note: This article first appeared on IntranetJournal, an internet.com site.