Content without it? The doubts about Vignette

By David Walker (Google profile)

They were an interesting crowd. Among them was the webmaster for a government institution assessing site management tools, the tech team from a large private school assessing intranet solutions, a smattering of IT consultants assessing the next likely gravy train, and several corporate managers trying to make their existing Web processes work better. They'd all turned up for the launch of Vignette's V/5 system, an assemblage of software which aims to solve the site content management problem. It does this for a minimum of roughly $A200,000, and often more like $A1 million ($US500,000) when consulting costs are included.

What does a private school need with such a pricey piece of software as Vignette? Probably nothing. Vignette's probably the wrong solution for many of the audience at the V/5 launch. But they turned up nevertheless, because the content management problem is looming larger and larger in the minds of Web builders.

The site content management problem happens like this. Your site starts off with a bunch of pages. If you're smart or well-advised, you put the content of those pages into a database right at the start. Then you build an interface to add content to the database and pour it into pre-defined templates for serving up as Web pages. Then you add some code to allow the right people to edit the right pages. Then you add a script library. Then you arrange frequent back-up for the database, and version control for the templates and database scripts. Then you add more code to allow your content to be syndicated, perhaps through RSS. Then you start wondering about the easiest system for editing the templates. And then ... you start to wonder whether some off-the shelf software could replace the bowl-of-spaghetti mess you've created.

Vignette were just about the first people to identify this problem. They started by simply hacking together some code from the CNet site's proprietary system, which they sold to company executives. They stretched the truth a little to make their first sales: Web database guru Philip Greenspun memorably described an early Vignette marketing spiel as "akin to hearing Adobe pitch Photoshop as a payroll check processing system". Later versions became more sophisticated, though never elegant; many developers today hate Vignette, some passionately. The company retains a reputation for delivering more in its marketing than it does in the software. Nevertheless, Vignette has emerged as the content management market leader of sorts.

Why "leader of sorts"? Because Vignette - and rivals like Interwoven TeamSite and Broadvision - still often run a poor second to the do-it-yourself solution. You see, the core content management task is relatively simple: create a database schema in a robust relational database, shove your content into it and whack on a front-end interface to manage elements such as workflow. Except in rare cases, you can skip many of the fancier Vignette features, such as the much-overrated "personalisation", and save yourself much of that $A200,000. In-house development lets you create the exact functionality you need (for instance, you might want to associate content assets with more than the single metadata keyword allowed by V/5). If you're using Microsoft's ASP or Allaire's ColdFusion, you'll have a decent pool of developers to call on. And in-house development may also let you tie your content management to your Web site's transaction engine and other applications.

Vignette and similar systems simply cannot offer this degree of flexibility yet. The most vicious criticisms of Vignette and its rivals come from developers who have tried and failed to tailor them for a site's business needs. (One not attypical example: in late 2000, Vignette settled a legal action from a user reportedly unhappy that Vignette didn't accomplish what he'd been told it would.)

Then there's that price. $A200,000 is plenty of money for a tool which many users complain does almost nothing straight out of the box. So much customisation does Vignette's software require that the company makes half its money from consulting services. Over-ambitious Vignette installations can endanger launch plans: at least one Australian start-up ended up going out of business before its developers could ever get their Vignette-driven site up and running. So Vignette is vulnerable to new entrants like Allaire's ColdFusion-based and keenly-priced Spectra system and even to open-source solutions like Zope and the Disney Internet Group's recently-released Tea.

Vignette's ambition seems to extend as far as becoming the underlying platform for Web sites. It's a not completely implausible ambition. But it will require Vignette to do more, probably for less. V/5 makes a start on that task, allowing Vignette functions to be coded in ASP, and the company promises Java support by mid-2001. Whether all that will justify a $A200,000-plus price tag remains unclear.