The article certainly gave me food for thought. The whole time we've been working to bring greater consistency to how University units organise their content, I've advocated audience first, then task - largely inspired by McGovern's work. The big exception in the University context (I feel) is when a service offers the same thing to multiple audiences, or when an academic unit optimises the way they profile themselves to suit priority audiences. In these instances, as Gerry says, task comes first.
So, it doesn't matter who you are if you want to get a replacement library card or work out how to access the wireless network or apply for a parking permit. Similarly, writing research activity profiles a particular way and grouping them together doesn't have a lot to do with audience segmentation.
However, we do have an awful lot of information where the answer to a website visitor question is (initially) - 'it depends who you are'. And we are organised in many many areas to deal specifically with certain target audiences. So in a devolved publishing framework, the person who administers postgraduate applications is often the person responsible for the management of the webpages.
So on balance, I'd have to conclude that much of Gerry's argument - persuasive as it is in the context of the examples he gives - doesn't apply to us much of the time. Well, not the outward facing aspect anyway. His tales of Dell could strike more of a chord with our service units.
What reassures me most is his closing statement:
If you're considering audience-based navigation, ask yourself these questions:
Are my audiences totally distinct and separate from each other?
Are their tasks totally different?
Do I have the budget and resources to build and maintain unique content for my audiences?
Why audience navigation usually doesn't work - article by Gerry McGovern