Wednesday, July 9, 2008

My Boston KM Forum Presentation on Enterprise Tagging

It's been a busy couple of weeks. Right before I left on a week's vacation, split between a chamber music camp called "Heaven" and Cape Cod, I presented to the Boston KM Forum, which has a blog as well as its Web 1.0 website, on tagging inside the enterprise. My slides are also available there.

If you want to jump directly to other blogger's reactions, please see Sadie Van Buren's post and discussion about tagging as part of search activity and Suzanne Minassian's concise summary.

My main purpose was not to discuss how to implement tagging in the enterprise (my firm hasn't, so I can't), but rather to identify what I think are current business needs that could be met by tagging, some of the ways that enterprise tagging is different from tagging on the web, and to identify some of what I am looking for in tagging software.

I tied tagging to Prof. Andrew McAfee's SLATES paradigm of Enterprise 2.0, by which in my opinion any social software package needs to be assessed. SLATES=Search, Links, Authorship, Tagging, Extensions, and Signals; for more information, read the Dawn of Emergent Collaboration. In that paper McAfee focuses on tagging on the Web and the way that it enables people to assign their own context to web sites, which can then be shared with others. Looking back at the paper it is striking that McAfee focused on how intranet and internet sites might be tagged and leveraged inside the enterprise, but did not extend the concept of tagging to more typical inside-the-firewall content such as documents, matters, people, or tasks.

Tagging will help law firms provide context and enhance findability with respect to the large number of similar documents generated by legal work. By having tags linked to actual identified people (unlike Delicious) tagging should help us identify internal experts, either through their work being tagged or through their own tagging activity. My hope is that it may also enhance our ability to find out about new work, for instance, through having an internal RSS feed on a particular tag that would signal, for instance, when a new brief about civil rights was filed or finished.

Another driver of enterprise tagging may be employee engagement. One theme of the Enterprise 2.0 conference is that employees just entering the workforce, members of the "Millenial" and "Y" generations, enter the workforce with a great deal of enthusiasm but can get disillusioned by the difficulty of making measurable contributions inside a large and beureaucratic organization. Tagging, along with other social collaborative activities, may be a way to enhance engagement and derive value from these employees, in a fashion similar to that they may already be familiar with on the web.

That is not to say that tagging inside the firewall will be the same as it is on the web. Tagging must to adapt to the enterprise.

Professional service firms of all kinds have to be able to have ethical "walls" for all their content at the individual, matter, and client level. Such "walls" can prohibit a particular person from viewing a set of content, or can prohibit all but a very small set of people from viewing the content. That is, a tag might itself reveal the existence of a confidential client or matter, or show to someone who shouldn't know that a particular event has happened on a matter.

Ideally, enterprise tags will extend to more than "just" web sites. I would like to see tags (in descending order of importance) on 1) our document management system 2) the other document repositories in the firm 3) people 4) intranet pages and other content 5) matters and 6) tasks and projects. Ideally, external web sites would also be taggable.

While tagging people is third on the list, I believe that attribution of tags to people, either through their own activity or through others, will be a huge benefit to large professional services organizations. That's because finding the friendly experts is particularly important for us. One major difference of enterprise tagging is that there is no expectation of privacy within an organization as to one's activities; accordingly, authorship of a tag should definitely be exposed.

Some firms may experiment with somehow giving tags different weights depending on the seniorty or institutional position of the tagger. I'm not sure that's a good idea, but I would like to see some way to expose who made the tag, and exposing the role of the tagger could be quite beneficial. In evaluating work product, a first year associate's tag should not be accorded as much weight as the junior partner's tag in the mind of the person viewing the tag, but reflecting that in a system poses many obvious problems.

I also spoke about some features that should enhance enterprise adoption of tagging.
As with any new technology, providing help and guidance at the time that users need it will be key. Ideally there will be training features built right in, as you see on Amazon, ranging from a "What's this?" link, to a very basic description of tagging built in to the feature, to automatic feedback on a user's tags. This last might be something like "This is the 149th item you've tagged with the word "document". Do you want to try another word?"

There was an interesting discussion afterwards about the timing of tagging. Again, this is not an issue that I have addressed directly. The users I work for apply some metadata to the documents at the time they save them (principally a matter number and a document type), and we do our best to leverage that matter number to apply other metadata to the documents. Tagging could be applied to some extent at the time a document is saved, but typically our users don't want to actually have to think at that point as they need to print, file, or send it. It may be better to have tagging applied by other users at the time they search for and view the document, person, or other content. That would be a better approach where search is federated across many different content sources as it would be difficult to have "native" tagging in many different legacy systems.