I’m in Palm Springs attending Intelligent Content 2009, a very interesting conference hosted by The Rockley Group. The venue, Le Parker Meridien, is a very nice “boutique” hotel. The conference rooms are well set up and private, and the grounds are beautiful.
The conference started off with an introduction from Ann Rockley, whose company is the sponsor. Her objective is to provide a small, focused conference. To that end, her team worked with the speakers to make sure the talks were of high quality and focused on the objective. So far, most of what I have attended has done that. This entry will provide a few highlights from the first day of talks.
Salim Ismail gave the opening keynote. The question he addressed was “What makes content intelligent.” He defines intelligence as the ability to take patterns from one context and use them in another. In a sense, Intelligent Content boils down to the idea of using embedded information, along with information about the user (i.e., context), to deliver customized information. Although it wasn’t clear at that point, the rest of the day reinforced this as the central element of intelligent content.
Next up was Scott Abel, who provided examples, both good and bad, to illustrate the idea of intelligent content. His recent post on The Content Wrangler describes the “bad” in detail. Apple provided most of the “good” examples, with the central theme being that Apple uses all of the information available to the company about a customer to provide a personalized experience targeted at increasing sales. As always, Scott gave a spirited and informative talk.
This was followed by three parallel sessions. The first I attended described a system being developed for breastcancer.org, a site devoted to helping women who have been diagnosed with breast cancer and their families. Derek Olson and Ron Daniel discovered some interesting, and not immediately intuitive, things about personalization. The central problem they faced was how to gather accurate information from each user so they could select the most applicable information for that particular person. The obstacle was that users often incorrectly characterize the state of their illness, which makes it difficult to give them the right information. They built a flexible taxonomy, which helped considerably, but in the end, the most useful help to users was input from other users, who independently devised a reasonably standard way of describing the critical factors about their situation in signature lines. This ad hoc metadata helped users help themselves. Overall, a nice example of user created design.
Rober Lee, of Symantec, gave an excellent talk about optimizing search results. Unlike standard SEO, which seems to be primarily aimed at drawing the maximum number of hits, his objective is to analyze search results to make sure users get good answers for their searches. Here are a couple of key points:
- Look at search logs
- Look for top search terms
- Test those terms (and look for terms that yield no results)
- Optimize content, then re-test
Optimizing content involves getting the right content for the right search terms. Suggestions for doing this include placing search terms in titles, in the first 100 words of content, and in the filename. One counter-intuitive point is that minimalist writing might not be the most effective way to get search engine hits. Repetitive text early in the content, what he called “stupid text,” helps with this, even though it offends the best instincts of tech writer. Overall, an excellent talk.
There were a couple of other talks, but I needed to duck out early, though I did leave a copy of Managing Writers to be used as a door prize during the evening festivities.
More to follow tomorrow.