Our latest webinar, “Organizing Help Content: Overcoming the Problem of Findability” was a discussion between Scott Abel and Tom Johnson, technical communication expert and creator of the ultra-popular technical communication blog, I’d Rather Be Writing. The two talked about the problems that impede content findability and suggested strategies for helping attendees improve their technical support sites by making their content easily discoverable.
There were a handful of questions asked by the live audience that were unable to be answered during the webinar. I enlisted Tom to help answer them. You can find them below. Thank you again to Tom and Scott for an insightful webinar.
Live Audience Questions:
Q: Tags have been around for a while. how useful are they in creating user-generated topics that others can use?
A: If you can control the tags, so that users choose from a list of available tags, this will help standardize the metadata for your content. In contrast, if users fill in the blank for tags, spelling the same terms in different ways, the variations will probably cause the topics to be grouped differently. The problem is that controlled taxonomies may not always include the terms a user needs to use.
Q: Please elaborate more about how lists are a bad thing.
A: Lists are typically sorted in alphabetical order or by date. You might have a metadata property that defines the sort order (you can see this with topics in the Semantic Enterprise Wiki), but the sort order for one list, such as Beginner Topics, may be different from the sort order in another list, such as Most Popular Topics. Sort order isn’t an absolute across all lists.
Generally, a long list of topics to choose from isn’t attractive because there’s no visual hierarchy for the reader to quickly make sense of the content.
You can evaluate whether lists work for a help system I created here. I feel that there is something missing or empty about having a bunch of lists.
Q: I can use metadata to search my CMS. Do you have any experience in implementing a metadata search engine experience for products such as Robohelp?
A: If your CMS doesn’t allow you to attach metadata with your content, nor enter metadata in the search queries, that I’m not sure how you would do that.
With your second question, building a metadata search engine experience, you could include some advanced search options for the user to select before searching. But implementing a faceted search might be more helpful, because it would allow the user to filter the results based on the product he or she is looking at. You would need to add custom metadata to XML files to accomplish this, among other things.
Q: Aren’t we really talking about the battle between taxonomy vs. folksonomy? How do we reconcile the two?
A: I assume this question relates to our discussion about search being problematic because the company’s terms may not match the user’s terms. Many systems allow you to connect synonyms with each other.
Another solution might be to provide instant results as users type entries into a search box (similar to Google Instant Search). If users can see that their term isn’t retrieving any results, they can quickly experiment with other terms until they get a better result set.
What’s great about faceted search/browsing is that it helps teach users the terms as they search. The search and browse experiences goes hand in hand. When a user searches for a term, the faceted navigation teaches the user the correct terms, allowing the user to make more accurate searches using the newly learned terms.
Q: Are you aware of any tools or technologies in place for tracking and analyzing user’s Help Search patterns, learning from the data, and then anticipating what they might need? Similar to the Pandora example, a Helpbot that tracks what the user searched for, topics accessed, and builds an understanding the user’s information needs so that Help can proactively recommend other learning opportunities for the user?
A: To build a smart system like Pandora that tracks our usage and then proactively recommends results, you would probably have to custom build something like that through a team of programmers. That’s part of the genius of systems like Pandora.
It might be possible, however, to have users input a profile that identifies their role, interests, and other descriptions. When a user logs in to a site, the help topics might automatically be personalized to their profile.
On a related note, Radiolab’s latest podcast, Talking to Machines, might be interesting to you. It explores ways that computer programmers develop machines with the ability to respond intelligently with conversational feedback based on input scenarios. I don’t think there are any tools or technologies out there that do that out of the box. You’d probably need a team of dedicated programmers to do it, and if you’re able to get budget for that, it would really be amazing.