The conclusion of the Ask4Stuff experiment

Back last June, the OCLC Innovation Lab announced the availability of a Twitter-based service called Ask4Stuff. The idea (in a nutshell) was to let people tweet a request for information on a particular subject to the service using the #Ask4Stuff hashtag. The service would then return a link to a WorldCat.org set of resources based on a search of that subject. Later, we added a more complex multistep analysis of the request matching to various classification and ranking schemes. It was an experiment in developing a more "social search".

The experiment was a success... in that we learned about many things, including:

• Creating this new type of social search has a measurable positive impact on search engine ranking
• Getting library services to work and play nicely with mainstream consumer applications can be done
  
• The switch to Twitter's OAUTH authentication from their basic authentication was painful and time consuming
  
• The usage of a social media service linked to library materials needs more thought
• Users outside the library community participated in the experiment with as much vigor as those inside our community
• Some natural language processing mapped through classification systems such as Dewey produce improved results 

What we also learned, though, was that while people found the Ask4Stuff service very interesting, in the long run it wasn't deemed to be sufficient as a standalone service. It wasn't "sticky" in consumer terms. In the days immediately following the announcement of the service, we saw 200-300 tweet-requests per day, excluding the hundreds of tweets about the service. That leveled off, and for the next month we saw below one hundred requests per day. The average user only used the service one time. The highest usage for a single user was 17 tweets. Recently? Well, let's just say that we have dropped well below a "per day" measurement. All in all, there were around 1500 tweet-requests from just under 500 Twitter-ers. More than a fourth of the users that tried the service also followed the Ask4Stuff twitter account. Again, an interesting service but not sticky enough to sustain usage.

This kind of "failure," though, is just the kind of project the Innovation Lab is set up to run and observe. We hope for successful adoption in every project but we don't go into experiments like this one thinking, "This will take the world by storm!" We go into it thinking, "This sounds interesting; but we have more to learn before we commit to a direction." Which is exactly what happened. one of the goals for the Innovation Lab: "Go out there and learn from everything you do, even when it fails." As long as we're honest about our goals, and scrupulous about transparency, we really can't fail. We just kind of succeed... sideways ;-)

To conclude: as of April 1, 2011, we won't be supporting the Ask4Stuff Twitter account/service. We'd like to very much thank all the folks who tried it out, provided feedback and linked-to/friended/followed the project. One of the most pleasant things we learned from this is that there are a *LOT* of people who like to play with things like this. We'll be counting on that for future Innovation Labs projects.

If you can think of any interesting uses for the service that we didn't touch on, or want to know more about how we did the experiment or what our specific results were, please feel free to shoot me an email at innovation@oclc.org. One of the other things we're dedicated to at the Innovation Labs is sharing our work. If you think you've got a home for Ask4Stuff (or a related step-child application), don't hesitate to let me know. We are continuing to incubate ideas for libraries' participation in social services.