Several months ago, I ran a complete review of easily available search widgets for my Typepad blog. At the time, Technorati provided superior results and ease of use. Today I've replaced my Technorati search with Lijit.
Since the Search Widget Death Match, I've had a number of frustrations with the results' data quality. Primarily, the fact that Technorati's searchlet did not search old blog posts - only sufficiently recent ones. This is not acceptable for my need of a complete, thorough search tool for my blog content. I often search my own content in order to remind myself of past company snapshots or to reference old notes, and it's very annoying to get an empty search result on a topic that I know I've written about. Search on 'burning man' in the Technorati widget (I'll leave it in the right column for a few more days) and you'll only get this month's post. Search for 'burning man' in the Lijit widget, and you get all of my burner posts.
Here's more of a download on what I liked and disliked about Lijit search:
Pros
- Interesting (and good) that Lijit searched the comments content in pulling out results
- Easy setup via the Typepad widget gallery
- Beautifully done (easy and surprisingly thorough) Web 2.0 content aggregation
- Loved the enabled-by-default 'naughty word filter' that will keep icky bits from appearing in the search cloud. (Maybe we should pass this to the Spock guys?)
- Smugly pleased that the service is new enough for 'christine' screen name to be available (that's always extra fun)
- Lijit is presenting me with my search statistics (also powered by Google?) in a friendly format, and for free
Cons
- Still some tweaky bugs; entering 'http://christine.net" as my blog URL during setup popped up an invalid URL error (worked fine with the www added)
- Results pages include lots of duplicative clutter; e.g., all pages that mentioned 'burning man' in a post link were included. Since I by default include links to my last 10 posts in the left column menu, that means a lot of trash results being dragged in for each blog post using the search phrase in its title. Fortunately, these hitchhikers weren't ranked as highly as the actual posts in the results set.
So....thanks to Brad Feld for passing me the Lijit link to check out. It'll be kept as the alpha search dog for this site. Though I haven't yet used the search heavily enough to warrant the ongoing results quality, this search implementation looks much more promising than the last batch tested.