In my last post I mentioned that people from the BC Rich Players Forum were visiting Craig’s Little Buddy to find used BC Rich guitars. I actually signed up check out there forum and saw that someone had links to a BC Rich Stealth for sale in New Jersey.

That’s awesome… except it outlined a shortcoming in Craig’s Little Buddy.

The “browse” links determine which for sale item to show based on an index variable in the URL (i=xxx). Also, if you’ve never used my craigslist search before it will default to searching the 22 main craigslist cities.

Now… this brought up 2 shortcomings.

1. The index thing was stupid. The browse pages should operate on permalinks not ever-changing index values.
2. People may publish a browse link to an ad which sits on a non-default craigslist location (New Jersey in this case)

So I fixed this tonight…

1. The browse links now operate off permalinks (p=http://xxxxxxx).
2. Since the browse links contain the permalink… if the particular ad is not in a default city list I add that location to the searchable set as if the user had customized which cities they wanna search.

This works way better. The index thing was just a brain fart. The index will always change… if a new ad shows up the indexes are all different… that was just stupid on my part. Plus, expanding the searchable city list on the fly is pretty good stuff. The odds of the browse link being valid are way higher now. The only way it’ll be invalid is if it’s not found via Craig’s Little Buddy at all.

This just strengthens my previous rant on the analytics thing. Seeing the context of your inbound links helps you follow the same flow random Joe’s gonna follow when visiting your site. In certain instances (like this) bugs, areas of improvement/clarification & the need for better error text/detection are revealed.