12:03 pm - Fri, Dec 5, 2008

URL Shorteners Redux: Winning Strategy

I applaud your verve Kortina, and you know I also think a bit(ly) about the utility and future of URL shorteners. We can probably agree that the URL shortening market has been and will continue to be driven in the short term by the limits imposed by microblogging formats like Twitter. The promise of something more in terms of historical metrics, collective insight, search and discovery has yet to catch on, but it will come and many of the amazing things we’ve chatted about will come to pass. However, the game right now will be won by focusing first on the bread and butter shortening which is the focus of my comment.

As I’ve also mentioned, URL shortening tied to a specific domain “brand” is a strategy that will eat itself as it becomes more popular. But here’s the detailed explanation (less for you as I know you get it, but I never really quantified it for my own amusement):

<nerd> A quick back of the envelope calculation reveals that given the character set [_0-9a-zA-Z+-] has 65 possibilities per character, the maximum number of URLs shortened is 65^number of characters (each character adds 64X the previous availability). Currently, tinyurl is up to 6 short string characters representing about 68 billion URLs (well, less because they stupidly don’t use +,-, or _ and you could use / too, but not in the first or last character positions). </nerd>

So what does all this math mean in terms of strategy? It really hammers down the value of the domain parking/multiplexing strategy I wrote about. Run, don’t walk and buy up as many short URLs as possible bit.ly, is.gd, zz.gd folks, name recognition be damned. Why? To map URL shortening across multiple short domains of course! For every 65 5-character domains I purchase like zz.gd (I’m including the dot in my count), I can reduce my short string character length by one. So, I buy 260 5-character domains, I can beat every tinyurl.com shortened URL by 10 characters with zero loss of capacity - I save 6 on the domain length and 4 on the shortened string by using 4*65=260 domains. In a 140 character limit game, that’s a HUGE and easy win, especially when you think about advertisers and marketers who want maximum room for content and will PAY for the analytics you speak of (trust, soon you will find marginal value calculations for each tweet character). However, there is a finite number of resources (domains) to implement such a strategy. This is THE barrier to entry for competition within your market and should not be glossed over, so buy up all the .me, .gd, .ly etc. domains you can.

So, that’s the case for domain parking + URL shortening. Without going into as much detail, you can probably see that from a performance standpoint, there is a strong case for CDN’s + URL shortening also. I’ll leave that as an exercise for the reader.

Finally, your claim that bit.ly can track anything anywhere isn’t completely true: bit.ly can track any URL anywhere a click occurs. Ergo, bit.ly and other URL shorteners measure click-confirmed intent, but they don’t measure visual consumption the way Google Analytics can. Tracking codes and bits confer different powers than mere URLs, though URLs can be more ubiquitous across platforms. You say that GA can’t track views on your YouTube videos, but neither can bit.ly unless somebody clicks a bit.ly link (and besides, viewing statistics is available via YouTube’s API). Further, URL shorteners cannot reveal any context of their destination URL within the distribution channels. The value of URL shorteners is that they are the only viral friendly tracking device in a world where distribution channels are completely uncertain in the sea of current innovation.

As always man, you keep my brain clicking and whirring. :-)