When I wrote my first three posts on bitly (as “bit.ly”) nearly five years ago, they offered a serious-yet-humorous attempt to answer a simple question: Given that bitly’s shortened URLs were only 6-character strings and given the effort by 301Works (through the Internet Archive) to preserve those “mappings,” how long would it be before bitly ran out of strings? My assumption-dependent answer back then — roughly the year 2030 — is now as irrelevant as 301Works seems to be invisible, but you can find the third post here with backlinks to the first two: “Soooo big!” Counting on bit.ly, Part III.
The humor was embedded in some of my tongue-in-cheek suggestions for how bitly could solve the supply side of the problem:
* add more domains in Libya (LY): unfortunately most of my favorites — gig.ly, hig.ly, pig.ly, and wig.ly — were taken, but the balance of the “6-vowel family” was available — bat.ly, bet.ly, bot.ly, but.ly, and byt.ly;
* use other countries: domains in Estonia (EE) “would appeal to a certain demographic (and leave others merely puzzled): hu.ee, dew.ee, and lou.ee”; in this scenario, bitly could add characters yet leave the string length unchanged, which sounds like a Zen koan in the making;
* add character sets, instead of characters: Greek, Cyrillic, Hebrew, even Dingbats; or
* just wait: perhaps before there is a self-driving car, there will be nanobots in neo-natal eyedrops to install an iLobe in each of us at birth, connected forever to VerizoNeural.
Unknown to me, at essentially the same time as those first posts (early July 2010), Twitter introduced its own shortening service, t.co (June 2010). Because Twitter’s 140-character limit on tweets was the initial driver for URL-shortening, this event precipitated volatility in bitly’s traffic. I documented that chatter in three more posts:
* Maybe not “Soooo big!” But still counting on bitly (October 2011);
* The bitly dea(r)th watch (February 2012); and
* Long day’s journey into bitly (July 2012).
What started as an amusement in 2010, became progressively “somber, morbid, even sepulchral,” as the three titles above suggest. The only redeeming factor was a mid-2012 infusion of $15 million in venture capital from a group that included Vinod Khosla. Now, after a couple more years of check-ins on compete.com to look at monthly data for “unique visitors” — my surrogate metric, using the only public traffic data I have found — it’s time for one last update of the graphs from previous posts.
Here is the graph for all three URL-shortening sites and their sum:
Clearly, t.co has taken over and is on a steep growth path.
Here is the overall graph for the two bitly domains, with a brief data gap (due to inattention on my part) just at the time the VC money was committed (month 49):
It seems safe to infer that during months 46-64 (Mar 2012-Oct 2013) things were flat.
Finally, here is the segmented graph, with ups and downs demarcated by individually-fitted lines:
Some, but not all, of these have higher correlation coefficients than the single-fit line in the second graph. To do a finer-grained interpretation of this pattern would call for a more sophisticated analysis, but neither the problem nor the available data nor my interest in it are enough motivation to fire up SPSS.
Instead, what started as a bit of a lark — and three small ducks — now seems largely irrelevant. I assume that bitly’s upward track, which certainly remains real overall, is a result of its utility to branded marketers (Nike, eBay, GE, Ben & Jerry’s among others identified on their website). Good for them. And presumably good for Vinod Khosla and partners. But for me that long day’s journey is over. Farewell, bitly.