Urchin

» see urchin.com as it lived, on the wayback machine

~ the last words ~

urchin is a name most people met in a hosting control panel rather than as a product they went looking for. shared hosting gave you webalizer or awstats for nothing, and urchin sat one row above them as the option that cost money. i never bought a license. i have also typed its initials several thousand times without once thinking about where they came from, which is true of almost everybody reading this.

the origin is better than the product. paul muret and scott crosby started a small web hosting company in san diego in december 1995 under the name quantified systems. they billed clients by bandwidth consumed, and working that out by hand across a shelf of servers is miserable, so muret wrote something to parse the access logs and total it up per customer. the billing tool turned out to be more interesting than the hosting business. by 1998 it was a product called urchin, and brett crosby and jack ancone had come in to sell it.

it shipped in two forms, a paid edition aimed at internet service providers and a free one for smaller sites, and that split is why it spread. urchin did not chase individual webmasters. it went to the companies that already had thousands of customers and a control panel to put things in, and let them resell it. the company took the urchin name for itself in 2003, after the quantified systems label failed to mean anything to anyone.

the technical position was log analysis at a time when the industry was moving to page tags. logs are what the server already knows, which means no javascript and no third party holding your data, but they also count every crawler and every cached hit as traffic. urchin ended up doing both, reading logs and accepting tag data, and stitching the two into one number. that hybrid is the thing google wanted.

~ the record ~

~ what google actually bought ~

google acquired urchin software corporation in april 2005. the price was never published. scott crosby has said the earliest investors saw returns on the order of a hundred times their money, with the average investor around thirty, which tells you enough. in november of that year the hosted version came back as google analytics, free, and the paid analytics market lost its floor overnight.

the self-hosted software did not die there. google kept selling it as urchin from google for another seven years, to universities and to anybody whose lawyers would not accept traffic data living on somebody else's servers. version 6 arrived in april 2008, 6.5 added adwords reporting in february 2009, and version 7 in september 2010 brought a rebuilt interface with event tracking. paul muret, who had written the original log parser, was by then running analytics at google.

~ the three letters ~

the part of urchin that outlived urchin is a naming convention. the urchin tracking module needed a way to tell where a visitor had come from when the referrer was useless, so it read parameters off the end of the url. utm_source and utm_medium, with a campaign name to tie a run together. every marketing email you have deleted unread and every tracked share link you have ever clicked carries a prefix named after a defunct san diego log parser.

nobody chose this. google analytics inherited the parameter names because rewriting them would have broken every existing customer, and then google analytics became the measurement layer of the commercial web, and the prefix went everywhere the web went. it is now in url specifications, in privacy tools that strip it, in the muscle memory of people who were not born when quantified systems was billing for bandwidth.

~ the postmortem ~

on 20 january 2012 paul muret posted a note on the analytics blog titled the end of an era for urchin software. no further development, no new licenses or upgrades after 28 march. the last build, 7.200, had gone out days before the announcement. google's reasoning amounted to focus, which is the word it uses when a product is not worth the engineers standing next to it, and by then log-file analysis had shrunk to a corner of the market anyway.

the people who ran urchin on their own hardware were mostly the ones with a policy reason for it, and they got pushed toward a hosted product that solved the policy problem by ignoring it. that trade got made all over the industry in those years. urchin is just the version where you can watch a company buy the alternative first, then switch it off.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »