Urtak

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

~ the obit ~

Urtak was a poll that lived under other people's articles. A publisher dropped in the widget, and a reader who scrolled far enough got a question with two obvious answers and a third one, don't care. Answer it and another question arrived, pulled at random from the pile, then another. The average session ran 23 questions deep, which is not the behavior of somebody filling in a survey.

Aaron Gibralter and Marc Lizoain built it in New York starting in 2008. The feature they kept pointing at was not the voting but the asking. Any reader could submit a question of their own, and once the publisher approved it, it went into rotation alongside the ones the newsroom wrote. That inverts what a poll is for. An editorial poll asks what the desk already suspects. Urtak let the audience decide what the desk should be curious about.

It got real distribution. Andrew Sullivan ran it, along with the Huffington Post, the CBC in Canada, El Tiempo in Colombia and The Blaze. Roughly one reader in ten touched it, which for anything embedded halfway down a page is a high number. By the end it had gathered more than 22 million answers across upward of 77,000 questions in something over 13,000 polls. In January 2012 it raised 500,000 dollars, with Esther Dyson among the names on the round.

Nobody chooses a widget. You meet it inside something you were already reading, which is how the poll box under the article became furniture on news sites everywhere, Turkish ones included, always a single question, always written by the desk. Urtak's version of that furniture sat on English-language media sites I was not reading in 2012, so I met it much later and as a footnote. The mechanic is what stayed with me rather than the brand, and the mechanic was giving away the right to ask.

~ the rap sheet ~

~ why don't care mattered ~

A two-option poll forces everyone into a position, including the people who do not have one. The third option was an escape hatch and it did two useful things at once. It let a reader keep going without lying, which is why sessions ran to 23 questions instead of one. And it produced a figure nobody else was collecting: the size of the group that had read the question and declined to hold an opinion about it.

That figure is more interesting than the split sitting next to it. A question where seventy percent say they do not care is a question the newsroom should probably stop writing about, and no traditional survey instrument will ever tell you so, because a respondent who does not care has already closed the tab.

~ who was paying ~

Nobody, is the short answer. The widget was free to publishers and the company had not turned on revenue by the time it took the seed round in early 2012. The plan discussed at the time was sponsored questions, which is the plan every question-and-answer product eventually writes down, and which needs enough volume that a brand will pay to have its question shuffled into the deck.

Twenty-two million answers sounds like that volume until you divide it. Spread across five years and a few dozen publishers and 77,000 questions, it is a very large number of very thin engagements, almost none of them worth anything on their own. The economics of a widget are the economics of a rounding error on somebody else's page.

~ the pause ~

On 16 September 2013 the founders announced that the service would be suspended from the 20th. The wording was careful and it was not a goodbye. They wrote about needing to pause and come back on a winning trajectory, which is what people write when they have not yet admitted to themselves that the thing is finished.

It never came back. The domain no longer resolves, and the archive of 22 million answers to 77,000 questions, most of them written by readers rather than editors, is not sitting anywhere you can query it. That is the actual loss. The widget itself was replaceable and got replaced within a year by a dozen others. The corpus was not.

@}-->--

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

« previous  ·  random  ·  next »