Google Surveys

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

~ the obit ~

The number worth keeping from this one is five cents. That is what a publisher earned when a reader answered the question standing between them and the article. Google charged the buyer around ten cents for a response from the general US population and fifty for a targeted one. It kept the spread and handed the rest down. An article, priced by the sentence in every argument about journalism since 1995, turned out to be worth about a nickel if you paid in attention instead of currency.

It launched on 29 March 2012 as Google Consumer Surveys with roughly twenty publishers signed up, Pandora and the Christian Science Monitor among the first of them. The mechanic was one screen: a question slid into the page, shading out most of the text, and stayed until you answered it or declined. Then it vanished and the article opened, and stayed open for however long the publisher had set.

The clever part was not the wall. It was what sat behind it. Every consumer research company in the world was paying people to sit through twenty minutes of questions, and getting a sample made entirely of people willing to sit through twenty minutes of questions. Google had a different pool: everyone trying to read something. Ask one question at a time, across enough pages, and you assemble a national sample out of people who never agreed to be a panel.

None of this reached the sites I read. Turkish news publishers were not metering anything in those years and had no reason to start, since the business was ad density and the ads were already everywhere. The surveywall was an English-language product from the publisher side out, and it arrived where the paywall argument was already loud. In Turkey that argument was still a decade away.

~ the rap sheet ~

~ the polling detour ~

In 2012 the thing accidentally became a pollster. Nate Silver ranked Google Consumer Surveys second for accuracy among everybody who polled that presidential election, above a long list of firms whose entire business this was. The reason is not mysterious. Telephone polling was already struggling with the fact that the people who answer unknown numbers are not a cross-section of anything, and Google's respondents were sampled from something much closer to the general reading public.

In 2016 it did badly, along with most of the industry, and the accidental pollster stopped being cited. What survived that four-year arc is a fairly uncomfortable result: a market research tool built to sell soap questions did better at reading an electorate than the specialists, once, and nobody could keep it there.

~ why the publishers walked ~

Google's own case study named GoErie.com, the local paper in Erie, Pennsylvania, at an estimated $200,000 a year in survey revenue with no measurable drop in traffic. For a paper that size that is a newsroom salary line. But by the mid-2010s the Wall Street Journal and the New York Times had made the hard subscription paywall normal, and once your reader will pay ten dollars a month there is no argument for renting the doorway out at five cents a head.

There was a second squeeze, and it came from inside. Google launched Opinion Rewards in late 2013, which paid respondents directly in Play Store credit and cut publishers out of the arrangement entirely. A publisher speaking to The Outline said survey revenue fell hard in 2017. The company that built you a business had also built the version of it that did not need you.

~ the day it died ~

The name changed once, on 19 October 2016, from Google Consumer Surveys to Google Surveys, when the product was folded into the Analytics family. The end was announced in September 2022, operations stopped on 1 November, and customers had until 1 December to pull their historical data out. Google's stated reason was that it still believed in cheap custom market research and no longer believed this was the way to do it, and that the team would go and find scale through the ads products instead.

Which is where the shape came back. Google has since put the same trade into Ad Manager as Offerwall, letting a reader clear a paywall by watching something or answering something. The mechanism was never the part that failed. What failed was the idea that a newspaper would rather be paid by Google than by its readers.

@}-->--

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

« previous  ·  random  ·  next »