Is Google's "Survey Wall" Experiment Brilliant or Evil?

The search giant is partnering with online publishers to replace their sign-ins and paywalls with market research micro-surveys.

Paywalls may be working for the New York Times, but other publishers looking for a digital lifeline may not want to rely on them. Still, what’s the alternative? Google has quietly cooked one up that asks online readers to answer one market-research question as a "toll" for accessing content. Nieman Journalism Lab is calling it a "survey wall" model. It’s like a micropayment, but with attention and data instead of cash out of your pocket.

Google’s interaction design, at the very least, struck me as savvy. Upon accessing an article through Google search, the user is presented with a choice: go through the publisher’s annoying log-in rigamarole, or answer a quick market-research question (e.g., whether you use SMS, social networks, or instant-messaging for casual conversations). Surveys are irritating, but if it’s only one question—and if you can answer/dismiss it with one click (Google provides large, Tumblr-like buttons for this purpose), dealing with the survey may actually offer the more fluid user experience, compared to filling out a log-in form or being forced into a purchase decision about the content itself.

There are obvious chinks in the armor, though: the "survey wall"'s ease of use may perversely incentivize readers to provide crappy data ("who cares what button I click, just get me to the article"), which won’t be worth much to the marketing partners who are paying publishers for the privilege of harvesting it. But if Google can continue improving the UX design of that one-question, one-click mechanism, they may be onto something. It’s all about framing: Providing intel to a body-spray company in order to read about the latest United Nations resolution is crass and offputting, but what if the survey question was posed by an NGO with a mission related to the article you were about to read? It’s like the difference between CAPTCHAs (annoying) and ReCaptchas (tolerable because it’s for a good cause), which Google clearly understood when it acquired the latter as an in-house product in 2009.

Survey walls, nagwalls, whatever kinds of walls—the point is, experiments that pit empathetic interaction design (and knowledge of cognitive biases) against the harsh realities of digital economics are going to become more common, not less. Attention and money are slippery, unpredictable things on the web, and experiments like Google’s latest may blow up in their partners’ faces. But for better or worse, they’re the only way forward.

[Via Nieman Journalism Lab; top image by Joost J. Bakker]

Add New Comment


  • Satanfornoreason

    The author of this article is incorrect when he claims that you can "answer/dismiss it with one click". You can't "dismiss" the ad. There is no button to do so. You can close the web page, but that's closing the web page, not dismissing it.
    I disagree because I've purposely answered two survey questions incorrectly, but I've taken the time to read it and the answers. I don't know how often others would do the same, but my behavior alone makes the data invalid. You can't distinguish good (i.e. valid) data from bad (i.e. invalid) data based on the amount of time it takes a user to click. First, you have a ton of variables just based on the connection speed, the user's hardware, browser, etc. Second, choosing a cutoff point will inevitably leave you with the same problem. Users who read the question quickly but answer accurately will be left out of the responses that they count, and users that take the time to read it but then stop to decide whether or not to proceed (or leave the site) might be included in the responses, though it took them a little longer to decide to go through with it. There are also problems with logging in - if that's the first option taken, but you forget your password or mistype and decide to answer the question with a random click instead of trying again.

  • Matt

    They might be able to correlate 'good data' with a certain amount of time taken to answer.  People just clicking through would answer faster than you could read.

  • Carterrosenberg

    I love it, who said you to answer honestly, or even read the question? It might as well say "Do you want to see this website?" "YES", "YES", "YES"

  • Chris Dowsett

    If Google can overcome the 'crappy data' problem that you mentioned and the question is designed well, this could be really valuable to companies. It also means a bunch of extra data.