Jon22: The Website of Jonathan Dobres

Today’s post started very differently, one of those days where I start out making one point and then realize I’m actually talking about something else entirely. I started this post wanting to talk about Google’s new disappearing, reappearing homepage (check it out for yourself), but I ended up reading about this fascinating usability issue that Google discovered a couple of years ago. When Google tested putting thirty items on the results page instead of the usual ten, they found that Google AdSense revenue dropped a whopping twenty percent. The reason for the plunge turned out to be a real surprise. The larger number of search results meant that the page would take, on average, 0.5 seconds longer to load, and in this half-second delay Google lost money. The take-away lesson? “People do not like to wait.”

That’s basically true, in the same way that sentences like “I’m all for job creation” and “I dislike murderers” are also true, but I don’t think that simple impatience explains the drop in revenue. Whether displaying ten results or thirty, your results still show up in under a second. I doubt this is enough time to cause the kind of anguish that would encourage a user to speed through the page with a radically different reading strategy.

On the other hand, in the visual system half a second is an eternity. Attentional cueing effects have been thoroughly studied since around 1980 (thanks, Michael Posner!), and the findings are numerous and complicated. To put it really, really broadly, these cueing effects can powerfully alter the way you deploy your attention, and the effects change dramatically depending on the length of the gap between the cue and the thing the cue indicates. Cueing effects show the most drastic change across a time window of—wait for it—about half a second.

So, picture it. You’re typing a search query into Google for the millionth time. You press “Google Search,” and the screen goes blank. This is the cue that tells you the results page is on its way. You’ve been to Google plenty of times before, and you know that not everything on the results page is relevant. Specifically, you know that the right-hand side of the screen is all AdSense territory, and you aren’t looking for ads. The longer the screen remains blank, the longer your brain has to prepare itself and focus your attention, and the more likely it is that you’ll lock onto the left half of the screen and inhibit the ads. In this context, 500 milliseconds could mean the difference between ad revenue and banner blindness.

In fairness, I should note that there does come a point where the cue isn’t helpful. Attentional cues typically have a Goldilocks zone—not too big, not too small—where they are most beneficial. Cognitive scientists studying these phenomena in labs typically find that Goldilocks is happiest in a zone somewhere well under one second, which would suggest that a longer page load time should create an increase in AdSense revenue (since, according to the research, you should show stronger attention-narrowing effects around 400 milliseconds, not 900). Then again, search engines are not laboratories. This sort of attentional cueing research is rarely done in real-world usage scenarios, and the timings could well be different. There’s also an extremely small (but possibly important) difference between when the browser begins to display the page and when it finishes displaying the page. Maybe the cue isn’t the blank screen. The start of page display could be the cue and the end of page display could be the cue gap. These timings would probably be more consistent with the empirical research, but I can’t know for sure unless I have access to Google’s data. Which could only happen if Google hires me. I’M JUST SAYING.