Google’s RNC ‘Trends’ Didn’t Mean That Much—Again

When everything is trending, is nothing trending? Are all newsy “spikes” equal or are some spikes spikier than others? When it comes to a big event like the Republican National Convention, it can be hard to tell what’s true and what’s not. Covering the RNC is hard—more than 15,000 credentialed journalists were in Cleveland, all vying for a tiny slice of differentiation in their take, with thousands of others aggregating content from afar.

Amid that frenzy, if journalists are offered ready-made news nuggets with convenient data points, they become hard to resist. Every spike becomes a story, and more popular than most are the data nuggets offered up by Google Trends, often showing spikes in interest on an hour-by-hour basis around major news events. And if you were looking for RNC trend stories, data without the context was on the menu again, some of it dug out by media outlets themselves, others served up by the Google team.

The main event on Thursday night was the father-daughter pairing of Donald and Ivanka Trump. It seems pretty natural to assume that her name might spike in searches as she took to the stage to introduce a soft, gentle avatar of her father. But the level to which her name spiked was made out to be like a SpaceX launch; the reality was maybe less dramatic.

Searches for @IvankaTrump have overtaken searches for @realDonaldTrump https://t.co/glMUYyjZML pic.twitter.com/Fiy8A4DLGd

— GoogleTrends (@GoogleTrends) July 22, 2016

Put beside trends for her father, over time, searches for Ivanka peaked briefly before dropping off again. In this case, reporting the spike wasn’t disingenuous, rather premature, and not all that interesting. It’s pretty predictable stuff, unless you inflate it to be something that seems more impressive than it is. It wasn’t the only surge touted that, with the proper context, was easily rendered pretty meaningless.

Controversial tech titan Peter Thiel was one of the speakers, announcing amid an otherwise bland speech that he was “proud to be gay, and proud to be a Republican.” Thiel is a huge figure in the dot com world, but if you’re not up to speed on internet executives, you may not know who he is, and hence resort to Google. So a spike like the one described below may suggest that Thiel is suddenly very much worth searching for, a real internet search hot property! Get searching before there’s nothing left to search! All the kids are searching for Thiel right now!

+1,100% spike for searches for @PeterThiel in past hour

— GoogleTrends (@GoogleTrends) July 22, 2016

Here’s the context. Thiel’s search spike was short-lived, and rose from a comparatively low base—which is an important concept to grasp. Something spiking by 1,100 percent should be huge! Imagine if your rent spiked by 1,100 percent—that would be insane! But if one cent spikes by 1,100 percent, you have…11 cents. Where you start from is important. Again, when you put it in the context of other searches, it seems smaller, and not unexpected—someone who’s well known but not Trump-size-famous gets a national platform on the biggest night in politics so far this year, people are going to want to know who the hell he is and why he’s there.

Let’s do one more. Apparently, searches for Jerry Falwell spiked by 2,300 percent on the last night of the convention. Damn, that’s a lot of searching! Praise the lord! The question is: where did that spike rise from? Were people searching for Jerry Falwell at all, or is this a rise from one single search per hour to 2,300 in one hour? We had a quick look at the Google Trends and found that prior to the spike, Jerry Falwell was searched for about half as much as early 90s sitcom “Saved By The Bell.”

Does that mean anything? Maybe—if you know how many people are searching for Saved by The Bell, or even care to ask the question. It’s important for journalists who care about putting data about politics in the proper context, which seems more important than ever given the amount of terrible inaccuracies that surround the current election process.

Data analyst Danny Page certainly seems to care—he penned an angry but accurate screed recently entitled “Stop Using Google Trends,” in which he castigated their team and journalists at large for spitting out data bereft of context. Page used the example of all the stories about a spike in U.K. residents searching “What is the EU?” after the U.K. voted to leave the EU in a referendum.

Said Page:

“The peak was merely ~1000 people! It’s ludicrous that so few people get turned into a massive story, but it underscores the need for context.”

Simon Rogers, Google’s data editor, responded to that post and agreed that putting trends data in context is useful to understand how any given spike or surge stacks up against related search volumes.

Those “spikes” are a sudden acceleration of search interest in a topic, compared to usual search volume. We know these are interesting because they are often reflective of what’s going on in the real world … To get a sense of relative size, we can add additional terms, which helps put that search interest into perspective.

That last part is really important, but rarely followed through on. A look back at the Google Trends twitter account shows that spikes, when they’re presented to their 112,000 followers, are rarely given any context at all. And news outlets are using that data more and more often. That’s got the potential to be problematic. Or maybe it’s just up to editors to realise that if a reporter comes to you with a quick-hit story idea built around a “spike” with no context, there’s only one thing to do.

Spike it.