The Quest for the Perfect Headline

The best headline ever written is “Headless Body in Topless Bar.” It appeared on the front page of the New York Post in 1983—simple, symmetrical, and intriguing. Five words that tell a story but still compel you to find out more.

Today, you’d probably never see that headline, at least not online. Most publishers now favor either conversational titles or listicles, ideally with some sort of curiosity gap. If Upworthy, the master of the curiosity gap, wanted to make the same story go viral, it would’ve rolled out something like “A Man Was Shot in a Topless Bar. What Happened Next Is Even Crazier.” BuzzFeed might’ve gone with “11 Reasons This Was the Craziest Murder of the ’80s.”

A lot of people would click on those. If I came across them at the right time, I would too.

For better and worse, headlines are the movie trailers of the Internet—and like trailers, many of today’s titles look and sound the same. What was once a creative art has become a science concerned with maximizing engagement by any means necessary.

“A lot of online headlines are injected with things you must do, things you must not do, and it exploits insecurity,” said Patrick Burke, who worked as the deputy editor of digital applications at the New York Post for three years before becoming the senior editor of CIO Insight in 2015. “What I appreciate the most is when headlines honor the reader, the reporter, the subject matter, and—at the same time—the web crawlers that help get that story out more.”

But publishers may not care much about honor in an era in which they’ve had to battle for survival, especially since following headline patterns and templates has helped some sites grow at unprecedented rates. In 2013, Upworthy surged to 90 million readers in November 2013 behind the appeal of conversational clickbait that took off thanks to Facebook’s algorithm. ViralNova hit 100 million. Last month, BuzzFeed registered 177 million uniques, according to Quantcast.

“There’s an overwhelming amount of content being published every day, and there’s not as big of a demand for the amount that’s published,” said Nathan Ellering, content marketing lead for CoSchedule, a social media technology company. “What is the way to stand out for people who have absolutely no time to read every single headline? It’s just to cut to the chase. It’s a completely different playing field now.”

But the evolution of Facebook has complicated that playing field. Over the last few years, publishers have had to rethink the way they approach headlines. In 2014, a leaked innovation report revealed that homepage visitors to The New York Times dropped from 160 million to 80 million between 2011 and 2013 while the number of pageviews had held steady. Readers don’t necessarily find content from homepage headlines, coming instead from social media feeds and email newsletters. According to data from Shareaholic, Facebook drives at least 25 percent of all traffic to publisher sites, a figure some media analysts think could be closer to 50 percent once you account for unrecorded referral traffic from the Facebook app.

“What we’re seeing is a fragmentation of the whole idea of what a headline is.”

As a result, publishers are tasked with finding the perfect headline for every distribution channel—or at the very least, the one best programmed for maximum virality. It’s not uncommon for one article to get dozens of “headlines” if you add up all of the email subject lines and pithy descriptions posted on Facebook, Twitter, and LinkedIn. What works on one platform won’t necessarily connect with users elsewhere; in that sense, the perfect headline as we once knew it may not even exist anymore.

“I think what we’re seeing is a fragmentation of the whole idea of what a headline is,” said Scot Petersen, a former editor for the Boston Herald and PC Week. “A headline is still important, but it’s read differently.”

In some cases, headline creation has even been reduced to formulas. There are a number of tools out there that will score your titles based on algorithms and provide instant feedback on how you can improve.

The most popular of those tools—or the one with the best SEO, at least—is CoSchedule’s Headline Analyzer. In addition to offering a headline score on a scale of 1-100, the feature also tracks factors like word balance, character count, word count, keywords, sentiment, and style (generic, listicle, how-to, etc.). According to Ellering, people spend an average of five and a half minutes testing headlines when they use the analyzer.

CoSchedule’s algorithm, which culls data from over a million digital headlines, operates under a few key assumptions that are supposed to lead to more shareable content. Ideally, titles should consist of nine words and 55 characters. They should convey positivity, is possible. (If not, negativity ranks better than neutrality.) They should also aim for a specific balance: 20-30 percent for common words (how, what, why); 10-20 percent for uncommon words (awesome, world, baby); 10-15 percent for emotional words (wonderful, valuable, worry); and at least one power phrase (what happened to, will make you, in the world).

Scores over 70 are considered good; between 60 and 69 is solid; anything below 60 needs improvement. (For context, “Headless Body in Topless Bar” only gets a 46.)

The tool is hellaciously addictive, although it’s also possible to abuse the algorithm for better results. In fact, some of the highest scores come from headlines that don’t make any sense. One night, Contently’s editorial team spent an hour competing to see who could get the highest score. The winning entry, which got a 91, was “How to Be the Best Baby in a World Full of Babies.” Unfortunately, we’ve yet to find a story that works for that subject.

Point is, striving for the perfect headline is a dangerous exercise, even if you avoid gibberish and get more social shares in the short run. If all publishers use the same tools and check off the same boxes in a never-ending game of escalation, then won’t all headlines just blur together? Even The New York Times, known for its buttoned-up style, has resorted to social media updates that resemble clickbait. (On article pages, however, the Times has mostly stuck to muted titles that are interrupted by stiff commas.)

When you expect to see a shocking tabloid, it’s really not that shocking or interesting anymore. There are only so many times a news outlet can tell readers they won’t believe what happens next.

Readers (and algorithms) also seem to be developing an immunity to these tricks. If users click a headline and bounce right back to Facebook, for instance, the publisher is downgraded by Facebook’s algorithm; if they stay on the site for a few minutes, Facebook rewards that publisher with a boost.

Upworthy, once the viral king, now serves as a cautionary tale for the problems with clickbait headlines. By 2014, the site’s traffic had dropped from 50 million to 20 million due to Facebook’s algorithm change, Google’s crackdown against unoriginal content, and widespread exhaustion from headlines like “You Won’t Guess How One Ingredient In Your Doughnuts Could Be Leaving Thousands of People Homeless.” (A real title, in case you’re wondering.)

In some ways, it’s the modern equivalent of passing the National Enquirer in the supermarket—when you expect to see a shocking tabloid, it’s really not that shocking or interesting anymore. There are only so many times a news outlet can tell readers they won’t believe what happens next.

Today, Upworthy’s monthly readership sits under 13 million, per Quantcast, despite recent efforts to produce original content about social good. By the time Upworthy’s executives tried to switch their approach, the clickbait collapse was already underway. Upworthy’s reputation was significantly damaged, and now, the publication is a media-industry punchline.

So what does happen next? Is there any way for publishers to create headlines that are simultaneously appropriate and entertaining?

Perhaps—as long as they produce content that matches the headline. As Ben Smith, BuzzFeed editor-in-chief, wrote two years ago: “The only real trick is that the work has to be good. And the only thing, other than mediocrity, that can really sabotage this strategy is writing a headline that overpromises and a story that underdelivers.”

Ellering believes headlines will evolve for the better by becoming more straightforward and less sensationalistic. “There’s a lot of data out there that says list posts do really well, but what’s happening is people are sick of the ’10 Ways’ format,” he said. “I think that list posts will get more mature.” To accomplish this, he expects to see longer, more robust lists in the future. Think of listicles full of 99, 100, 150 tips.

That tactic may work for certain publications, but those that have to cover the news cycle won’t be able to only focus on longform listicles that rank well in search.

“Organizations are trying to appeal to as many people as possible, but that sometimes brings out the lowest common denominator. I hope they evolve in a more intelligent way that respects the reader more,” Burke said. “Or maybe they’re just going to evolve into symbols of dollar signs and celebrity butts. Time will tell.”

After all, “Headless Body in a Topless Bar” was five words of perfection. But as a string of emojis? Well that could lead us on a whole new quest altogether.