She Published the Same Article Twice. The Second Title Drove Four Times the Traffic.
She had published the same article twice. The first version ran under a descriptive title that told readers exactly what they would get. It pulled modest traffic for two weeks, then dropped off. Months later, she published the same piece with a different angle title, the kind that withheld just enough information to create tension. The second version drove four times the traffic in the first week.
The difference was not the content. It was the headline.
This kind of split result has been studied at scale. The Upworthy Research Archive, a dataset of experiments run by the digital publisher Upworthy between 2013 and 2015, contains records of more than 8,000 headline variations tested on real audiences during a period when the site was reaching around 50 million readers per month. The dataset is one of the most detailed records of headline performance ever made publicly available, and the findings consistently point to the same underlying variable: information gap.
The concept comes from George Loewenstein's 1994 paper "The Psychology of Curiosity: A Review and Reinterpretation," published in the Psychological Bulletin, which described how the brain experiences a form of discomfort when it perceives a gap between what it knows and what it wants to know. A headline that tells you just enough to make you aware of what you do not yet know triggers that discomfort. Clicking resolves it. The mechanism runs in about two seconds.
The History of Headlines as a Discipline
Headlines were not always an object of systematic study. In the newspaper era, writing headlines was considered a craft rather than a science, passed from experienced copy editors to newcomers through apprenticeship. The yellow journalism of William Randolph Hearst and Joseph Pulitzer in the 1890s demonstrated that sensationalized headlines could dramatically increase circulation, but the measurement was blunt: edition sales, not click-through rates.
The transition to quantitative headline testing happened gradually. Reader's Digest, founded in 1922, conducted extensive reader surveys to understand which articles people found most valuable, which informed their cover line selection. David Ogilvy, the advertising executive who founded Ogilvy and Mather in 1948, argued that the headline was the single most important element of any advertisement, citing its function as the first and often only piece of text most readers would encounter. His book "Ogilvy on Advertising," published in 1983, codified what copywriters had been learning through decades of direct mail testing: headlines needed to communicate a benefit, include specific information, or create curiosity.
The internet created the first environment where headline performance could be measured precisely and at scale. A/B testing, which had been used in direct mail since the 1920s, became standard practice for digital publishers when platforms like Chartbeat and later Upworthy automated the process. An editor could write five different headlines for the same article, publish each to a fraction of visitors, and have a winner identified by click-through rate within hours. The empirical study of headlines became possible at a scale that newspaper editors could never have achieved.
What the Headline Research Complicates
A 2024 study published in Scientific Reports analyzed 8,977 headline experiments from the Upworthy archive and found that the relationship between headline specificity and click-through rate is not linear. Headlines that were vague benefited from becoming more concrete. But headlines that were already moderately specific saw click-through rates fall as they became more explicit. The researchers described this as a curvilinear relationship: the optimal headline gives enough to make the gap feel real but not so much that it closes the gap before the reader clicks.
This is the practical difficulty of headline writing. Too vague and readers do not understand what they would be clicking toward. Too specific and they already feel they have the information. The optimal point shifts depending on topic, audience, and context. A technical audience reads headlines differently than a general interest audience.
The Upworthy analysis also found that emotional resonance affected click behavior. Each additional negative word in a fifteen-word headline correlated with a roughly 2.3 percent increase in click-through rate, up to a point. Headlines that frame a problem, name a mistake, or identify a failure consistently outperform equivalent headlines with neutral framing. The researchers speculated that negative framing triggers a loss-aversion response, making the information gap feel more costly to leave unresolved.
BuzzSumo's analysis of hundreds of millions of social media shares, published in several blog reports between 2014 and 2020, found that phrases like "will make you" and "this is why" consistently outperformed descriptive headlines across platforms. Number-based headlines, such as "5 things" or "23 studies," generated significantly more shares than equivalent non-numbered titles. Research by Jonah Berger at the Wharton School suggests that specific numbers trigger a sense of precision that increases perceived credibility: an article titled "the 27 greatest films" seems more carefully researched than one titled "the greatest films."
SEO Headlines Versus Social Headlines
A headline optimized for social sharing often differs from one optimized for search. Search engine optimization favors headlines that contain the primary keyword phrase in the first few words, use natural language that matches search queries, and fall within 50 to 60 characters to avoid truncation in search results. A search-optimized title for an article about credit card debt might read: "How Long to Pay Off Credit Card Debt with Minimum Payments."
A social-optimized title for the same article might read: "She Paid the Minimum for 27 Years and Spent $9,000 in Interest on $6,000 of Debt." The second version is longer than the SEO optimal range and does not front-load the keyword phrase. But it creates a specific, vivid information gap that a social reader in a news feed will feel compelled to resolve.
Choosing between these approaches requires knowing where the traffic will come from. A blog post that primarily attracts visitors through organic search should prioritize the search-optimized title. A post that will be promoted through social media and email should use social framing, sometimes with the keyword title used as the meta title separately from the displayed heading. Many content management systems support different titles for on-page display and for the HTML title tag, which is what search engines index, enabling both strategies simultaneously.
Platform Differences in Headline Performance
Research by LinkedIn in 2020, analyzing engagement across professional posts, found that headlines using second-person framing ("you") outperformed third-person equivalents by approximately 21 percent on that platform. LinkedIn's audience is in professional mode, and direct address emphasizes personal relevance.
Email subject lines follow somewhat different rules. Personalization tokens and questions outperform statements. Short subject lines, under 50 characters, tend to perform better on mobile where the preview is truncated. Capitalization of the first word only performs better than title case in many studies, because it reads as a message rather than a headline.
Conclusion
The hardest part of headline writing is that you are too close to your own content to judge the information gap from the outside. You already know everything the article contains, so the gap does not feel like a gap to you.
ToolHQ's Title Generator offers angle variations built around the information-gap principle, negative framing, and platform-specific conventions, giving you options to evaluate with fresh eyes rather than working from the single framing you had when you first sat down to write.
Frequently Asked Questions
What makes a blog title get more clicks?
Research from the Upworthy archive shows that information gaps drive clicks. Headlines that reveal enough to make a reader aware of what they do not know, without closing that gap entirely, consistently outperform fully descriptive titles.
How specific should a blog title be?
A 2024 Scientific Reports study found a curvilinear relationship: too vague or too specific both reduce clicks. The optimal title conveys enough context to feel relevant but withholds enough to create curiosity.
Do negative words in headlines increase clicks?
Upworthy research found each additional negative word correlates with roughly a 2.3 percent increase in click-through rate, up to a point. Headlines framing a mistake or problem outperform neutrally worded equivalents.