Headline craft is one of the oldest disciplines in publishing. The form has evolved — wire-service ledes, magazine cover lines, newspaper rack toppers, web headlines, social-share titles — but the underlying job has not. A headline earns the reader's attention and accurately previews what follows. Get that right and the piece compounds. Get it wrong and the writing underneath never gets read.
This is a working reference on what actually makes headlines work, across the platforms where they now have to perform.
The Twitter-optimized headline era
Around 2010, a new constraint reshaped headline craft for online publishers. Twitter's 140-character limit meant a headline had to fit inside a retweet — with the user's handle, the "RT @" prefix, and a "via @" attribution all consuming characters. Get the headline tight, keep it shareable, and a piece compounded through the network. Miss the constraint and the piece never traveled.
Brian Clark's Copyblogger headline formulas — published in 2007 and widely shared through the 2010s — defined the era. "How to" structures, list patterns ("10 Ways…"), question framings ("Are Your Titles…"), and number-anchored claims ("3 Reasons…") were the building blocks. They worked because the retweet was the test, and the retweet had to fit inside the character constraint.
The discipline was unforgiving. A 140-character headline left no room for filler. Every word had to earn its place. The publishers that internalized the constraint produced sharper, faster headlines than they had produced in the longer-form web era that preceded it.
The 280-character era and the slow decline of retweet distribution
Twitter doubled the character limit in November 2017. Most publishers loosened headline discipline in response. The retweet still worked, but the cap stopped enforcing brevity. The trade-off was visible: longer headlines carried more context but lost the punch that made the 2010-era titles travel.
By 2020, Twitter's algorithm had begun deprioritizing tweets with external links — the platform was optimizing for in-feed engagement, not outbound traffic. Publishers watched referral traffic from Twitter decline through the early 2020s. The platform that defined a meaningful chunk of headline craft for a decade stopped functioning as a primary distribution layer for most categories of publishing.
The current headline environment
Headlines now live across multiple surfaces simultaneously — the publisher's own site, search results, social platforms (X, LinkedIn, Threads, Bluesky), newsletter aggregators, content recommendation widgets, and creator-amplification networks. Each surface has its own conventions and its own optimal headline patterns.
The publishers that perform well across the working set tend to produce headlines that share a few characteristics.
They name something specific. Named brands, named people, named numbers, named places, named time frames. Generic headlines underperform across every platform. Specific headlines earn attention because they tell the reader what they will actually find.
They match how readers actually search and ask. The headline that mirrors the natural-language pattern of a reader's question — "best [category] in [city]," "how to [outcome]," "what happened with [event]" — outperforms the headline that doesn't.
They keep the reader's promise. A headline that overpromises against a thin body fails the same way in 2026 as it did in 2010. The reader leaves. The publisher loses trust. Subsequent headlines from the same source get discounted.
They are credible. A claim that can be checked, a named entity that exists, a number that survives audit. The readers that publishers compound with are the readers who can tell when a headline is bullshitting. Avoiding that perception is the foundation of every long-term publishing brand.
What didn't change
Three rules from the 2010 era still apply to current headline craft.
Make titles relevant. A headline that overpromises and underdelivers fails consistently. The reader leaves. The publisher's brand erodes.
Make titles specific. "10 Ways to Peel Potatoes" beat "Some Cooking Tips" in 2010 because it named the count, the action, and the object. The same specificity wins now.
Make titles credible. A claim that can be checked, a named entity that exists, a number that survives audit. The headlines that earn trust are the headlines that deliver what they promise.
The modern headline stack
The headline craft that compounds has four moves.
Match how readers actually search and ask. Mirror the natural-language pattern of the reader's question. "Best [category] [year]" works. "Top 10 [thing] in [city]" works. "How to [outcome]" works. The question and the headline should rhyme.
Front-load the named entities. Brands, people, categories, years, methodologies. The first six words carry the weight. Burying the named entity in the back half of the headline weakens the piece.
Make the value proposition specific. Numbers, scope, methodology, time frame. "PR Industry Survey: 2026 Results" beats "Our New Study."
Structure the piece so the headline's promise is recoverable in the body. The reader who clicks expects the headline to be honored within the first few paragraphs. A headline that overpromises against a thin body damages the publisher's brand across all subsequent work.
The bottom line
The 2010 question was whether your title fit inside a retweet. The question now is whether your title performs across the working set of surfaces where it has to live — search, social, newsletter, the publisher's own site, the broader recommendation infrastructure. The medium has evolved. The discipline — specific, relevant, credible, structured — has not.
The journalists and publishers who internalize the discipline compound their authority across years. The ones who treat headlines as an afterthought watch their work underperform regardless of how strong the writing underneath actually is.
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
