A headline has one job, and it is not to be admired. It has to earn the next line. Everything else — the wit, the rhythm, the allusion — is decoration on top of that job, and decoration that costs the reader's attention is a net loss.
Most writing advice about headlines is about cleverness. The useful advice is about specificity. A specific headline almost always beats a clever one, because specificity gives the reader a reason to continue that does not depend on them sharing your sense of humour.
The specificity test
Take a vague headline and ask what it actually promises.
How to Improve Your Writing
This promises nothing. Improve in what way? For whom? Compared with what? The reader has no reason to believe the article knows more than they do.
Seven Edits That Make a Draft Sound Finished
Now the promise is concrete: seven specific edits, and a result the reader can recognise. The reader can decide whether they want that result.
The test is simple. Read your headline and ask: what will the reader be able to do after reading that they cannot do now? If the answer is vague, the headline is vague. If you cannot answer at all, the article may be vague too, and the headline is telling you something useful.
Verbs beat nouns
Compare these pairs.
The Importance of Proofreading Why Proofreading Catches What Grammar Tools Miss
The first is a topic. The second is an argument. Arguments get read; topics get skipped. The difference is usually a verb and a specific claim.
This does not mean every headline needs a verb. It means every headline needs a claim. "The Quiet Cost of Wordiness" has no verb and still makes a claim. "Wordiness" does not.
A useful exercise: write the headline as a sentence first, then compress it. "Grammar tools miss errors that only a reader catches, and that is why proofreading still matters" compresses to "Why Proofreading Catches What Grammar Tools Miss." The sentence forces you to know what you are arguing. The compression is the craft.
Length and the character limits
On the web, headlines get truncated. Search results cut around 60 characters, social cards around 70, and email subject lines around 45 on a phone. A headline that only works in full is a headline that fails in most of the places it appears.
Count characters as well as words. The character counter is the right tool for this, and the word counter helps with the body copy underneath.
The practical compromise is to put the essential meaning in the first 45 characters and the interesting detail after. "The Three Edits That Fix Most Drafts" survives truncation. "How I Learned to Stop Worrying and Love the Three Edits That Fix Most Drafts" does not.
Subheads are promises, not labels
A subhead does a different job from a headline. The headline earns the click; the subhead earns the next paragraph. Readers scan subheads to decide whether to read the section, so a subhead that only labels the topic ("Introduction," "Background," "Conclusion") gives them nothing to decide with.
Compare.
Label: Common Mistakes Promise: Three Mistakes That Make a Draft Sound Unfinished
Label: Pricing Promise: What the Price Includes, and What It Does Not
The promise version tells the reader what they will know at the end. That is the information they need to keep going.
Subheads should also make sense read alone, in sequence, as a summary of the argument. This is the oldest test in the book and still the best: cover the body text and read only the subheads. If they tell a coherent story, the structure is sound. If they read like a table of contents for a different article, the structure needs work.
The pattern that works for most articles
Most useful articles follow one of four headline shapes. Knowing them speeds up the writing and keeps you from reaching for a headline that promises something the piece does not deliver.
The how-to. "How to Cut Word Count Without Losing Meaning." Promises a method. Requires the article to actually have steps.
The number. "Five Sentences That Signal a Weak Argument." Promises a finite, scannable list. Requires the number to be accurate; readers count.
The correction. "Why the Passive Voice Is Not the Problem." Promises to overturn a belief. Requires the article to name the belief and address it directly.
The question. "Should You Use the Oxford Comma?" Promises a decision framework. Requires an answer, not a survey of opinions.
Each shape sets an expectation that the body must meet. Choosing the shape before writing is a way of choosing what the article is.
What to avoid
The tease. "You Won't Believe What This Editor Found." Curiosity is borrowed attention; it has to be repaid immediately and it rarely is.
The pun that hides the topic. Clever headlines work when the reader already knows what the article is about. On a search results page, they do not, and the click goes elsewhere.
The vague superlative. "The Ultimate Guide to Everything." A guide cannot be ultimate, and readers know it.
The question with no answer. A headline that asks a question the article never answers is a broken promise, and readers remember.
The final test
Before you publish, read the headline as if you were the most impatient reader on the internet. Would you click? If the answer depends on you already caring about the topic, the headline is not doing its job.
Then read the subheads and ask a second question: does each one give me a reason to read the section underneath? If any subhead could be replaced by the word "Stuff," it needs rewriting.
Headlines and subheads are the cheapest edits in any draft. They cost minutes and change whether anyone reads the expensive parts.