Medium Headline Tips: The Formats Linked to More Claps

“Stop…” and “Don’t…” headlines had the strongest positive association with reaching 10 or more claps on Medium: 44% reached that mark, compared with 35% of other headlines. That was 1.3× the comparison group’s rate.

We call reaching 10 or more claps traction throughout this article. Number-led titles and first-person “How I…” headlines also reached traction more often than their comparison groups, while colon headlines and titles mentioning a year reached it less often.

For writers choosing a headline, those distinctions offer something more specific than a list of familiar title formulas. Some styles had positive associations, some had negative associations, and others had no statistically reliable difference.

“Stop…” and “Don’t…” led the positive comparisons

The strongest positive result belonged to an uncommon style. “Stop…” and “Don’t…” headlines accounted for 1% of headlines, with a traction rate of 44% against 35% for everything outside that group.

Consider “Stop Giving Your Customers So Many Options”, which received 640 claps. Its author had 249 followers. The title makes the instruction explicit: stop offering so many choices.

Another example, “Stop Asking Customers If They Want to Buy”, received 616 claps; its author had 88 followers. Both titles address the reader with “Stop,” followed by the action the headline argues against.

The group also counts headlines that open with “Never”, such as “Never Tell a Woman She Can’t Do Something”, which received 1,193 claps; its author had 249 followers.

These stories illustrate the style rather than establish a rule. The reason to consider the format is the group-level comparison: it had the strongest positive association, despite being uncommon.

If you are comparing possible openings for an article that argues against an action, a “Stop…” or “Don’t…” version deserves consideration on that basis. Its low share of headlines and its higher traction rate are separate facts; the recommendation comes from the rate comparison, not from rarity itself.

Number-led titles and “How I…” also ran higher

Headlines that started with a number reached traction 43% of the time, compared with 37% for other headlines. That was 1.2× the comparison rate. Number-led headlines represented 4% of the total.

The examples make this category easy to recognize. Each starts with a number and names what the reader will encounter:

• “5 Manipulation Tactics That Make You Think the Decision Was Yours” received 1,047 claps; its author had 182 followers.
• “7 Warning Signs You May Be Ignoring From Your Body” received 908 claps; its author had 247 followers.
• “5 Types of Fruit That Can Harm Your Health If You Eat Too Much” received 898 claps; its author had 716 followers.

The titles concern manipulation tactics, bodily warning signs, and fruit consumption. What puts them in the same headline category is their opening structure, not a shared subject.

Correlation is not causation: a headline pattern often goes with a topic or format, so the association is not purely about the wording. The individual examples illustrate the formats, rather than prove that those openings produced their clap counts.

First-person “How I…” headlines had a positive association as well. They reached traction 41% of the time, versus 35% for other headlines—1.2× the comparison rate—and appeared in 4% of headlines.

Keep the first-person wording in view when applying that result. “How I…” belongs to the positive group; the more general “How to…” construction belongs to the group without a statistically reliable difference. They should not be folded into a single recommendation to use a “how” headline.

For your own shortlist, number-led and “How I…” titles are therefore additional formats to consider alongside “Stop…” and “Don’t….” Each recommendation rests on its own comparison with other headlines.

The most common pattern reached traction less often

Two-part headlines with a colon were the most common pattern, appearing in 23% of all headlines. Their traction rate was 31%, compared with 38% for headlines outside that category—0.8× the comparison rate.

That is a clear distinction between popularity and performance. A familiar-looking title structure can be common without having a higher rate of reaching the clap threshold.

If your reason for adding a colon is that it looks like an established Medium headline format, separate that preference from an expectation of more claps. The prevalence supports calling it common; the traction comparison does not support calling it an advantage.

Headlines mentioning a year also ran lower. They reached traction 26% of the time, versus 36% for other headlines, or 0.7× the comparison rate. Year mentions appeared in 4% of headlines.

Keep the category’s wording precise here, too: this result concerns headlines that mention a year. It is not a result about a particular opening phrase or the placement of the year within the title.

“Why…,” questions, and “How to…” had no reliable edge

Several familiar headline styles had observed differences that were not statistically reliable. These results belong in a separate group from the positive and negative associations above.

• “Why…” headlines: used in 3% of headlines, with a traction rate of 38% versus 35% for other headlines.
• Question headlines: used in 8% of headlines, with a traction rate of 39% versus 37% for other headlines.
• “How to…” headlines: used in 3% of headlines, with a traction rate of 33% versus 36% for other headlines.

Do not read the higher observed rates for “Why…” and question headlines as dependable advantages. Equally, the lower observed rate for “How to…” is not a dependable disadvantage. None of these comparisons provides a statistically reliable reason to favor or avoid the format.

When reviewing the detailed table, keep that distinction attached to the rates. The table lets you inspect the categories together, but the reliability of these comparisons remains part of the result.

Pattern	Share of headlines	Reached 10+ claps	Other headlines	Reliable?
“Stop…” / “Don’t…” headlines	1%	44%	35%	yes
Headlines that start with a number	4%	43%	37%	yes
First-person “How I…” headlines	4%	41%	35%	yes
“Why…” headlines	3%	38%	35%	not yet
Question headlines	8%	39%	37%	not yet
“How to…” headlines	3%	33%	36%	not yet
Two-part headlines with a colon	23%	31%	38%	yes
Headlines that mention a year	4%	26%	36%	yes

Length offers a comparison point, not a formula

The median Medium headline was 9 words long. Across the length bands, the best-performing band reached traction 38% of the time, while the longest band had the lowest rate, 33%.

The median and the traction comparison answer different questions. The median describes the typical headline length; it is not itself a recommended target. The rates describe how often headlines in different length groups reached the threshold.

Use length as another comparison when reviewing title options, rather than letting the typical word count dictate the title. For details on how we measured these comparisons, see our methodology.

What to do with your next Medium headline

• Consider “Stop…” or “Don’t…” when comparing suitable openings. This category had the strongest positive association with traction.
• Include number-led and first-person “How I…” alternatives. Both reached traction more often than their respective comparison groups.
• Do not select a colon or year mention expecting a clap advantage. Both categories had lower traction rates, even though colon headlines were the most common pattern.
• Keep “Why…,” questions, and “How to…” out of the winner-and-loser lists. Their differences were not statistically reliable.
• Use length as a comparison, not a prescription. Typical length and traction by length are distinct measures.

How we measured this

Pattern detection uses simple, transparent rules on 18,639 headlines. Each pattern is compared with all headlines that do not use it, standardised to the same author-audience mix.

Limitations.

• A headline pattern often goes with a topic or format, so the effect is not purely the wording.

Data: WriterPulse scan of 2026-10-20. Full details in our methodology.