LaunchRanked
Launch free

Headline analyzer

Paste a headline, page title, tagline or Show HN title. AI scores it on clarity, specificity, benefit and hype, then suggests rewrites you can test.

0 / 150

What is it?

Adds an audience-fit score.

Rewrites only use facts from your headline and this box.

Scores by Jev 1.13 (TypeSafe’s decision model) via OpenRouter; rewrites by Claude Haiku 5.5 from Anthropic (Meta Llama 3.3 70B on Workers AI steps in if it’s unavailable). Scores are probabilities about how the headline reads, not predictions of clicks or rankings. We don’t save what you enter.

Enter a headline and pick what it’s for. You’ll get an overall score, a score for clarity, specificity, benefit and hype, quick stats, and five rewrites scored the same way.

What the scores mean

Every headline is scored on the questions a first-time reader answers in a second or two. Each answer is a probability over four steps, from worst to best, so you see both the most likely label and how sure the model is.

  • Clarity. Would a stranger know what it is, or what the article is about, from the headline alone? This carries the most weight for every type.
  • Specificity. Concrete nouns and details (“invoicing for freelance designers”) beat words that fit any product (“the all-in-one platform”).
  • Benefit. Does it say what the reader gets? It counts for more on taglines, ads and email subjects, and very little on Show HN, where HN readers want to know what you built.
  • No hype. Superlatives, exaggeration and clickbait lower this score. Higher means plainer.
  • Audience fit appears when you name an audience, and HN guideline fit when the type is Show HN.

The overall score is the weighted average of those bars, minus a deduction when the headline likely makes claims that are untrue or can’t be checked.

What makes a headline clear

Most weak headlines have the same problem: they describe how the writer feels about the product instead of what it is. “Revolutionary AI app that changes everything” could be anything. “Postgres-backed job queue in 300 lines of Go” tells a developer exactly what they’d get. A few habits help:

  • Name the thing and what it does, in the words your reader would search for.
  • Keep one real detail that sets it apart: who it’s for, how it works, a number you can back up.
  • Drop adjectives that only say it’s good. If it’s fast, say how fast.
  • Fit the space. Google shows roughly 600 pixels of a title on desktop, HN takes 80 characters, and Google search ad headlines take 30.

Using the rewrites

You get up to five rewrites from different angles, scored with the same questions and sorted best first. Use them as drafts: keep the angle you like, then put your own voice back in. “Analyze this one” runs a rewrite through the tool again so you can keep iterating.

For more options on one job, the title generator writes SEO titles, the tagline generator writes short product lines, and the SERP preview shows how a title looks in Google. Posting on Hacker News? The Show HN title checker checks HN’s written rules as you type.

Questions

Does a high score mean more clicks?

No. The scores say how the headline reads: whether a stranger would understand it, how concrete it is, whether it says what the reader gets and how much hype it carries. Clicks also depend on where it appears, what sits next to it, the audience and timing. We have no click data, and the tool doesn't predict it. A headline that scores well is a clear one, which is a good starting point for your own tests.

Who does the scoring?

Jev 1.13, TypeSafe's decision model, through OpenRouter. It doesn't write text. It answers set questions (how clear is this, how specific, how much hype) with a probability for each step on a four-step scale, and we combine those into the bars and the overall score. The weights are shown next to each bar. The rewrites are written by Claude Haiku 5.5 (from Anthropic) and then scored by Jev with the same questions.

Why did my headline lose points for untrue claims?

Jev also estimates the probability that the headline makes claims that are likely untrue or unverifiable, such as "changes everything" or "the fastest in the world". Up to 35% we take nothing off. Above that the deduction grows, to 25 points at 100%. If your claim is true and you can back it up, say how: a benchmark, a number or a named result reads as proof rather than hype.

Can the rewrites make things up?

They're told to use only what's in your headline and the "what the product is" box. We also drop any rewrite that contains a number you didn't give us, and Jev checks each rewrite for facts that weren't in your input. Rewrites that are very likely to add something are removed, and borderline ones carry a warning. Still read every rewrite before you use it.

Is it the same as the Show HN title checker?

Pick Show HN as the type and you get the same HN guideline score the Show HN title checker uses, plus benefit, stats and rewrites. The Show HN checker adds instant checks for HN's written rules (the prefix, the 80-character limit, capitals, emoji) as you type.

Free launch

Launching something? Get a page that ranks.

Launch free on LaunchRanked: a permanent, SEO-ready product page, a weekly leaderboard slot and a followed link while our badge is on your site.

Launch freeSee Autopilot

Already live? See what your site is missing on Google, free.

Related free tools

See all free tools