Short answer: a headline’s job is not to be interesting, it is to let someone decide within a second whether this page is about their problem. Naming the subject and the outcome does that. Intrigue makes them lean in, which only helps if you were going to deliver something anyway.
The filter, not the hook
There is a persistent confusion between two different jobs, and most bad headlines come from mixing them up.
A headline is a filter. Its purpose is to stop the right reader and let everyone else go. That is why “Our Services”, “Why Choose Us” and “Get In Touch” are among the most expensive words on a website: they are not filters at all. They are labels on the furniture. Every visitor passes them, so they have told nobody anything.
A hook is a different device, and hooks work on an audience already reading. Reading research on the web is unambiguous about how little of a typical page gets read in sequence, which is the whole reason the filter approach wins. [1] Most of the time you do not have that audience. Someone landed from a search for a specific problem, they are in a hurry, and they are deciding whether to keep going.
Key takeaway: write the headline for the person who has just arrived and is scanning, not for the person who is already reading.
What actually earns its place
Looking at headlines that work, the useful ones share a small number of traits, and none of them are unusual — they are simply what good descriptive writing already does. [2]
- A named subject. “Fixed-price boiler replacement” not “Total heating solutions”. The reader can tell in three words whether they are in the right place.
- A real constraint. Location, timescale, price basis, or scope. Constraints are what make a page findable and worth reading, and they are usually the part a competitor left out.
- A concrete outcome. What the reader ends up with, described as something they will recognise.
| Headline | What the reader learns |
|---|---|
| ”Our Services” | Nothing |
| ”Heating and plumbing services in Kent” | Area and trade, no differentiator |
| ”Fixed-price boiler replacement in Kent, including removal” | Trade, area, pricing basis, scope |
| ”The Secret Sauce of Search” | Nothing until they read it |
The fourth one is not bad writing. It is just wasted on a first impression, because a visitor who cannot tell what the page is about cannot decide to read it.
Be suspicious of your own numbers
Headlines with numbers in them have a specific problem: the reader treats them as claims rather than as decoration, and will hold you to them.
“Increase bookings by 40% in 90 days” is the kind of headline that generates a conversation about whether 40% is right, which is a worse first impression than no number at all. If you have measured something and can stand behind it, the number is genuinely strong and you should use it. If you have not, use something else — a scope, a location, a timescale, a price basis. Constraint works as well as magnitude and does not require you to be right about a percentage.
This is a general rule about the whole site: a number you cannot source is worse than no number, because it costs you the credibility of every true number nearby. Building a justification web covers the reasoning.
Clarity is the citation strategy
There is a persistent claim that headlines need optimising for AI in some technical sense. There is no such mechanism. A heading has no property that makes a system cite a page, and anyone selling one is selling an unverifiable promise.
What is real is much simpler. When something summarises your page, it has to work out what the page is about, and it does that from your headings. A descriptive heading gets that right without effort. A metaphorical one makes it guess. So clarity buys you accuracy in summaries — the same reason it buys you comprehension from a human in a hurry. One habit, two audiences, no special technique.
This also applies to the search results title, which Google may rewrite from your page title and meta description. [3] There is no way to force it and no benefit from fighting it; a clear title is usually the version the search engine chooses anyway. Featured snippets have done this for years, lifting the sentence that answers the question. [4]
The rest of the headings are not headlines
Your H1 filters. Your H2s tell someone whether the answer they need is on this page. Your H3s signpost inside a section. Treating all of them as headline opportunities produces a page that scans as a wall of promises, which is a worse experience than one with four clear headings and no cleverness at all.
Two habits are worth more than any headline rule:
- One idea per heading. If a heading needs a conjunction to make sense, it is two headings.
- Make the headings the table of contents. This is the single most useful structural habit on a content site. [5] Someone skimming should be able to see from your H2s alone whether their question is answered here. If they cannot, the page needs restructuring, not better adjectives.
What to do first
- Open your five most important pages and read only the H1 of each. Could a stranger tell what the page is about? Rewrite any that answer “no”.
- Replace every instance of “Our Services” and “Why Choose Us” with the actual subject and the actual constraint.
- Delete any number you cannot source.
- Read your H2s in sequence and check they form a sensible table of contents.
- Watch for the week after a change: an immediate traffic drop means the headline promised more than the page delivered.
The test is simple and does not require analytics: show your headline to someone who has not read the page and ask what they think it is about. If they are right, it is doing its job.
How this connects
- Capture and book covers what happens once the headline has done its work.
- Clicks versus direct answers covers why clear claims carry into summaries.
- A justification web covers putting evidence next to the claim the headline makes.
- Direct directives covers the last line of the page, which has the same job as the first.
