Header tags are the HTML elements H1 through H6 that give a page its structure. The H1 states what the page is about, H2s divide it into major sections, and H3s break those sections down further. They matter three times over: search engines read them to understand how content is organized, screen reader users navigate by them, and everyone else scans them before deciding to read at all. This post covers what each level does, what Google actually does with headings, the multiple-H1 question answered from the record rather than folklore, the accessibility rules that carry legal weight, the mistakes that flatten real pages, and how heading structure now feeds AI search.
The short version
- Header tags create a page outline from H1, the main topic, down to H6, the finest detail.
- Google says any number of H1s ranks fine. One H1 is still the right call, for people.
- Headings are a light ranking signal and a heavy comprehension signal, for crawlers and readers alike.
- Keep levels in order. Skipped levels break navigation for screen reader users.
- Most content needs only H1, H2, and H3. Reaching for H4 usually means the page needs restructuring.
What the six levels are and what each one does
The markup is as simple as HTML gets, and the jobs divide cleanly:
<h1>What is on-page SEO</h1>
<h2>The elements that matter most</h2>
<h3>Title tags</h3>
<h3>Headings</h3>
<h2>How to audit your pages</h2>| Level | Job | How often it appears in practice |
|---|---|---|
| H1 | Names the page's one topic | Once per page, by convention |
| H2 | Marks the major sections | Several per page; secondary keywords sit here naturally |
| H3 | Breaks an H2 section into subtopics | Common in long-form content |
| H4 | Detail under an H3 | Mostly dense technical material |
| H5, H6 | Fine-grained documentation structure | Rare outside reference docs |
If you reach for H4 and beyond regularly, the honest fix is usually restructuring the content or splitting the page, not going deeper into the outline.
What search engines actually do with headings
Google has said directly that headings help it understand what each part of a page is about, and heading text carries more interpretive weight than the body text under it. That makes headings a comprehension signal more than a ranking lever: they do not lift a page the way links or content quality do, but they shape what queries each section can answer. Two concrete payoffs follow. Heading text phrased as a question is a common source for featured snippets, and clean section boundaries help Google surface a specific passage of a long page for a specific query. Headings are also what Google reaches for when it rewrites a weak title, which means your H1 is an understudy for your title tag whether you planned it or not.
The newest reader is AI retrieval. Systems that feed AI answers split pages into chunks before storing them, and heading boundaries are the natural seams. A section whose heading states plainly what the section answers is a section an AI engine can retrieve and cite on its own. Vague headings like "More info" produce chunks that answer nothing in particular, for machines and people alike.
The multiple H1 question, answered from the record
Ask this in a room of SEOs and you will hear one H1 stated as law. The record is looser. John Mueller of Google said you can use H1 tags as often as you want, with no upper or lower bound, and that a site will rank fine with five H1s or none. HTML5 permits multiple H1s, and Google has repeated for years that heading count is not a ranking factor. So the SEO case for exactly one H1 is thin. The case that remains is about people, and it is a good one: one H1 keeps the page's main topic unambiguous for screen reader users who jump straight to it, and it mirrors how a document actually has one title. Part of the confusion traces to the HTML5 outline algorithm, a spec idea where nested sections would let multiple H1s form a hierarchy. No browser or screen reader ever implemented it, and it has been removed from the spec, so the heading number is the only structure assistive tech sees. The full history is in our post on whether multiple H1 tags are ok. The short answer: you will not be penalized, and you should still use one.
Screen readers treat your headings as a table of contents
Screen reader users move through pages by jumping heading to heading, and WebAIM's screen reader surveys have found heading navigation the most common way users find information on a page, ahead of reading straight through. That turns your heading outline into the primary interface for part of your audience. A skipped level, H1 straight to H3, sounds like a missing section. A decorative H3 chosen for its font size announces a section that does not exist. A page of bolded paragraphs with no headings at all offers nothing to navigate by. None of these are cosmetic problems, and under the accessibility rules now carrying real deadlines for public-sector sites, heading structure is part of what conformance reviews check. The same discipline that helps a crawler parse your page is a legal and human requirement first.
A drafting habit that fixes most heading problems before they exist: write the headings first, as full natural sentences, and check that they retell the page's argument on their own. We draft every page on this site that way. If the heading list does not make sense as a summary, the structure is wrong, and no amount of body copy will rescue it. Readers who scan, screen reader users who navigate, and retrieval systems that chunk are all reading that same skeleton.
The rules that keep an outline working
- One H1 per page, carrying the primary topic, agreeing with the title without duplicating it word for word.
- Keep the hierarchy in order. H2 follows H1, H3 follows H2. Never pick a level for its size.
- Make each heading a promise about its section, then keep the promise in the text below it.
- Keep headings unique on the page. Two sections with the same heading help no one decide where to look.
- Write for the scan. A reader should reconstruct the page's argument from the headings alone.
- Style with CSS, not with levels. The level states what the text is; the stylesheet states how it looks.
The mistakes audits keep finding
- The logo is the H1 on every page. A common theme-template default, and it makes every page claim the brand name as its topic. Move the H1 to the page's real headline.
- Levels chosen for font size. Outlines like H1 to H4 to H2 come from picking whatever looked right in the editor. Fix the levels, then restyle.
- Label headings. "Overview," "More info," and "Details" describe nothing. Say what the section actually covers.
- The same keyword stuffed into every heading. It scans terribly and reads as spam to every audience the headings serve.
- Hidden headings for SEO. Headings positioned off-screen for crawlers still get read aloud to screen reader users, as noise. Structure you would hide is structure you should delete.
- A clean three-level outline whose headings retell the argument, with the keyword appearing where it genuinely describes a section.
Finding the problems takes minutes. A crawler like Screaming Frog lists every page's H1s in one report, and the free HeadingsMap browser extension draws the outline of any page you are viewing. Browser dev tools work too: search the elements panel for h1 and count the results.
Where header tags fit in the bigger picture
Headings are one strand of on-page SEO, doing on the page what the title does in the results, and they share their double duty with alt text: both exist for accessibility first and earn their search value as a side effect of doing that job honestly. Structure the page for the person navigating it, and every machine reading it benefits on the same pass.
Questions people ask about header tags
What are header tags in SEO?
Header tags are the HTML elements H1 through H6 that define the heading structure of a page. They tell search engines how the content is organized and what each section covers, and screen readers use them as a navigation map. They are structure, not styling.
How many H1 tags should a page have?
One, as a working convention. Google says rankings survive any number, including zero, and HTML allows as many as you like. The reasons to keep a single H1 are people: it keeps the main topic unambiguous for screen reader users and for anyone skimming, and it gives Google one clear candidate when it rewrites your title.
Do header tags affect rankings?
Lightly and indirectly. Google has said headings help it understand what each section of a page covers, and heading text is a common source for featured snippets, but headings are not a heavy ranking lever. The structure they give readers and assistive tech is the larger share of their value.
Should headings contain keywords?
When it reads naturally, yes. A heading should first describe its section accurately; if the target keyword belongs in that description, include it. Forcing exact-match keywords into every heading reads as spam, scans badly, and has not worked in years.
Can I use header tags just to make text bigger?
No. Headings are structure, not styling. If you want large text without a new section, use CSS. Choosing an H3 for its font size breaks the outline that screen readers and crawlers depend on, and it is one of the most common problems accessibility audits find.
Does skipping a heading level hurt SEO?
Google has said it copes with imperfect heading order, so the direct SEO cost is small. The real cost is accessibility: screen reader users navigating by heading level hit a gap where a section seems to be missing. Keep the order clean because people use it, and the crawler benefit comes along free.
