What heading tags actually are
Strip away the SEO folklore and an H1 to H6 element is one thing: a declaration of document structure. It tells a browser, a crawler and an assistive technology that a block of text is a section title, and where it sits in the hierarchy. The HTML spec gives you six levels, H1 for the top, H6 for the deepest nesting. The visual size a browser applies is a default stylesheet convention, not the point. You can restyle an H3 to render enormous with CSS and it stays an H3 in the document outline.
The distinction that trips people up: a heading tag is semantic, a styled <div> with a large font is not. A crawler reading your HTML builds an outline from the heading elements in source order, and that outline is what gets used downstream for passage extraction, snippet generation and screen-reader navigation. If your visible « title » is a <span> carrying inline styles, it does not exist in that outline, no matter how big it looks to a human.
For a fast visual primer on the markup itself, this walkthrough covers the syntax cleanly:
One more piece of vocabulary worth nailing down, because competitors blur it: the heading elements are not the <header> landmark and not the <title> tag. The <header> is a structural region (a banner, a card top), it can contain a heading but is not one. The <title> lives in the <head> and feeds the browser tab and the SERP, never the visible page body. Conflating these three is the root of half the heading audits we read.
How headings work in 2026
Here is the senior stance: headings are a weak direct ranking signal and have been for years. Google's own Search Central documentation frames headings as a way to help the engine understand the structure of a page, not as a scoring input you optimise to climb. John Mueller has said on multiple occasions that having more than one H1 on a page causes no problem for Google. If your audit tool still flags « multiple H1 » as an error in red, recalibrate the tool, not the page.
What changed the calculus is how search and answer engines now consume content. Passage indexing, rolled out by Google in 2020, lets the engine rank a specific section of a long page for a query even when the whole page is not focused on it. Featured snippets and the AI overviews that increasingly sit above the ten blue links are assembled from discrete, well-bounded chunks of text. A clean H2 followed by a tight two-sentence answer is exactly the shape those systems prefer. That is the mechanism that makes headings matter in 2026: not weight in the ranking formula, but parseability for extraction.
A historical footnote that still confuses people: the old HTML5 document outline algorithm, which promised that each <section> would reset heading levels, was never implemented by browsers or assistive tech and was removed from the spec. So you cannot lean on sectioning to manage hierarchy. The flat H1 to H6 order in your source is the outline. Treat it as a strict, sequential structure, not a nested one that tools will infer for you.
Headings in an SEO and netlinking operation
On the on-page side, headings do three operational jobs. They map the topic so the engine can match sections to queries, they create natural targets for snippet and passage capture, and they give you in-body anchor points for internal linking. That last one is underused: a well-named H2 is the obvious destination for a contextual link pointing from a related article on the same site, and it doubles as a fragment URL you can deep-link to.
This walkthrough from a practitioner who has optimised heading structures across a large page set lines up with what we see at scale:
The H1 versus title tag question deserves a direct answer because the SERP and the page are different surfaces. The element that shows up as the clickable headline in search results is the title tag, optimised for click-through and length limits. The H1 is what the visitor reads at the top of the page once they arrive. They can carry different wording: a punchy, query-matched title for the SERP, a more natural, fuller H1 for the reader. Forcing them identical is a self-imposed constraint, not a best practice.
On the netlinking side, this is where structure stops being theory. When we place a sponsored article, the heading skeleton is decided before a sentence is written, because it dictates where the contextual link sits, which section can win a snippet, and how the host page reads to an editor. The way we structure the articles we publish in-house across our owned media always starts from an H2 outline that earns the link its surrounding context. The same discipline applies when you calibrate a campaign across several publishers over the long run: consistent, topical heading structures on the destination pages are what let the linked passage rank, not just the link itself. And as answer engines pull more from structured chunks, that same outline discipline is what gets a brand surfaced inside AI-generated answers.
Accessibility, the part most SEOs skip
Accessibility is not a compliance box, it is the same outline doing double duty. Screen-reader users navigate a page by jumping from heading to heading, exactly the way a crawler walks the outline. A blind visitor pressing the « next heading » shortcut expects a logical, sequential structure: H1, then H2s for major sections, then H3s nested under them. Skip from an H2 straight to an H4 and you have created a gap that reads as a broken or missing section to that user.
This short explainer ties the semantics and the accessibility side together well:
The governing references are concrete: WCAG 2.1 success criterion 1.3.1 (Info and Relationships) requires that structure conveyed visually, like a section heading, be available programmatically, which is precisely what a real heading tag does and a styled div does not. Criterion 2.4.6 (Headings and Labels) asks that headings describe their topic. The practical rule that satisfies both, and that happens to please crawlers, is simple: use real heading elements, name them descriptively, and never skip a level. The SEO win and the accessibility win are the same action, which is the rare case where you do not have to choose.
What we see go wrong in audits
The recurring failure, from what we see in audits, is using headings as a styling shortcut. A designer wants a bold callout, reaches for an H3 because it renders bold by default, and now the outline has a phantom subsection that means nothing. The reverse is just as common: a genuine section title built as a styled paragraph, invisible to the outline. Both break the machine-readable structure for no reader benefit.
Keyword stuffing in headings is the second one, and it is a legacy habit that should be dead. Cramming the target term into every H2 reads as spam to an editor and adds nothing for the engine, which long ago stopped rewarding exact-match density. Write the heading for the human scanning the page, the relevance follows from the content, not from repetition. Empty headings, headings that wrap an image with no text, and duplicated identical H2s across a page round out the list, each one a small dent in both the outline and the accessibility tree.
The last pattern worth flagging is over-nesting. H4, H5 and H6 exist, but a page that reaches H5 usually has a depth problem: the content should be split into separate pages or the hierarchy flattened. Deep nesting is rarely a structural necessity, it is more often a sign that one page is trying to be three. Keep the outline shallow and legible, and both the crawler and the reader will thank you.