The Single H1 Rule and Why Search Engines Care

The rule says every page should have exactly one H1, and it should state what the page is about. Like most rules that get repeated for twenty years, it has accumulated a fog of half-remembered justifications, some of which are wrong. It’s worth separating what is actually true here, because the rule survives the corrections, and the reason it survives is more interesting than the rule.

Start with what’s false. HTML does not forbid multiple H1s. The specification allows as many as you like, and for a while the HTML5 spec even encouraged a pattern where each nested section got its own H1, on the theory that browsers would compute an outline from the nesting. Browsers never implemented that outline, and the advice was eventually walked back, but the tags themselves were always legal. Nor will Google penalize you for having two H1s. People at Google have said plainly that multiple H1s are fine, in the sense that no ranking demerit is applied. So the rule is not a law of HTML and not a law of Google. If you stop there, it looks like folklore.

But look at what an H1 actually does for a machine reading your page. A search engine’s core problem with any document is figuring out what it’s about, so it knows which queries the document might answer. It has several sources: the title tag, the URL, the body text, the links pointing in. Among these the H1 has a special property. It is the author’s own answer, placed in the document’s most prominent position, marked with the most prominent tag. The title tag is similar but lives in the browser chrome and is often mangled by CMS templates into “Page | Category | Brand | Tagline.” The H1 is what the author chose to put in large text at the top of the actual page. It is the closest thing to asking the author directly.

Now consider what happens to that signal when there are two H1s, or none. With none, the machine falls back to weaker evidence, and so does a human. My scanner, GazeSite, emits “No H1 heading found” constantly, and the pages it fires on almost always have the same shape: there is text that visually functions as the headline, but it is a div styled to look big, so nothing in the markup says it matters. The information exists on the screen and is absent from the structure. Machines read structure. So does every screen reader, which is why blind users, who navigate by jumping between headings, land on such a page and find it has no top.

With two H1s, something subtler happens. The signal doesn’t disappear; it splits. The page is now claiming to be about two things at once, with equal emphasis. Sometimes that’s an accurate claim, and then nothing bad follows, which is why Google doesn’t penalize it. But usually it’s not a claim at all. Usually it’s an accident: the logo is wrapped in an H1 by the template, and the article headline is another H1, so the page announces it is equally about “Acme Corporation” and “How to Repot a Fiddle-Leaf Fig.” No one decided that. It fell out of two developers not talking to each other.

And this is the real content of the single-H1 rule. It is not a requirement of the format or of the ranking algorithm. It is a discipline that forces a decision. A page can only be found for what it’s about, and “what is this page about” turns out to be a question many pages have never answered, because answering it means choosing. The product page that is about the product but also the brand story but also the seasonal promotion is about nothing in particular, and it will be found for nothing in particular. Requiring exactly one H1 is a mechanical trick for surfacing that unmade decision. If you can’t say which single line of text is the heading, the problem is not your markup. The problem is the page.

I find this pattern all over software: rules whose stated justification is wrong but whose effect is right. “One H1 per page because Google requires it” is false, and following it is still correct, because the rule launders a hard editorial question into an easy technical one. Developers who would wave off “make sure this page has a clear singular topic” as marketing fluff will happily enforce “exactly one H1” in a linter. The linter check and the editorial question are the same question wearing different clothes.

So when an audit flags your H1s, the mechanical fix takes a minute. Demote the logo to a span or a p, promote the real headline from a styled div to an H1, done. But the minute is worth spending with the real question in mind. Of everything on this page, which one sentence would I keep if I could keep only one? Put that in the H1. You are not doing it because the spec demands it. You are doing it because it is the one place where a machine, or a person in a hurry, or a blind user skipping from heading to heading, will look to learn what you think your page is for. It seems wise to have decided.