We know that the overuse of
<h1> can lead to an over-optimization penalty, but now, in HTML5 it's allowed to use multiple
<h1> in the same page (inside nested sections).
What will Google do with pages containing
<h1> titles only?
According to Matt Cutts, Google doesn't penalize sites just for having multiple
<h1> tags. It's possible that their indexers may be programmed to detect egregious overuse of
<h1> — like, say, having all your text inside
<h1> tags — and to penalize such pages, but I've seen no direct confirmation of that. It is very likely that their human staff, if alerted to such a page, will see it as abuse and act accordingly.
I don't work at Google and I've no idea how their algorithms actually work (beyond what can be inferred from observations and posts by folks who do know), but if I were to write their heading handling code, I'd allocate a certain amount of "heading weight" per page and divide that weight proportionately across all the headings on the page. That way, having two
<h1> tags would have exactly the same effect as having a single long one, and using only, say,
<h3> tags on a page would have exactly the same effect as using only
<h1>. It's such an obvious solution that I'd be rather surprised if Google didn't do something more or less like that.
It's generally accepted best practice to have only a single
<h1> on a page, or only 1
<h1> in a section for HTML5. Google will not penalise you for meeting the HTML standards, nor will it penalize you for having multiple
<h1>'s - it may penalise you for having ONLY
The important point is that it's about the requirements of the content and the requirements of the users - be they sighted or not, that should carry the day.
Here's Google's Johnathan Mueller confirming that Google can index HTML 5, and that it doesn't do anything special for it, i.e. reward you for using HTML 5.
Our crawling and indexing systems currently don't do anything special for HTML5, so there is no "bonus" for using HTML5 constructs, but similarly also generally no downside. That said, most of the pages with HTML5 markup that I've seen tend to be very clean - with little "cruft" and unnecessary elements. While having clean markup isn't something that our algorithms explicitly look for, we occasionally see pages that are almost unparseable. Also, clean markup makes it much easier to maintain the website, easier to add new elements like microformats, and frequently makes the pages more portable across a variety of browsers and devices. So with that in mind, there's nothing holding you back (from a search point of view) from creating an awesome site in HTML5 :).
Everyone always points to the Matt Cutts post from 2009. Here is a video from 2011 during a live web cam chat where Matt Cutts specifically states his preference for a single
<h1> tag per page.
It's always been allowed to use multiple
<h1> in a document, and was never arbitrarily penalized. Overdoing it is what could make itself a problem.
As far as Google, you're just going to have to trust that the engineers are doing their jobs and if they're indexing HTML5 properly, Googlebot knows that nested
<h1> are effectively pushed "down" a level. If you don't trust that, then don't do it and just manually make them
<h2> until you get a clear, direct answer on it. (The link to Johnathan Mueller's post in @toomanyairmiles' response is interesting, but not complete enough for me.) I haven't seen this happen yet.
I assume this question is based on the logic that an
<h1> tag represents the one main title of the page. While that may have been true in previous days, HTML5 has changed that type of logic. In HTML5, the
<h1> tag is the main title of a section. This could be a full page, a subpage, a module, etc. A good example of why this difference matters is when building a modular system of different sections dedicated to different activities.
Imagine a system with multiple "pages" lined up next to each other. Each page serves a different purpose: one for the weather, one for organizing tasks, one for display customers, etc. Each of these pages can be further drilled down into (hour-by-hour forecast, single task, single customer) and these "subpages" could be visually placed alongside the actual pages.
Before HTML5, you would need to logically determine which header tag was appropriate and, if you went more than 6 levels deep, you ran out of header tags. With HTML5, you can set them all as
<h1> without extra logic or a nesting limitation.