Headings are the outline a machine reads before it reads anything else. A crawler building an understanding of your page, a screen reader announcing sections to a blind user, and a language model deciding which passage answers a question all lean on the same hierarchy.

When that hierarchy is broken — two H1s competing, a jump from H2 straight to H4, a heading left empty by a theme — the outline stops making sense. The page still looks fine to a sighted visitor, which is exactly why these problems survive for years.

An H1 checker parses the actual markup and reconstructs the outline as a machine sees it. That view is frequently unrecognisable to the person who built the page.

The rules are simple and rarely followed completely:

• Exactly one H1, describing what the whole page is about.
• H2s for main sections, each a distinct part of the topic.
• H3s nested inside H2s, never floating on their own.
• No skipped levels. H2 to H4 is a break in the outline.
• No empty headings. Every heading tag contains text.
• Descriptive, not decorative. A heading should make sense read out of context.

Think of it as a table of contents. If you extracted every heading and read them in order, would someone understand the shape of the page? If not, the structure needs work.

A page with no H1 gives search engines no primary statement of topic. They will infer one from the title tag and content, but inference is weaker than declaration. Pages built with page builders and drag-and-drop sections are the usual offenders — the visual title is styled text, not a heading tag.

HTML5 permits multiple H1s within sectioning elements, and browsers handle it without complaint. For SEO the practical guidance has not changed: use one. Two H1s make two competing claims about what the page is about, and the second is usually a leftover from a template rather than a deliberate choice.

Going from H2 to H4 creates an orphan. Assistive technology announces a structure that does not exist, and parsers lose the parent-child relationship between sections. This almost always happens for visual reasons — someone wanted smaller text — which is precisely the wrong reason to change a heading level.

Heading tags containing nothing at all, usually artefacts of a theme or a deleted section. They contribute nothing and clutter the outline. They are invisible on the page and obvious in an audit.

A heading running past seventy characters stops functioning as a signpost. It also makes a poor extraction candidate — models prefer headings short enough to quote as a label.

This is where heading structure has quietly gained importance. Answer engines segment content before they extract from it, and headings are the primary segmentation signal available.

A page organised as a sequence of question-shaped H2s, each followed immediately by a two-to-three-sentence direct answer, is close to the ideal shape for citation. The model can identify the question, lift the answer, and attribute it — all without ambiguity about where the answer begins and ends.

Compare that to a page where the same information exists but is buried in long paragraphs under vague headings like "Our approach" or "More information". The content may be better. It will be cited less.

A practical exercise: take your three most important H2s and rewrite each as the question a customer would actually type or ask aloud. Then make sure the first paragraph beneath each one answers that question completely, before any context or qualification.

This single change tends to move more than any other on-page adjustment, and it costs nothing but editing time.

The most common objection to fixing headings is that it will break the design. It will not, because heading level and visual appearance are independent.

If your H2s look too large, change the tag to H2 and style it to render at whatever size you want. The semantic level tells machines about hierarchy; CSS tells browsers about appearance. Conflating the two is the root cause of most broken outlines.

For the logo-in-H1 problem, change the logo wrapper to a div or span and promote the real page title to H1. On Shopify this usually lives in the header snippet and takes one edit to fix sitewide.

Checking one page at a time works for spot fixes. For a full audit, crawl the site with a tool that exports heading data, then filter for pages with zero H1s, more than one H1, or missing H2s entirely.

Patterns matter more than individual pages. If every product page shares the same structural fault, that is one template fix rather than hundreds of manual edits. Sorting issues by template rather than by URL usually collapses a long list into a short one.

Screen reader users navigate by heading far more than by scrolling. Assistive technology exposes a list of headings and lets someone jump straight to the section they want, which makes the outline a primary navigation mechanism rather than a stylistic detail.

A page with no H1 offers no starting point. A page that skips from H2 to H4 announces a subsection of something that does not exist. An empty heading appears in the list as a blank entry with no indication of where it leads.

This matters beyond compliance. The same structure that makes a page navigable by keyboard makes it parseable by a crawler and extractable by a language model. Accessibility work and technical SEO work overlap here almost completely, which is unusual and worth exploiting when you are arguing for the time to do it.

The H1 and the title tag do related but separate jobs, and treating them as duplicates wastes an opportunity.

• The title tag is written for the search result. It competes for a click against nine other results and gets truncated around 600 pixels.
• The H1 is written for someone who has already clicked. It confirms they landed in the right place and sets up the page.

They should agree on topic without being identical. A title tag might read "Free H1 Checker: Audit Your Heading Structure" while the H1 simply reads "H1 Checker" — same subject, different jobs. Where they diverge in topic rather than phrasing, that mismatch is usually worth investigating: it often means the page is trying to do two things at once.