On-page SEO checklist: what to check on every page, and why Google says so

What is on-page SEO?

On-page SEO is the work you do on a page itself so search engines can read it correctly and searchers want to click it. It covers the HTML that search engines read (title, meta description, robots tags, canonical, structured data) and what people read (headings, text, images, links). Off-page signals such as links from other sites are a separate job; technical SEO such as crawl budget and site architecture overlaps with it.

This on-page SEO checklist is in the order that matters most: first whether the page can be indexed at all, then how it appears in results, then what is on it. Every item links to the Google Search Central page it comes from, so you can check the source rather than trust a rule of thumb. For the whole-site version of this work, see how to do an SEO audit.

1. Can the page be indexed?

Before anything else, confirm nothing on the page tells search engines to leave it out. Google's noindex documentation lists two ways a page opts out: a <meta name="robots" content="noindex"> tag, or an X-Robots-Tag: noindex HTTP header. The header is easy to miss because it never shows in the page source.

Two details catch people out. First, Google does not support noindex in robots.txt, and if robots.txt blocks the page, Googlebot never sees the noindex at all. Second, a removed noindex only takes effect after Google recrawls the page, which the same page says can take months for less important URLs. Use URL Inspection in Search Console to see the HTML Google actually received.

2. Is the title distinct and descriptive?

Every page needs its own <title> that says what the page is, in plain words. Google's title link guidance asks for descriptive, concise titles, warns against vague labels like “Home”, keyword stuffing and boilerplate repeated across pages, and suggests a short brand name at the start or end with a separator.

Google says there is no limit on how long a title can be, but the title link is cut to fit the device. It also builds title links from more than the <title>: the main visible heading, <h1>, og:title, prominent text, anchor text pointing to the page and WebSite structured data. When your title and your main heading say different things, expect Google to pick its own. Preview how yours will be cut with the free SERP snippet preview.

3. Does the meta description summarize the page?

Write a unique meta description that summarizes the page, because Google may use it as the snippet when it describes the page better than the text does. Google's snippet documentation says snippets come mainly from the page content and change with the query, so the same page can show different snippets for different searches.

• Make each one unique. If time is short, start with the home page and your most visited pages.
• Include specifics such as the price, author, date or other key facts, clearly labelled.
• Avoid keyword lists. Summarize what the page actually covers.
• Ignore length rules. Google says there is no limit on meta description length; the snippet is truncated to fit the device.

If you need to keep text out of snippets, Google supports nosnippet, max-snippet and the data-nosnippet attribute.

4. Do the headings describe the sections?

Use a clear main heading and subheadings that tell a reader what each section answers. Google's SEO Starter Guide says heading order does not matter to Google Search, and there is no ideal number of headings. Semantic order still helps screen reader users, so keep it tidy anyway.

The practical rule: one visible main heading that matches the title's topic, then subheadings phrased the way people ask. That helps readers scan, and it gives Google a consistent signal for the title link.

5. Is the content helpful to the person who searched?

The page should answer the search fully, with original information, from someone who knows the subject. Google's helpful content guidance gives self-assessment questions: does it add original information or analysis, is the title descriptive rather than exaggerated, would someone bookmark or recommend it, and would a reader leave satisfied?

• There is no preferred word count. Google says so directly. Write as much as the question needs.
• Show who wrote it and how. Bylines where readers expect them, and a note on how content was made where they would wonder, including substantial use of automation.
• Do not fake freshness. Changing dates without substantially changing the content is a warning sign in Google's list.
• Do not stuff keywords. The Starter Guide says keyword stuffing is against Google's spam policies.

6. Do images have alt text?

Use HTML <img> elements with short, specific alt text that says what the image shows in the context of the page. Google's image SEO best practices give alt="Dalmatian puppy playing fetch" as the model and warn that filling alt attributes with keywords is keyword stuffing.

Two more points from the same page: Google does not index CSS background images, so an image that matters belongs in an <img> tag; and descriptive file names such as black-kitten.jpg give Google light clues, while IMG00023.JPG gives none.

7. Is the page linked, and does it link out well?

Every page you care about needs at least one crawlable link from another page on your site, with anchor text that describes it. Google's link guidance says Google can generally crawl only <a> elements with an href; links that rely on onclick or a <span> may not be followed.

• Avoid “click here” and “read more”. Read the anchor text alone: if it makes no sense, make it specific.
• Do not stuff keywords into anchors or stack many links side by side.
• Link in context, from the paragraphs where a related page is genuinely useful.

To find pages that deserve more internal links, see internal link suggestions; to find links that point nowhere, use the broken link checker.

8. Is there valid structured data?

Add structured data, preferably JSON-LD, for the content types that have rich results, with every required property filled. Google's introduction to structured data recommends JSON-LD where your setup allows it and says markup makes a page eligible for enhanced results; it does not guarantee them. Validate with the Rich Results Test, then watch the rich result reports in Search Console.

Mark up only what is visible on the page. An FAQ, product or article schema that describes content readers cannot see is a mismatch, not an optimization.

9. Does the canonical point where you want?

Put one absolute rel="canonical" link in the head, pointing at the URL you want indexed, including on that page itself. Google's canonicalization guide calls the tag a strong signal, not a guarantee: redirects are stronger, sitemaps weaker. It also says not to send mixed signals (a sitemap URL that differs from the canonical), not to use robots.txt or noindex to choose canonicals, and that the link only counts in the <head>.

A canonical tag is optional. If you leave it out, Google picks a version itself. Two different canonical tags on one page are worse than none.

10. Is the page experience solid?

Serve the page over HTTPS, make it readable on a phone, avoid intrusive pop-ups and aim for good Core Web Vitals. Google's page experience guidance says there is no single page experience signal; its core ranking systems look at several. Core Web Vitals are used, but good scores do not guarantee top rankings, and Google still shows the most relevant page even when its experience is poor. Page experience matters most when many helpful pages compete.

For field data on LCP, INP and CLS and what “good” means for each, see Core Web Vitals thresholds.

How to run an on-page SEO audit of one page with MrKeyword

Paste the URL into the free on-page SEO checker; it fetches the page live, reads its HTML and lists every finding with the reason. No account is needed. It is the quickest way to run this on-page SEO checklist against a real page.

1. Enter the page address

   Any public URL, with or without https://. The checker follows redirects and reports on the final address.

2. Read the findings

   Each finding is a problem, a fix or a note, in plain words, such as “No meta description. Google will pick a passage from the page instead.”

3. Fix problems first

   Problems are things that stop a page working in search, such as a noindex. Then fixes, then notes if you have time.

The score starts at 100 and loses 12 points per problem, 6 per fix and 1 per note. It measures how much is left to tidy, not how well the page will rank. The word-count flags are a prompt to look, not a target: Google has no preferred word count, and a short page that fully answers a short question is fine.

To check every page at once, the free site audit crawls your site and runs its checks on each page; in the app, the site audit tracks them over time.

What MrKeyword does not do

• It does not run JavaScript. The checker reads the HTML the server returns. Content, links or tags added later by scripts are not seen; use URL Inspection in Search Console to see Google's rendered HTML.
• It does not judge content quality or intent. It counts words, headings and alt text; whether the page is helpful is a human call.
• It does not validate structured data against Google's required properties. It lists the types it found; use the Rich Results Test for eligibility.
• It does not measure Core Web Vitals in this check. It times the server response only; field data lives in the Core Web Vitals feature.
• Its score is not a ranking prediction. Links from other sites, relevance and competition are outside what one page fetch can see.

Questions and answers