Use the page paths in your Scan to find the affected templates. Fix the template when it causes the same issue on several pages, then check an example of each page type. Clear text helps tools interpret a page; it does not guarantee a citation or ranking.
Readable text
This compares readable text with the size of the initial HTML. A low ratio can mean wrappers and navigation crowd out useful content. It can also describe a legitimate short page, so inspect the page before adding words.
Keep the actual offer, identity and proof readable without scripts. Remove unnecessary markup where practical. Do not pad a page to satisfy a ratio. Read the raw response and confirm its main facts survive. Our writing for extraction guide covers text structure in more detail.
Page titles
The HTML title helps a reader or search engine identify the page. Missing titles, repeated titles and heavy boilerplate make that job harder.
Write a distinct title from each page’s subject. Put the useful distinction before a long brand suffix. Change the template if it repeats the same title everywhere. Inspect the rendered <title> on the flagged paths. Search engines may choose a different title link; Google’s title link guidance explains the signals they use.
Meta descriptions
A description offers a short preview of the page. Search engines can choose other text, but a specific description gives them a useful option. It should not promise something the page does not contain.
Write a separate summary for each flagged page. Remove repeated sales copy and name what the visitor will find. The Scan’s length band is an editorial check, not a fixed Google limit. Verify the tag in the initial HTML. See Google’s snippet guidance.
Heading structure
Headings provide a page outline. The Scan checks for one main H1 and missing intermediate levels. It reads static heading counts, not a full accessibility tree.
Give the page one descriptive main heading, then organize sections with H2 and subsections with H3. Use CSS for appearance instead of choosing a heading level for its font size. Read the outline on its own and check that it still explains the page. MDN’s heading reference covers the elements.
Distinct page titles
Several valid titles can still say almost the same thing. A city, service or product page needs a real distinction, not a different number attached to boilerplate.
Compare the flagged titles, main headings and descriptions beside the page content. Name the subject that genuinely differs. If two pages have no distinct purpose, consider combining them rather than inventing differences. Check the whole template family after editing it.
Image descriptions
Alternative text carries information for people who cannot see an image and for text-based tools. It should explain the image’s role on this page, not list search keywords.
Describe informative images briefly. Use alt="" for decoration. For a linked image, make its destination or action clear. Review charts and diagrams in the surrounding text as well. The Scan cannot judge whether a description is useful just because it exists. See the W3C images tutorial for examples.
Page language
lang identifies the page’s language. hreflang connects actual translated versions. Wrong labels make language selection and pronunciation less reliable for tools and assistive technology.
Set the HTML language to the language used on that page. Use valid language tags and absolute translation URLs. Include a self-link and return links between translations. Do not add languages you do not publish. Verify the linked pages as a set. Google’s localized version guidance includes examples.
Mobile viewport
A viewport setting tells a narrow browser how to size the page. Without width=device-width, the visitor may see a tiny desktop layout.
Use <meta name="viewport" content="width=device-width, initial-scale=1">. Then test the actual page at a narrow width. A correct tag does not fix fixed-width tables, images or navigation by itself. Check for horizontal scrolling and clipped content. MDN’s viewport reference explains the options.
Link previews
Open Graph fields supply link-preview titles, descriptions and images. They give sharing tools explicit metadata, but are not evidence that an AI engine will cite the page.
Add the missing fields in the affected template. Use an absolute public image URL and text that matches the page. Check that the image loads without signing in. Preview a real shared URL after clearing any platform preview cache. The Open Graph protocol lists the fields.
Browser zoom
Restrictions such as user-scalable=no can stop people enlarging text. A responsive page should still work when someone zooms in.
Remove unnecessary zoom restrictions and low maximum-scale limits. Test enlarged text and browser zoom, especially forms and navigation. Fix clipping in the layout rather than preventing zoom. The viewport reference above explains the settings; this static check is not a substitute for testing the page.
Accessible controls and page regions
Unnamed controls and unclear main landmarks make navigation harder for people and software. This check reads a few static signals; it is not an accessibility certification.
Use native button text and form labels. Add one main content region with <main>, and avoid multiple competing main landmarks. Test keyboard navigation and accessible names in your browser’s accessibility tools. Follow MDN’s HTML accessibility guide, then test with people and assistive tools where possible.
Internal links
Links between sampled pages can point to errors, redirects or a URL that declares another page as canonical. Direct links to the right page help readers and crawlers follow your structure.
Update the source link to the working preferred destination. Keep a useful replacement when a page moves. Do not redirect every deleted page to the homepage. Recheck both source and destination. The Scan only knows targets in the pages we checked; it cannot establish site-wide orphan pages.
Link relationships
Link relationships describe the destination and browser behavior. An explicit opener on a new-tab link can grant the destination access to the original window. External links do not automatically need nofollow.
Remove unnecessary opener permissions. Keep sponsored or ugc where those relationships apply. Test legitimate new-tab workflows before changing them. MDN’s rel reference explains the tokens. The Scan reports static attributes, not a browser exploit test.
Similar page content
The Scan compares main text on sampled pages. High similarity is a reason to inspect them, not proof that a page should be deleted. Product variations and legal text can legitimately overlap.
Keep pages separate when they answer different needs. Improve their useful differences, or consolidate true duplicates with appropriate redirects and canonicals. Update internal links after a merge. Read both pages before deciding; do not add filler just to lower a similarity score.
