How to Fix Key Content Only in PDFs: Publish HTML Versions

This check flags sites where a noticeable share of crawled URLs are PDF files, which means important content such as specs, pricing, guides or policies is only available as a document download. By the end of this guide you will have published proper HTML pages for your important PDFs, pointed links and sitemaps at them, and told search engines which version to index.

By Updated ⏱ 1-3 hours for 5-10 PDFs🎯 Intermediate● Low impactGEO / AI
Illustration: Key content only in PDFs fix guide
GEO / AI · FIX GUIDE Key content only in PDFs IMPACTEFFORT TIME1-3 hours for 5-10 PDFs LEVELIntermediate

Why it matters

Google can index PDFs, but they rank poorly for rich results, carry no structured data, have weak titles, and send users into a file with no navigation back to your site. AI crawlers and answer engines such as ChatGPT search, Perplexity and Google AI Overviews extract text from HTML far more reliably than from PDFs, where multi-column layouts, tables and scanned pages often turn into garbled or missing text. PDFs are also hard to use on phones and often fail accessibility requirements.

An HTML version makes the same content quotable, linkable, measurable in analytics and readable by screen readers.

✕Beforehttps://example.com/files/product-specs-2026.pdfis the only place the product✓Afterhttps://example.com/products/specs isa full HTML page with headings andtables, links to the PDF as an

What you need

A list of PDFs on your site, from a crawler or Google Search Console
Access to your CMS to create pages
Access to server config, CDN rules or an SEO plugin to set HTTP headers on PDFs
The original source files (Word, Google Docs, InDesign) if available

Step-by-step fix

1

Inventory every PDF on the site

Find all PDFs that crawlers can reach. Use several sources, because PDFs are often linked from old pages or only from the sitemap:

  • Run a crawler (Screaming Frog: filter Internal > PDF) on your domain
  • In Google Search Console open Indexing > Pages, click View data about indexed pages and filter URLs containing .pdf
  • Search Google for site:example.com filetype:pdf
  • List files on the server if you have shell access
text
find /var/www/example.com -iname '*.pdf' -printf '%s\t%p\n' | sort -nr > pdf-inventory.tsv
Pro tip: Add columns for organic clicks (Search Console > Performance, filter page contains .pdf) and inbound links so you can prioritise in the next step.
2

Decide which PDFs need an HTML version

Not every PDF needs converting. Prioritise documents that answer questions people search for or that AI assistants would cite: product specs, pricing, service descriptions, how-to guides, FAQs, research reports, policies and terms. Low-priority files such as printable forms, invoices, archived press kits or brochures that duplicate existing pages can stay as PDFs.

Mark each row in your inventory as Convert, Keep as download only or Remove.

Pro tip: If an existing HTML page already covers 80% of a PDF, merge the missing content into that page instead of creating a new one.
3

Extract the content into clean HTML

Start from the original source file whenever possible, because text extracted from a PDF loses headings and breaks tables. From a Word file, pandoc produces clean semantic HTML. If only the PDF exists, use poppler's pdftotext to get the raw text and then rebuild the structure by hand.

Scanned PDFs need OCR first (for example ocrmypdf).

text
# From the original Word document
pandoc product-specs-2026.docx -t html5 -o product-specs.html

# If only the PDF exists
pdftotext -layout product-specs-2026.pdf product-specs.txt

# Scanned PDF: add a text layer first
ocrmypdf product-specs-2026.pdf product-specs-ocr.pdf
Pro tip: Paste the output into your CMS's code or HTML view rather than the visual editor to avoid inline styles and empty spans.
4

Build a proper HTML page for each document

Create a normal page in your CMS with a descriptive URL, a unique <title> and meta description, one <h1>, <h2>/<h3> sections that mirror the document, real <table> elements for data, and alt text on images. Add structured data where it fits (Article, Product, FAQPage). Keep the PDF available as a clearly labelled download link for people who want to print it.

html
<main>
  <h1>Example Co X200 Technical Specifications</h1>
  <p>Last updated: <time datetime="2026-01-15">15 January 2026</time></p>
  <h2>Dimensions and weight</h2>
  <table>
    <thead><tr><th>Property</th><th>Value</th></tr></thead>
    <tbody>
      <tr><td>Width</td><td>420 mm</td></tr>
      <tr><td>Weight</td><td>3.2 kg</td></tr>
    </tbody>
  </table>
  <p><a href="/files/product-specs-2026.pdf" download>Download the printable PDF (240 KB)</a></p>
</main>
Pro tip: Put the key facts (numbers, dates, prices) in the first paragraph or a summary table. AI answer engines quote short, self-contained statements most often.
5

Point the PDF to its HTML version

So search engines index the HTML page instead of the file, send an HTTP header on the PDF. Use Link rel="canonical" when the PDF has an HTML equivalent, or X-Robots-Tag: noindex for PDFs you want kept out of search entirely. Use one or the other per file, not both.

nginx
# Nginx: canonical for a specific PDF
location = /files/product-specs-2026.pdf {
    add_header Link '<https://example.com/products/specs>; rel="canonical"' always;
}

# Nginx: noindex for download-only PDFs in a folder
location ^~ /files/forms/ {
    add_header X-Robots-Tag "noindex" always;
}

# Apache (.htaccess) equivalents
<Files "product-specs-2026.pdf">
    Header set Link "<https://example.com/products/specs>; rel=\"canonical\""
</Files>
<FilesMatch "^(order-form|returns-form)\.pdf$">
    Header set X-Robots-Tag "noindex"
</FilesMatch>
Pro tip: Do not block PDFs in robots.txt. If Google cannot fetch the file, it cannot see the canonical or noindex header.
6

Update internal links and the sitemap

Change navigation, footer, product and blog links so they point to the new HTML pages, not straight to the PDFs. Remove converted and noindexed PDFs from your XML sitemap and add the new HTML URLs. For PDFs marked Remove, return a 301 redirect to the closest HTML page so existing backlinks are not lost.

nginx
# Nginx: retire an obsolete PDF
location = /files/brochure-2022.pdf {
    return 301 https://example.com/products/;
}
Pro tip: Search your database or theme files for ".pdf" to catch hard-coded links the crawler might miss.
7

Check the result

Confirm the headers and links are live, then re-run the audit. The share of crawled URLs that are PDFs should drop, and your key content should now appear on HTML pages. In Google Search Console, request indexing for the new HTML pages with URL Inspection > Request indexing.

text
curl -sI https://example.com/files/product-specs-2026.pdf | grep -iE 'link|x-robots-tag|content-type'
Pro tip: Expect Google to take a few weeks to swap the PDF for the HTML page in results. Track both URLs in Search Console > Performance.

Platform-specific notes

WordPress

PDFs uploaded through Media live under /wp-content/uploads/ and are served directly by Apache or Nginx, so add the headers in .htaccess or your server config. Make sure attachment pages are disabled (default since WordPress 6.4 for new sites; in Yoast use Settings > Advanced > Media pages). Use Pages > Add New for the HTML versions and remove PDFs from the sitemap in your SEO plugin's settings.

Shopify

Files from Content > Files are served from cdn.shopify.com, where you cannot set HTTP headers. Create HTML versions under Online Store > Pages (or as metafield content on product pages), link to them instead of the PDF, and keep the PDF only as a secondary download link.

Webflow

Uploaded PDFs are hosted on Webflow's asset CDN and cannot receive custom headers. Build the HTML version as a static page or a CMS Collection item, update links, and rely on internal linking and the sitemap to signal the preferred URL.

Cloudflare

If your origin cannot set headers, go to Rules > Transform Rules > Response Header Transform Rules, create a rule with the expression ends_with(http.request.uri.path, ".pdf") (or a specific path), and set X-Robots-Tag or Link. This works in front of any host, including platforms that lock server config.

Next.js

Files in /public are served as static assets. Add headers in next.config.js with async headers() returning e.g. { source: '/files/product-specs-2026.pdf', headers: [{ key: 'Link', value: '<https://example.com/products/specs>; rel="canonical"' }] }. On Vercel you can also use the headers array in vercel.json.

Check that it worked

✅ How to verify

  • Run curl -I on each PDF and confirm the Link rel="canonical" or X-Robots-Tag header is present and the status is 200 (or 301 for retired files).
  • Use Search Console URL Inspection on the new HTML page and confirm it is crawlable, indexable and shows the rendered content; inspect the PDF and check the Google-selected canonical once it has been recrawled.
  • Crawl the site again and confirm navigation and sitemap links point to HTML pages, not PDFs.
  • Re-run the audit and confirm the proportion of PDF URLs among crawled pages has fallen below the warning threshold.

⚠️ Common mistakes

  • Blocking /files/ or *.pdf in robots.txt, which stops Google from seeing the canonical or noindex header so the PDF can stay indexed.
  • Pasting PDF text into one long paragraph with no headings or tables, which gives AI and search engines little more structure than the PDF had.
  • Creating the HTML page but leaving every internal link and the sitemap pointing at the PDF.
  • Letting the PDF and HTML versions drift apart after updates; edit the HTML first and regenerate the PDF from it.

Is this still an issue on your site? Run a free audit — ZORO checks this and 294 other SEO, AI-visibility, accessibility and security signals, with fixes for your platform.

Check my site free

FAQ

Does Google index PDF files?

Yes. Google indexes text-based PDFs and can rank them, but they lack structured data, site navigation and reliable titles, and often show poorly on mobile. An HTML version usually ranks and converts better.

Can ChatGPT and other AI search tools read PDFs on my site?

Some AI crawlers fetch PDFs, but many skip them or extract text poorly, especially with tables, columns or scanned pages. Clean HTML is the most reliable way to get your content quoted and cited.

Will having both an HTML page and a PDF cause duplicate content problems?

Not if you send a Link rel="canonical" header on the PDF pointing to the HTML page. Google then consolidates signals on the HTML URL. Duplicate content is not a penalty, but without a canonical Google may choose the PDF.

Should I delete my PDFs after creating HTML versions?

Usually not. Keep them as a printable download for users who need them, and set a canonical header. Only delete PDFs that are obsolete, and 301-redirect those URLs to the most relevant HTML page.