Skip to content

Social

Link previews for PDF and file links: why they are blank and the fix

A link straight to a PDF, DOCX or ZIP shares as a bare URL because files carry no Open Graph tags. Give each document a landing page with a cover image, and keep the file where it is.

getReport teamUpdated 25 Sept 20269 min read

You send the new price list to a client as a link to price-list-2026.pdf, and the chat shows a grey line with a file name. The same happens with a brochure on LinkedIn, a menu in WhatsApp or a whitepaper in Slack. Nothing is broken: a file simply has nothing for a preview to read. This guide explains why, what each platform does instead, and the one pattern that fixes it for every document: a small landing page per file, with the file itself left untouched.

Quick answer

  • Social platforms build previews from meta tags in an HTML page. A PDF, DOCX, ZIP or image file has no <head>, so there is nothing to read.
  • There is no way to attach Open Graph data to a file with an HTTP header. The fix has to be an HTML page.
  • Give each document a landing page: a title, one or two sentences, a cover image as og:image, and a download button. Share the page, not the file.
  • Make the cover from page 1 of the PDF at 1200×630 px; one command does it (below).
  • Set the PDF's own Title property, because Google shows PDFs in search and uses it.
  • Test the landing page with the link preview checker before you share it.

When someone pastes a link, the platform's crawler fetches it and looks for Open Graph tags (og:title, og:description, og:image) or, failing that, the page's <title> and meta description. Those live in the HTML <head>. A PDF is a binary document with its own internal metadata that social crawlers do not read. So the platform falls back to whatever it can show without tags:

  • Most platforms (Facebook, LinkedIn, X, WhatsApp) show the bare URL or the domain and file name, with no image.
  • Some apps show a file icon or a small thumbnail of the first page. iMessage can, depending on the file and the device. Slack renders PDFs that are uploaded into a conversation, which is a different feature from unfurling a link.
  • Document services (Google Drive, Dropbox, Notion) never share the file itself: their share links open an HTML viewer page with the service's own tags, so you get the file name as the title and the service's thumbnail or logo as the image.

A bare file link is also a worse experience after the tap. On a phone the PDF opens in a viewer or downloads silently, the visitor has no way to reach the rest of your site, and your analytics sees nothing unless the server logs are read.

Headers do not help. Content-Disposition decides whether the browser shows the file (inline) or saves it (attachment); it changes what happens after the tap, not the preview. A Link: header can carry a canonical for search engines, but social crawlers do not read Open Graph data from headers, and there is no header version of og:image.

How getReport checks it

The link preview checker fetches the address you give it and reads the Open Graph and Twitter card tags from the response. Run it on a PDF and you see what a social platform sees: the file has no tags to read, so the title, description, image and card findings all come back missing. Run it on the landing page instead and each finding should pass.

Social panel for an address without Open Graph tags: warnings for the missing Open Graph image, title and description, the Twitter card type, the X image and og:type, and a grey info line for the Facebook App ID
An address with no tags gets the same six warnings, whether it is a bare PDF or an HTML page nobody filled in.

The three findings that decide whether a shared document looks like something worth opening:

For the landing page's search side (a title, a meta description, a canonical, indexability) run the SEO audit on the same address.

Step by step

1. Make a landing page per document

One HTML page per important document: the annual report, the price list, the menu, each whitepaper. It needs:

  • A clear heading that says what the document is and how current it is ("Price list 2026, valid from 1 March").
  • Two or three sentences on what is inside and who it is for.
  • The cover image, shown on the page and used as og:image.
  • A download button with the file type and size ("Download PDF, 2.4 MB"), which also meets the reader's expectation on a phone.
  • Optionally the key content as HTML (the prices, the table of contents, the summary), which helps search and accessibility more than the PDF can.

The file stays at its current address, so existing links keep working. The head of the landing page:

HTML
<head>
  <title>Price list 2026 – Example Joinery</title>
  <meta name="description" content="All kitchen and wardrobe prices for 2026, valid from 1 March, with delivery and fitting costs.">
  <link rel="canonical" href="https://example.com/price-list/">
  <meta property="og:type" content="website">
  <meta property="og:url" content="https://example.com/price-list/">
  <meta property="og:title" content="Price list 2026">
  <meta property="og:description" content="Kitchen and wardrobe prices for 2026, including delivery and fitting.">
  <meta property="og:image" content="https://example.com/share/price-list-2026.jpg">
  <meta property="og:image:width" content="1200">
  <meta property="og:image:height" content="630">
  <meta name="twitter:card" content="summary_large_image">
</head>

And in the body, a plain link to the file:

HTML
<a href="/files/price-list-2026.pdf" download>Download the price list (PDF, 2.4 MB)</a>

2. Make the cover image from page 1

A 1200×630 px image is the shape Facebook, LinkedIn and X show largest. A portrait A4 page does not fit that shape, so either crop the top of it or place the whole page on a background. With Poppler's pdftoppm and ImageMagick 7 (convert instead of magick on ImageMagick 6):

Shell
# Render page 1 at 150 dpi to cover.jpg
pdftoppm -jpeg -r 150 -f 1 -l 1 -singlefile price-list-2026.pdf cover

# Option A: the whole page, centred on a light background
magick cover.jpg -resize x560 -background '#f4f1ea' -gravity center -extent 1200x630 -quality 85 price-list-2026.jpg

# Option B: the top of the page, where the title usually is
magick cover.jpg -resize 1200x630^ -gravity north -extent 1200x630 -quality 85 price-list-2026.jpg

Option A reads as "this is a document" at a glance, which sets the right expectation; option B shows more detail but can cut the title in half on some layouts, so check it. Keep important text away from the edges, because platforms crop differently: Open Graph images: size and safe area has the margins. Export as JPEG, not WebP, so every platform can show it. For dozens of documents, run the same commands in your build, or follow generating OG images automatically.

3. Give the PDF itself a real title

Google indexes PDFs and shows them in results, and the title it shows often comes from the document's Title property or its first heading, not the file name. Set it before exporting (in Word: File → Info → Properties → Title) or afterwards:

Shell
exiftool -Title="Price list 2026 – Example Joinery" price-list-2026.pdf

Good file names help too: price-list-2026.pdf, not Final_v3_NEW(2).pdf. They show in chats, in downloads folders and in the address bar.

4. Decide what search should show

You now have two addresses for the same subject. Two reasonable setups:

  • Both indexed. The landing page ranks for the topic, the PDF for people searching inside documents. The landing page's canonical points to itself. This is the default and is fine for most sites.
  • Only the landing page. Send X-Robots-Tag: noindex on the PDF so it drops out of search, and make sure the landing page is indexable and canonical to itself. Google documents the header for non-HTML files on its robots meta tag page.

On nginx, in the server block:

nginx
location ~* \.pdf$ {
    add_header X-Robots-Tag "noindex" always;
}

On Apache, in .htaccess at the site root (with mod_headers):

Apache
<FilesMatch "\.pdf$">
    Header set X-Robots-Tag "noindex"
</FilesMatch>

5. Share the page, everywhere

Update the links in emails, ads, QR codes and social posts to the landing page. Keep the direct file link for places where people expect a download (an email to a client who asked for the file, a "Downloads" list).

Platform notes

WordPress

WordPress creates an attachment page for every uploaded file, and for years those were the accidental landing pages of PDFs: a thin page with the file name as the title. Since WordPress 6.4 new sites have attachment pages turned off; older sites often still have them, and SEO plugins can redirect them (Yoast: Settings → Advanced → Media pages; Rank Math: General Settings → Links → Redirect Attachments). WordPress attachment pages explains the choice. Either way, an attachment page is not a landing page: create a normal page with the cover, the summary and the download button, and link to the file from there. The file stays in the media library at its current address.

Shopify

Files uploaded under Content → Files are served from Shopify's CDN with no page around them. Create a page (Online Store → Pages) per document with the summary and a link to the file, and set the page's image through the theme or the SEO fields so it carries its own share image.

Static sites / custom

Generate the landing page and the cover image in the same build step from a small data file (title, summary, file path), so adding a document is one entry. Set long cache headers on the files and give each new version a new file name.

Verify

  • The link preview checker on the landing page shows the Open Graph title, description and image as passed, with the image at 1200×630.
  • The same checker on the PDF address still shows every tag missing. That is expected; the file is not what you share.
  • The Facebook Sharing Debugger and LinkedIn's Post Inspector show the cover image and title for the landing page.
  • curl -sI https://example.com/files/price-list-2026.pdf | grep -i x-robots-tag prints noindex if you chose the "only the landing page" setup, and nothing otherwise.

Common mistakes

  • Sharing the file URL in ads and newsletters. Symptom: the ad or post has no image and the click lands in a PDF viewer with no way back to the site. Share the landing page.
  • A WebP cover image. Symptom: the card shows no picture in some apps. Use JPEG or PNG for og:image.
  • Noindexing the PDF and leaving the landing page unfindable. Symptom: the document disappears from search entirely. The landing page needs to be indexable, canonical to itself and linked from the site.
  • A new file name for every small revision without a redirect. Symptom: old links in emails and on other sites return 404. Keep a stable address for "the current price list", or redirect old versions to the landing page.
  • Using the attachment page as the landing page. Symptom: the shared card shows the file name and a generic image. Build a real page with a written title, a summary and a cover.

Related guides

Check your site before and after Check