Foglift
All Tools
Free Tool
Open Graph Debugger & Preview
See exactly how your page looks when shared on Facebook, X/Twitter, LinkedIn, Discord, and Slack. Debug OG tags, find missing metadata, and fix social sharing issues.
How It Works
1
Enter Your URL
Paste any URL and we fetch the page server-side to extract all Open Graph, Twitter Card, and meta tags.
2
Preview Social Cards
See pixel-accurate previews of how your link appears on Facebook, X/Twitter, LinkedIn, Discord, and Slack.
3
Fix Issues
Get a prioritized list of missing or misconfigured tags with clear instructions on how to fix each one.
Why Open Graph Tags Matter for AI Search
Social Sharing Drives Traffic
Posts with proper OG images get 2-3x more engagement on social media. Without og:image, your links appear as plain text with no visual appeal, drastically reducing clicks and shares.
AI Models Parse OG Tags
AI search engines like ChatGPT, Perplexity, and Claude use OG metadata as structured signals when crawling your site. Your og:title and og:description help AI understand and summarize your page.
Brand Consistency Across Platforms
Each platform renders link previews differently. OG tags let you control your title, description, and image so your brand looks professional whether shared on LinkedIn, Discord, or Slack.
OG Tags Complement Schema Markup
While Schema.org JSON-LD helps search engines understand page structure, OG tags handle the visual presentation layer. Together, they create a complete picture of your content for both humans and AI.
Essential Open Graph Tags
| Tag | Required | Description |
|---|---|---|
| og:title | Required | The title displayed when shared. Should be concise (under 60 characters) and match or complement your page title. |
| og:description | Required | A brief summary of the page content (50-200 characters). This appears below the title in social previews. |
| og:image | Required | The image shown in the social preview. Recommended size: 1200x630px. Use JPEG or PNG format for best compatibility. |
| og:url | Required | The canonical URL for the content. Helps prevent duplicate shares and ensures engagement metrics are consolidated. |
| og:type | Optional | Content type (e.g., 'website', 'article', 'product'). Defaults to 'website' if not specified. |
| og:site_name | Optional | Your website or brand name. Displayed above the title on some platforms. |
| twitter:card | Optional | Card type for X/Twitter: 'summary', 'summary_large_image', 'app', or 'player'. Use 'summary_large_image' for maximum impact. |
| twitter:image | Optional | Image for Twitter cards. Falls back to og:image if not set. Recommended: 1200x675px for large image cards. |
How AI engines render Open Graph differently
ChatGPT, Claude, Perplexity, and Google AI Overviews each compose their citation cards from the same OG block, but they weight the fields differently and render them in different layouts. Five patterns below are where pages with a clean Facebook Debugger pass still lose AI citations.
Open Graph is a meta-tag protocol introduced by Facebook in 2010. It is the de facto standard for how a URL renders as a card when shared anywhere, including modern AI source panels. The fields are simple. The way each surface consumes them is not.
- •og:image renders inline; relative URLs break it. ChatGPT, Claude, and Perplexity fetch the og:image as-is. They do not resolve relative paths against the page URL the way a browser would. A relative og:image (/img/hero.png) renders as a broken thumbnail in the citation card. Use absolute URLs only (https://example.com/img/hero.png). A missing thumbnail visibly degrades the card, which lowers click-through by an estimated 30 to 50 percent in our internal tests.
- •Citation panels truncate the description. Perplexity source cards truncate around 160 characters. Claude citation panels around 200. ChatGPT browse-mode cards around 220. A 300-character og:description that reads fine in a Facebook share renders cut mid-sentence in an AI panel. Front-load the factual punchline before character 160 and you stay readable on every surface.
- •og:type tilts source-card confidence. og:type is technically optional, but AI engines treat its value as a hint for how to weight the rest of the block. og:type=article unlocks the article:author and article:published_time fields and lifts the citation from domain-level (the site says X) to source-level (the article wrote X). Source-level attribution is cited more often.
- •article:author is a person, not a brand. ChatGPT's browse-mode source card surfaces author and date when article:author is set to a real person name. If the field is missing, the engine falls back to domain attribution. Perplexity does the same: a populated article:author lifts the named credit into the source list, where bookmarkable surface area is highest.
- •twitter:card is the one field with no OG fallback. Twitter falls back to OG tags for title, description, and image. It does not have an OG equivalent for the card-shape selector. Set twitter:card to summary_large_image for content that has a strong hero image; set summary for compact preview lines. Without twitter:card the post renders as plain text on X.
What a high-AI-pickup OG block looks like
The block below scores cleanly across all five social platforms and all three AI source-card layouts: absolute og:image URL, factual description front-loaded before character 160, article authorship and freshness populated, and twitter:card explicit.
<meta property="og:title" content="Open Graph Debugger for AI Search" />
<meta property="og:description" content="Preview how your URL renders on Facebook, Twitter, LinkedIn, Discord, Slack, and AI citation cards. Find missing og:image, og:title, og:description." />
<meta property="og:image" content="https://foglift.io/og/og-debugger.png" />
<meta property="og:url" content="https://foglift.io/tools/og-debugger" />
<meta property="og:type" content="article" />
<meta property="og:site_name" content="Foglift" />
<meta property="article:author" content="Foglift" />
<meta property="article:published_time" content="2026-05-11T06:00:00Z" />
<meta name="twitter:card" content="summary_large_image" />
<meta name="twitter:site" content="@foglift" />
<link rel="canonical" href="https://foglift.io/tools/og-debugger" />Two patterns frequently break this block in practice. A relative og:image path renders as a broken card thumbnail in citation panels (use absolute URLs only). An article:published_time more than 18 months old can lower how confidently AI engines surface the page when freshness is part of the query.
The 5 dimensions of an AI-ready OG block
These dimensions describe what AI source cards actually consume. They are distinct from a generic Facebook Sharing Debugger pass, which checks rendering but does not weight fields by AI extraction value. Step 1 is the highest-leverage. Step 5 is the easiest to overlook.
- Step 1. og:image is an absolute URL
- og:image refers to a fully qualified URL beginning with https://. AI engines fetch the image without resolving relative paths. The recommended size is 1200x630 pixels for the 1.91:1 ratio that Facebook, LinkedIn, Discord, Slack, and AI source cards consume. Keep the file under 5MB and serve it over HTTPS.
- Step 2. og:description is a factual abstract under 160 characters
- og:description is the line AI engines render in the citation card. It should read like a factual abstract, not ad copy. Front-load the substantive claim before character 160 so that Perplexity, Claude, and ChatGPT all render the same readable line.
- Step 3. og:type drives source-card confidence
- og:type signals the page shape. og:type=article unlocks article:author and article:published_time as additional citation fields. og:type=website is the safe default for landing pages. Pages with the article shape get cited at the source level more often.
- Step 4. article:author is a person, not a brand
- article:author should resolve to a person name when possible. ChatGPT browse-mode and Perplexity surface author names directly in their citation lists. A populated author signal lifts attribution from domain-level to source-level, which compounds over multiple cited articles by the same author.
- Step 5. twitter:card has no OG fallback
- twitter:card is the one Twitter field with no Open Graph fallback. Set it to summary_large_image for content with a hero image; set summary for compact tile previews. Without it the post renders as plain text on X. Other twitter:* fields fall back to og:title, og:description, og:image automatically.
Frequently Asked Questions
What is an Open Graph tag and why does AI search care about it?
Open Graph is a meta-tag protocol that Facebook introduced in 2010 and that every major social platform now consumes. AI search engines consume the same block as a structured-summary signal. ChatGPT, Claude, and Perplexity all parse og:title, og:description, and og:image to compose the citation card that surfaces alongside a generated answer. When two pages have similar topical relevance, the one with a complete OG block is cited more often because the engine has a confident card to render.
Why does og:image need to be an absolute URL?
AI engines render og:image directly inside the citation card. They do not resolve relative paths against the page URL the way a browser would. A relative og:image ("/img/hero.png") renders as a broken thumbnail. Use absolute URLs only ("https://example.com/img/hero.png"). A missing thumbnail visibly degrades the source card a user sees, which lowers click-through and, over time, lowers how often the engine surfaces the page.
What size should og:image be for AI source cards?
The single safe size is 1200x630 pixels (1.91:1 aspect ratio). Facebook, LinkedIn, Discord, Slack, and most AI source cards crop or pad to this ratio. Twitter summary_large_image cards prefer 1200x675 (16:9). Stay above 200x200 minimum and keep file size under 5MB. The image must also be reachable from a public URL, since AI engines fetch it server-side at render time.
Do I need separate Twitter Card tags if I have OG tags?
Not for content. Twitter falls back to OG tags automatically: twitter:title falls back to og:title, twitter:description to og:description, twitter:image to og:image. You should always set twitter:card (typically summary_large_image) because there is no OG equivalent for that field. Set separate twitter:* tags only when you want different copy or imagery on Twitter than on other platforms.
How do article:author and article:published_time change AI citations?
ChatGPT's browse-mode source card surfaces author and date when the article: fields are present, lifting the citation from domain-level attribution (the site says X) to source-level attribution (the article wrote X). Source-level attribution gets cited more often. Pages without these fields are still ingestible, but they cite at a lower confidence weight. The same pattern applies on Perplexity, which credits the author name in its source list when article:author is a real person.
Why does my social preview look different from what this tool shows?
Social platforms cache OG data the first time a URL is shared. If you have updated your tags, Facebook may still show the old version. Use the Facebook Sharing Debugger to force a re-scrape, the LinkedIn Post Inspector for LinkedIn, and the Twitter Card Validator for Twitter. Adding a cache-busting query parameter to the share URL also works in a pinch. AI engines refresh their parse cycle independently, typically within hours to days of a re-crawl.
What happens if I do not set any OG tags?
Without OG tags, social platforms guess your title and description from page content, usually with poor results. The preview image is missing entirely, which reduces click-through rates on social platforms by 50 to 80 percent in published benchmarks. AI search engines fall back to body-text extraction for the citation summary, which is less reliable and which makes the page less likely to be surfaced when an engine has to pick a single source to cite.
Is og:type required, and what value should I use?
og:type is technically optional, but AI engines treat the value as a hint about how to weight the rest of the block. "article" unlocks the article:author and article:published_time fields and tilts source-card confidence higher. "website" is the safe default for landing pages and tool pages. "product" is appropriate for ecommerce. Setting og:type to "article" without populating article:author or article:published_time still helps, because the page is flagged as a substantive piece rather than a generic web destination.
Sources & Further Reading
Primary specifications and platform documentation behind the dimensions and risk flags above. Truncation lengths and citation-panel rendering are observed during product testing and may shift as AI engines update their UI; the underlying field semantics are anchored in the specs below.
- Open Graph Protocol. Canonical specification for
og:title,og:description,og:image,og:type, andog:url, including the article subtype witharticle:authorandarticle:published_time. - Meta for Developers, “Sharing on Facebook” webmaster guide. Reference for og:image dimensions, the 1.91:1 aspect-ratio recommendation, and the Facebook Sharing Debugger workflow used to clear the platform-side OG cache.
- X (Twitter) Developer Platform, “About Cards”. Canonical reference for
twitter:cardvalues (summary, summary_large_image, app, player), twitter:* field fallbacks to OG, and image sizing for summary_large_image cards. - LinkedIn Post Inspector. LinkedIn's OG cache-refresh tool. LinkedIn caches OG data per URL on first share, so any update to og:title, og:description, or og:image requires running the inspector to force a re-scrape.
- Schema.org Article type. Canonical definition of
author,datePublished, anddateModifiedproperties referenced by AI source cards alongside the OG article fields. - WHATWG HTML Living Standard, §4.2 Document metadata. Canonical specification for the
<meta>element and the property/name/content attribute semantics that OG tags rely on. - Slack API, “Link unfurling”. Slack's OG/oEmbed consumption model for in-message previews. Slack caches link previews per channel, so URL updates require either a Slack-side cache refresh or a cache-busting query parameter.
- Foglift product testing. Citation-panel character cuts, og:image rendering, and authorship/freshness weighting described above are observed across ChatGPT browse mode, Claude web search, and Perplexity source cards. Numbers are approximate, not benchmarked against a fixed sample, and may shift as engines update their UI.