Home Blog Search Engine Optimization Video Search Engine Optimization Guide

Video Search Engine Optimization Guide

Q: Are AI Overviews showing my videos?

Check Search Console's Performance report: the Search appearance filter includes AI Overviews as a filter added in 2025. If your videos earn impressions there, you are being cited. Videos with Clip or SeekToAction markup, strong transcripts, and high engagement on YouTube are the most common citation sources.

Last Edited April 19, 2026
by Garenne Bigby

How to get your videos to rank in 2026

Video is no longer a side-channel for search. YouTube is the world’s second-largest search engine in its own right. Google Search mixes video results into the main results page, and AI Overviews cite video timestamps directly in their answers. TikTok, Instagram Reels, and YouTube Shorts dominate short-form discovery, especially among audiences under 30. A modern video SEO strategy covers three surfaces at once: Google Search (organic video results plus Key Moments and AI Overview citations), YouTube (the platform’s own search and recommendation engine), and short-form vertical video (TikTok, Shorts, Reels). The techniques overlap but the playbooks are not identical.

The fundamentals Google needs to show your video in search results haven’t changed since Search Central first documented them: a title, a description, and a thumbnail. What has changed is the machinery around those three facts — the schema, the sitemap format, the tools for testing, the way Google renders JavaScript, and what winning a video result in 2026 actually looks like on a SERP.

Where video SEO happens in 2026

Four discovery surfaces matter, in roughly this order for most businesses:

Google Search — main results. Google mixes video thumbnails into the organic results page for queries where a video is useful (how-tos, product demos, reviews). A single Key Moment can earn a separate rich result — a timestamped snippet that jumps the user directly to the relevant section.
YouTube Search and recommendations. YouTube handles its own indexing. Watch time, click-through rate on the thumbnail, and engagement signals drive discovery there; schema.org markup is irrelevant inside YouTube’s walled garden. But a video that ranks on YouTube is more likely to rank on Google Search as well, because Google’s video-results panel pulls heavily from YouTube.
Google AI Overviews. Since May 2024, AI Overviews have cited video sources — complete with timestamped jump-links into Key Moments — for queries where video is the clearest answer. Being cited in an AI Overview is the 2026 equivalent of ranking in the top three organic results a decade ago.
Short-form platforms. TikTok’s in-app search is now used by roughly 40% of Gen-Z users for discovery. Instagram Reels and YouTube Shorts have their own search surfaces. None of these respect Google’s sitemap or schema conventions — they have their own on-platform SEO rules (captions, hashtags, trending audio, early-frame hook).

The rest of this guide focuses on the technical work needed to win surfaces 1–3 (Google Search, YouTube, AI Overviews) for video you host on or embed into your own site. Short-form platform optimization is a separate topic.

The three facts every video needs

Whether a video lives on YouTube (embedded into your page) or on your own server, Google needs three pieces of information to index it:

Title — descriptive, unique per video, not keyword-stuffed. Matches the title the user will see when they land on the page.
Description — at least a sentence, ideally a short paragraph that explains what the video covers. Duplicate descriptions across multiple videos reliably hurt indexing.
Thumbnail URL — a publicly accessible image file. Current guidance: minimum 60×30 pixels, but 1280×720 or larger is recommended for modern result surfaces; 16:9 aspect ratio performs best in Google’s video carousels.

You communicate these facts to Google through one (or both) of two mechanisms: on-page structured data using schema.org, or a video sitemap. Both are invisible to site visitors and do not affect layout.

VideoObject structured data

Google’s canonical recommendation is schema.org’s VideoObject markup, delivered as JSON-LD in the page’s <head> or <body>. The vocabulary is maintained jointly by Google, Microsoft Bing, and Yandex (schema.org originally included Yahoo, which has been powered by Bing since 2009 and is no longer a distinct search engine).

Required VideoObject properties (Google will not treat the markup as eligible for rich results without these):

name — the video’s title
description
thumbnailUrl
uploadDate — ISO 8601 format

Recommended properties (add as many as apply; they improve eligibility and result appearance):

contentUrl — direct URL to the raw video file (mp4, webm, etc.)
embedUrl — URL to the embeddable player (YouTube embed URL, Vimeo embed URL, etc.) — supply this or contentUrl, or both
duration — ISO 8601 duration, e.g. PT5M30S for 5 minutes 30 seconds
interactionStatistic — watch count, likes
expires — use this if the video should stop appearing in search after a certain date
publication — for live streams, nest a BroadcastEvent here to earn the LIVE badge in search results
hasPart — array of Clip objects specifying Key Moment start times and labels
potentialAction — a SeekToAction telling Google where timestamps appear in your URL (as an alternative to Clip markup)

Key Moments with Clip and SeekToAction

Google’s Key Moments feature, introduced in 2020 and expanded since, breaks a video into labeled sections that appear directly in search results. The user clicks a timestamp label like “How to replace the filter” and jumps straight to that point. Two ways to enable Key Moments:

Clip structured data. Specify each segment manually with name, startOffset, and endOffset, nested inside the VideoObject’s hasPart property. Best for videos where you control chapter boundaries and want specific, editorial labels.
SeekToAction. Tell Google how your URL encodes timestamps (e.g., ?t=120), and let Google detect key moments automatically from the video content. Supported in English, Spanish, Portuguese, Italian, French, German, Japanese, Chinese, Korean, Dutch, Turkish, and Russian. Best for long-form content where manual segmentation isn’t practical.

Pages using either mechanism frequently earn richer search results — multiple clickable jump-points per video, which dramatically increase CTR vs. a single result link.

Video sitemaps — still useful, with caveats

A video sitemap is an XML document that declares video content and its metadata independently of the page-level schema. Video sitemaps remain a valid way to feed video data to Google, particularly useful when:

You host many videos and want Google to discover them without crawling every page.
Your videos live on pages that also contain structured data — the two sources are complementary, not contradictory.
You need to signal platform restrictions (<video:platform>), country restrictions (<video:restriction>), or expiration dates.

The format requirements haven’t changed meaningfully since this article was first written. Each sitemap entry requires: loc (play page URL), video:title, video:description, video:thumbnail_loc, and either video:content_loc (raw video URL) or video:player_loc (embedded player URL). One sitemap file holds up to 50,000 entries and 50 MB uncompressed; large libraries need a sitemap index.

What changed: in June 2023, Google deprecated the sitemap ping endpoint (www.google.com/ping?sitemap=...). Google now discovers sitemap updates via the sitemap’s Last-Modified HTTP header and the frequency listed in <lastmod>. Submit sitemaps through Google Search Console (renamed from Google Webmaster Tools back in May 2015) and let Google pick up changes on its own schedule — there is no manual re-submission needed when content changes.

Google crawls the following video formats: mp4, m4v, mpg, mpeg, wmv, mov, flv, asf, avi, ra, ram, and rm. Note that swf (Adobe Flash) is no longer supported — Adobe ended Flash support on December 31, 2020, and every major browser has since removed the runtime. Any Flash-based video content will not be indexed and should be migrated to HTML5 video (mp4/webm).

YouTube SEO essentials

If you publish video on YouTube, you’re working a different ranking system. YouTube’s own algorithm weighs watch time, retention curve, session length, click-through rate, and subscriber behavior — not schema or sitemaps. Practical essentials:

Thumbnail and title are the entire click-through decision. YouTube measures impressions vs. clicks; a low CTR signals the video isn’t worth recommending. Custom thumbnails (not auto-generated frames) consistently outperform defaults.
First 15 seconds determine retention. A sharp hook keeps watch time high; watch time is the single largest ranking input.
Description field is a lightweight keyword surface — the first 150 characters appear in the search snippet. Timestamps in the description auto-populate as chapter markers.
Chapters (from description timestamps) power YouTube’s own Key Moments. Well-chaptered videos rank for a broader set of intent variations.
Captions and transcripts. YouTube auto-generates captions, but manually corrected transcripts rank better and are genuinely accessible. Transcripts also feed Google’s understanding of what’s in the video for web search.
End screens and cards drive session length — recommending related videos from your channel keeps viewers in your ecosystem, which YouTube rewards.

A video that ranks well on YouTube gets two SEO benefits for free: it’s more likely to surface in Google’s Search video panel, and it earns the watch-time signals that make it a candidate for AI Overview citation.

Self-hosting vs. YouTube embed

The practical choice most teams make:

Embed YouTube videos when discoverability on YouTube itself matters, you want free CDN delivery, and you don’t need to gate the video. You sacrifice on-site dwell-time signals (viewers can click through to YouTube), but you gain massive platform reach.
Self-host (or use Vimeo/Cloudflare Stream/Mux) when the video is core to a conversion funnel, you need precise control over playback, analytics, or access, or when YouTube’s ads next to your content are unacceptable. Self-hosted video is where VideoObject schema and video sitemaps do their heaviest work, because Google has no other way to understand the content.
Both — upload to YouTube for reach, then embed the YouTube player on your own page with on-page VideoObject schema. You get YouTube’s discoverability plus Google’s web-search indexing of the same video.

Core Web Vitals and video

Video playback is one of the most common Core Web Vitals problems on content sites:

LCP (Largest Contentful Paint) — a hero video that auto-plays often becomes the LCP element; if the poster image loads slowly, LCP fails. Preload the poster image, defer the video source, and set explicit width/height attributes.
CLS (Cumulative Layout Shift) — video players that resize on load cause layout shift. Reserve the aspect-ratio box in CSS (aspect-ratio: 16 / 9) before the player scripts load.
INP (Interaction to Next Paint) — replaced FID in March 2024. Heavy player scripts block interactions. Lazy-load the YouTube IFrame API using the lite-youtube-embed pattern or similar; the full player only mounts when the user clicks play.

A single hero video on a landing page routinely accounts for 1–2 seconds of LCP. Fixing it is often the highest-impact page-speed change available — see our page speed testing tools guide for diagnostics.

Testing your markup

Google’s Rich Snippet Testing Tool was replaced by the Rich Results Test in 2017, and the old Structured Data Testing Tool was fully deprecated in August 2021. Current tools:

Rich Results Test — shows which rich result types your VideoObject markup is eligible for, including Key Moments, LIVE badge, and Video Host carousel.
Schema Markup Validator (schema.org’s own tool) — validates any schema.org type, not just the subset Google uses. Useful for debugging broader markup.
Search Console Video indexing report — shows which of your pages have been identified as video pages, which are indexed, and why individual videos were excluded. This is the authoritative source once the page is live.
URL Inspection tool (inside Search Console) — shows exactly how Googlebot rendered the page, including whether your JavaScript-driven video player was successfully rendered.

The old advice to “view your page in Lynx to see what Google sees” is no longer useful — Google’s renderer is now an evergreen Chromium-based engine that executes JavaScript. Test with the URL Inspection tool instead.

Common pitfalls in 2026

Duplicate metadata across videos — same title, description, or thumbnail on multiple videos confuses Google’s deduplication. Every video needs unique identifying data.
Thumbnail dimensions — images smaller than 60×30 pixels won’t be used. For Google’s video carousels and AI Overview citations, aim for 1280×720 or larger in 16:9 aspect ratio.
Hidden or off-screen videos — Google now reports these in Search Console’s Video indexing report. Videos inside accordions, carousels that require interaction to show, or tabs that hide content on load often aren’t indexed as videos.
Missing uploadDate — one of the most common VideoObject markup errors; without it, the markup is invalid.
Expired-date bugs — a past expires value silently removes the video from search results. Omit the property entirely unless the video actually has a real expiration.
No transcript or captions — Google increasingly relies on spoken content to understand video topics. Auto-generated captions help; manually reviewed transcripts help more.
Client-side-only video injection — if the video element is inserted by JavaScript after rendering, Google may not see it. Include at least a stub <video> element plus VideoObject JSON-LD in the server-rendered HTML.
Relying on sitemap ping — deprecated June 2023. Changes are picked up automatically from Last-Modified headers.

Platform and geo restrictions

Video sitemaps still support two restriction tags that can be useful for rights-managed content:

<video:platform> with relationship="allow" or "deny" — restrict indexing to specific device categories (web, mobile, tv).
<video:restriction> — allow or deny specific countries. If the allow list is empty, the video shows nowhere; if the deny list is empty, the video shows everywhere.

For videos that genuinely should appear in all countries on all platforms, omit both tags entirely.

Frequently asked questions

Do I need both a video sitemap and VideoObject schema?

No — either one is sufficient for Google to index the video, provided the required fields are present. Most sites benefit from having VideoObject schema on individual video pages (because it powers Key Moments and rich snippets directly) and a video sitemap for discoverability at scale. If you only have the budget for one, prioritize on-page VideoObject schema.

Does YouTube embed count as on-page video for SEO?

Yes, if you add your own VideoObject schema referencing the embed URL. Google treats the page as video content and can index it. The video itself still lives on YouTube and accrues watch-time signals there; your page earns the search-result benefits of hosting a video thematically relevant to its topic.

How long should a thumbnail be cached-fresh?

The thumbnail URL must remain reachable for as long as the video should appear in search. A 404 on the thumbnail URL removes the video from rich results. If you move or CDN-migrate images, keep redirects in place until the sitemap/schema is updated.

Are AI Overviews showing my videos?

Check Search Console’s Performance report: the Search appearance filter includes “AI Overviews” (rolled out as a filter in 2025). If your videos earn impressions there, you’re being cited. Videos with Clip or SeekToAction markup, strong transcripts, and high engagement on YouTube are the most common citation sources.

What replaced Google Webmaster Tools?

It was renamed to Google Search Console in May 2015. The feature set has expanded considerably — it now includes URL Inspection, Core Web Vitals reports, video indexing report, AI Overviews filter, and the structured-data enhancement reports that let you debug VideoObject markup at scale.

Is Flash video still a thing?

No. Adobe ended Flash Player support on December 31, 2020, and every major browser removed the runtime shortly after. The .swf extension is no longer in Google’s supported video formats, and any surviving Flash video content needs to be migrated to HTML5 (mp4, webm) to be indexable.

Bottom line

Video SEO in 2026 is three jobs: rank in Google Search (VideoObject schema + video sitemaps + Key Moments markup), win on YouTube (watch time, thumbnail CTR, transcripts, chapters), and earn citations in AI Overviews (strong metadata, timestamps, and content that actually answers the query). The fundamentals Google documented years ago — title, description, thumbnail — remain the required inputs. What’s evolved is the richness on top: Clip and SeekToAction for Key Moments, BroadcastEvent for LIVE badges, Core Web Vitals for video playback, and the shift from Flash-era players to HTML5 everywhere. Get the basics right, add schema where it pays, and test with the Rich Results Test and Search Console before assuming the markup works.