Video SEO Services for YouTube and On-Page Video
A video that nobody finds is a production cost, not an asset. We optimise both sides of the same problem: the video on YouTube and the player embedded in your product, service and guide pages — titles, chapters, captions, transcripts, schema, thumbnails and the page weight your embed adds.
- YouTube titles, descriptions and chapter markers
- Human-edited captions and transcripts
- VideoObject and key moments markup
- Thumbnail testing for click-through
- Facade embeds and page speed
Channel audit plus page-level embed review. Works with existing footage or a new series.
The timeline is the sitemap of a video
Chapters = queriesTwo channels, one video, different rules
YouTube ranks inside YouTube and inside Google's video results. Your embedded player ranks inside a page that also has to load quickly and convert. They share footage and metadata, but they fail for different reasons.
On YouTube, the constraints are title length, the first two description lines, chapter timestamps and watch time. On your site, the constraints are the page the video supports, the schema that describes it, and whether the embed delays the largest paint by four seconds.
We treat them as one production brief with two outputs. The footage gets cut once, the metadata gets written once, and the transcript serves the video, the page and the AI answer surface at the same time.
What we usually inherit
Forty uploads with "Product Video 3" as the title, auto-generated captions with the product name spelled wrong, no chapters, a description that repeats the title, and an embed on the homepage that costs 1.4 MB before the hero image loads.
None of that is a production problem. It is a packaging problem, and it is fixable without a new camera, a new studio or a new editor.
What good looks like
A title that front-loads the query. A first line that states who the video is for and what they will be able to do. Chapters that map to real searches. Captions a human has read end to end. A poster image that loads in under 100 ms.
Metadata that does the ranking work
Every field below is editable today, on videos you have already published. This is the highest-return hour in most video projects.
| Field | What it must contain | Common mistake | Where it surfaces |
|---|---|---|---|
| Title | Query phrase in the first 40 characters, then the differentiator | Brand name first, topic last | YouTube search, Google video tab |
| Description | Two-line summary, then chapters, then links and specs | Copy of the title plus hashtags | Snippet text under the video result |
| Chapters | 0:00 start plus at least three timestamps, each segment over 10 seconds | Timestamps in a pinned comment | Key moments in search, YouTube player |
| Captions | Human-corrected, correct product nouns and units | Raw auto-captions with model numbers wrong | Silent autoplay feeds, accessibility, indexing |
| Transcript | On the page below the player, with headings | Hidden in a toggle nobody opens | Page text, AI answers, snippet selection |
| VideoObject | name, description, thumbnailUrl, uploadDate, duration, embedUrl | Markup for a video not on the page | Video rich results, entity understanding |
| Thumbnail | 1280×720, one subject, three words maximum, high contrast | Auto-selected frame with burned-in subtitles | Click-through rate everywhere |
Anatomy of a page that ranks its video
- 1A page, not a video dumpOne video per intent, on the page that already targets that query. A video library page ranks for nothing in particular.
- 2Poster and facadeThe player loads on click. The poster image is a real file with width and height set, so nothing shifts when it appears.
- 3Text that repeats the promiseAn H2 above the player and a 120-word summary below it. The video supports the text; it does not replace it.
- 4Chapters on the page tooA short list of what the video covers, each item linking to its timestamp where the platform allows it.
- 5VideoObject, onceOne block per page, matching the visible video. Duration in ISO 8601, upload date in full, thumbnail URL that resolves.
- 6Transcript with headingsSegmented by chapter, not one 4,000-word wall. This is the text search engines and answer engines actually quote.
"@type": "VideoObject", "name": "How to specify a flanged connection", "duration": "PT9M2S", "uploadDate": "2025-03-11", "hasPart": [ 6 chapter clips ]
Captions, transcripts and the AI answer surface
Speech is text you have already paid for. Right now it sits locked inside an audio track while answer engines quote competitors who published theirs.
Why auto-captions are not enough
Automatic captions mishear product nouns, units and part numbers — exactly the terms that carry commercial intent. "Four-inch flange" becomes "for inch flange", the transcript says nothing useful, and the video loses the query it was made to answer.
We edit captions by hand on the segments that matter: the specification, the installation step, the safety callout. The rest can stay rough if it is readable.
How the transcript earns its place
We publish it below the player, split under the same chapter headings, with the specification table pulled out as real text. That gives the page indexable content the video alone never provides, and gives answer engines clean sentences to quote.
The transcript also improves accessibility, which matters for procurement teams in regulated markets. It is the rare optimisation that helps search and users at the same time.
The writing standard here is the same one we apply to articles and guides: specific, sourced, no filler paragraphs. If the transcript reads like a script, the video will too — which is why the copy and the footage are briefed together in On-Page SEO Services.
Thumbnails decide whether any of it matters
A perfect description under a thumbnail nobody clicks is inventory, not marketing. We build three variants per video: a product close-up, a person mid-task, and a diagram frame with three words. Then we test them on the placements that show impressions and let the click-through rate decide.
Rules that hold across categories: one subject, high contrast at 120 px wide, no burned-in subtitles, no small text, and a face or a hand where the product alone is abstract. Text overlays stay under four words and never repeat the title.
Embeds and the speed budget
A standard YouTube iframe pulls roughly a megabyte of script before a frame renders. On a mobile connection that is the difference between a passing and failing Largest Contentful Paint, and it competes with the hero image for bandwidth.
We replace the iframe with a poster image and a click-to-load facade that injects the player on demand. The video still plays exactly as before. The page simply stops paying for it before anyone presses play.
For pages where video is central, we also check layout stability: fixed aspect ratios, reserved space, no injected banners above the player. And we make sure the embed never runs alongside three other embeds on the same page. This sits inside the wider page-level work we do in SEO Content Writing Services.
How the engagement runs
Every video, both channels. Titles, durations, chapters, caption quality, embed weights and the pages they sit on. We rank them by traffic potential, not by upload date.
Metadata across the catalogue. Titles, descriptions, chapter sets and captions for the top 20 videos first, then batches. Existing uploads keep their URLs and view history.
One video, one page, one intent. Facade embeds, transcripts, VideoObject markup, internal links from the relevant product and guide pages.
Sitemap and schema checks. Confirm the video URL is crawlable, the thumbnail resolves, and the markup validates before requesting indexing.
28 and 56 days. Impressions and CTR on video results, average view duration, plays from embedded players, and inquiries attributed to video landing pages.
What changed after the work
An industrial equipment manufacturer had 14 product videos with generic titles. After chapters, corrected captions and page-level markup, inquiries from video landing pages moved from 2 to 11 per month.
Industrial equipment / 6 monthsA SaaS onboarding tool published a 12-part series with transcripts. Video rich results appeared for 38 queries, and demo requests from those pages rose 64% quarter over quarter.
SaaS / 4 monthsA furniture exporter replaced three homepage embeds with a facade. Mobile LCP fell from 4.6 to 1.9 seconds, and video plays per session did not drop.
Furniture / 2 monthsSend one video link
We will tell you why it is not being found, and what the first three changes would be.
Scope: channels, batches and embeds
Video work is quoted per channel and per batch of uploads, plus page-level work for embeds. No price list, because a 14-video catalogue and a 400-video archive are not the same job.
We start with the videos that already have impressions but poor click-through — those are the fastest wins. New production is scoped separately and only if the existing library cannot answer the query.
Questions we answer before we start
Do we need a YouTube channel?
No. Self-hosted video with VideoObject markup can win video results on its own. A channel helps distribution and gives you a second search surface, but it is not a prerequisite.
How long until a video ranks?
Metadata changes usually show movement within two to six weeks. A brand-new video in a competitive category can take a quarter. We report leading metrics — impressions, chapter impressions, CTR — before positions.
Can you fix videos we already published?
Yes, and that is most of the work. Titles, descriptions, chapters and captions are editable after upload. Changing them does not reset views or delete comments.
Do captions really matter for search?
They are the only reliable machine-readable text of the audio. They improve accessibility, support silent viewing and give search engines words to match. Auto-captions help, but they mishear exactly the technical nouns that matter.
Will embedding video hurt page speed?
Only if it is left unmanaged. A facade embed with a poster image and click-to-load keeps the cost off first paint. Several autoplay embeds on one page will always hurt.
Is this white hat?
Yes. We optimise real video with accurate metadata and genuine transcripts. No cloaking, no fake durations, no markup for videos that are not on the page. Work runs per project or on a monthly retainer; you keep channel and CMS ownership and sign off on every metadata batch.
How do you measure success?
Video impressions and CTR, average view duration, plays from embedded players, and inquiries attributed to pages where the video is the main entry point. Views alone are not a business metric.