Why Google stopped showing your video, and what it takes to get it back

What changed, and what did not

The change is narrower than the panic around it suggested. Google did not stop indexing videos. It stopped showing a video thumbnail beside results where the video is secondary to the page, which is most marketing pages: a product page with a sixty second explainer under the headline, a blog post with a demo embedded halfway down.

Gabe's point in the aftermath is the one to hold on to, and he made it against his own headline: a drop in video snippets does not mean a drop in traffic. Rankings did not move. Sometimes an image thumbnail replaced the video one. What fell was a specific rich result, and reading a Search Console chart of video impressions as a traffic chart is how teams spend a quarter fixing something that was not broken.

What did change materially is eligibility. If you want the video result, the page has to be built as a page about that video.

A drop in video snippets does not mean a drop in traffic, and reading one chart as the other has cost teams a quarter.

The four conditions, in Google's words

Google publishes the criteria plainly, and they are worth quoting rather than paraphrasing. The watch page must be indexed. The indexed watch page must be performing well in Search before its video can be considered for indexing. The video must be embedded on a watch page, and it cannot be hidden behind other elements. The video must have a valid thumbnail that is available at a stable URL.

The second condition is the one that surprises people. Video indexing is downstream of ordinary page performance: a page that nobody ranks for does not get its video indexed either, so there is no video-shaped shortcut past a weak page. The video result is a reward for a page that already works.

The third condition rules out a common pattern. A player that only loads after a click, showing a poster image until then, gives Google nothing to measure, and Search Console reports it in as many words: cannot determine video position and size.

The thumbnail is half the failure modes

The video indexing report has 14 distinct reasons for a video not being indexed, and 6 of them are about the thumbnail. It must be at a stable URL, in BMP, GIF, JPEG, PNG, WebP, SVG or AVIF, at least 60 by 30 pixels with larger preferred, reachable by Googlebot without a robots.txt rule or a login in the way, and at least 80 per cent of its pixels must have an alpha value above 250.

That last rule catches teams who export a thumbnail with a transparent background because it looked neater in the design system. It is also the kind of thing nobody discovers by reading the documentation, only by reading the report after the video has been invisible for two months.

The stable URL requirement is worth a policy rather than a fix. Thumbnails generated per deploy, or served from a URL that carries a cache-busting hash, are a slow leak: each release quietly un-indexes the videos published before it.

Structured data: three required properties

VideoObject asks for three things and only three: name, thumbnailUrl and uploadDate. Google specifies that the name must be unique text per video, that thumbnailUrl points to the video's own thumbnail image file, and that uploadDate is ISO 8601.

Everything else is recommended rather than required: contentUrl, description, duration, embedUrl, expires, interactionStatistic, and the region properties. contentUrl and duration are the two worth treating as required anyway, because they are what let Google understand the file rather than only the page around it.

Key moments are a separate feature with three routes: Clip markup, where you give the start time, end time and label of each segment; SeekToAction, where you tell Google how timestamps appear in your URL structure and let it choose the moments; or timestamps in a YouTube description. Clip is the one that gives you control over the labels, which is the whole point of having them.

So where should the video live?

If the video matters to search, it needs a page of its own, and that page needs to be worth ranking on its own terms. A title, a real description that is not the marketing blurb repeated, a transcript, the player above the fold and loaded on arrival, a stable thumbnail, and VideoObject markup that agrees with all of it.

That is a different artefact from the video on your product page, and it is fine for both to exist. The product page video is there to convert someone who already arrived. The watch page is there to be found. Trying to make one page do both is how you end up with a page that is not the main content of anything.

The uncomfortable conclusion for most marketing sites is that the video in the hero, the one everyone argued about, was never going to appear in Search and never needed to. Judge it on what it is for.

What to do about it

1. Open the video indexing report in Search Console and read the reason, because Google names it in 14 different ways and each has a different fix.
2. Give any video that matters to search its own watch page, with the player loaded on arrival and a transcript on the page.
3. Freeze your thumbnail URLs. A hash that changes per deploy quietly un-indexes every video published before it.