How Google Understands Videos, Summarised by Danielle Marshak

Debate has raged for some time as to exactly how (if at all) Google reads and interprets the content of videos published online. Accompanying textual information is used for SEO purposes, video titles, metas, descriptions and so on, but what about the actual content of the video?

The question, which has played heavy on the minds of SEOs for years, was recently put to Google’s Search Product Manager for Videos, Danielle Marshak.  In the latest Search Off the Record podcast, Google Search Central’s Gary Illyes and Lizzi Sassman brought Marshak along for the ride, and put her on the spot to settle this longstanding debate.

In specific, Illyes asked how Google takes into account the actual content of a video, rather than just the accompanying textual information. To which, Marshak stated that various components of the video’s content are read and analysed by Google’s crawlers.

A brief summary of her explanation:

• Google is now capable of reading and extracting characters from videos, which may include things like headings and captions that appear at various junctures to display important information.
• The audio track of a video file is scanned by Google to give its crawlers a better idea of what the video is about and what kind of language it contains.
• On a fairly rudimentary level, Google extracts and analyses various forms of visual information from videos, which can be anything from animals to objects to indications of location.
• Google also uses structured data to build a better understanding of a video’s content, Marshak explained.

On the subject of structured data, Marshak was adamant in highlighting the importance of continuing to prioritise structured data for the time being.

“Don’t abandon structure data just yet,” she advised.

“We love structured data, because as we were talking about, even though we’ve come so far, we’re still at the tip of the iceberg, in terms of being able to really deeply understand videos.”

All of the above would therefore suggest that Google is gradually approaching a stage where its search crawlers can read and interpret video content with a good degree of accuracy. Given time, this could prove to be a game changer.

Traditionally, it has exclusively been the textual content (metas, filenames etc.) that has been picked up by Google and used to assign rankings. If and when this data takes a firm backseat to the actual content of the video, it will radically transform the way SEOs and marketers approach the use of video content in their posts and pages.

Though when this will happen is anyone’s guess, as Marshak admitted that Google is still only at the “tip of the iceberg” where accurately reading video your content is concerned.