Photo Credit: ladyhawker
I do not like them,
I do not like
green eggs and ham.
“Don’t judge a book by its cover.” is a lesson we’ve all been taught at some point in life. The basic premise is that one should not judge the worth or value of something by its outward appearance alone. Yet each day hundreds and thousands of videos are judged by search engines. The question is how do we get past the book cover?
In a webinar given by Marc Gosschalk, Sr. Analyst, Product Management, comScore Europe he states, “The value of the online video marketplace is booming, with the role of video in the consumer lifecycle becoming increasingly recognized. According to comScore’s Video Metrix, 34.7 million people watched online video content for an average of nearly 17 hours each in November (2010).” He goes on to say, “As the online video market grows the need for visibility, transparency and accurate measurement is now more critical than ever.”
So let’s talk about video and visibility because it represents a problem for all parties involved – content producers, platforms and search engines alike. Nielsen reports from last May have Americans setting another record by streaming over 15 billion videos for the month. Total online viewers also increased by nearly 3% from April to top 145 million unique viewers. By 2013, Cisco predicts 90% of all Web traffic will be generated by video (NOTE: this statistic could be considered slightly misleading). With this explosion of online video, why does the level of viewer engagement seem to be inversely proportionate to its growth? Is this due to a consistently decreasing attention span of the viewer or because they have no clue what’s inside most videos (i.e. what’s coming next)?
Some fuzzy math.
Let’s assume that 90% of those 15 billion videos are on average 30 seconds to three minutes in length, and the rest are three to 30 minutes in length. If my math is correct (not likely btw), we’re left with approximately 2,902,500,000,000 (yes, trillion) seconds of video content. For sake of argument, let’s assume a new action or moment (change of scenery, new speaker, etc.) occurs within each video every 10 seconds. This leaves us with 290,250,000,000 (billion) missed opportunities to search within a video for more content/context. The guts of every video on the internet are being utterly overlooked and ignored by search engines. All we are really left with is what’s currently indexed by most search engines (Title, Description and Keywords) – the book cover.
A search for solutions
No solution exists inside a vacuum and each could play a significant role in the future of video search. Though possibly more, these are the three I’m currently aware of:
1. Transcription – Converting speech to text and matching the text with the corresponding moment within a video.
2. Video chaptering / tagging – Calling attention to moments within a video and labeling each with a title, comment or keyword (like a DVD).
3. Image recognition – Recognizing changes and/or objects within a video
What companies are currently trying to solve this problem? How do we spotlight each of those moments (frames) within content and show search sites (i.e. Google, Bing) it is what’s inside that counts? If searching by keywords associated with a video is not the best solution to video search, what is the alternative? How do we turn the page?