Roughly 2 weeks ago, I had the pleasure of speaking with Tom Wilde, CEO of EveryZing. I was incredibly impressed both with Tom's expertise regarding the state of online video and general expertise with regards to SEO. In addition, it was enlightening to learn more about EveryZing and the technology it uses to index and serve online videos. The implications for on-page SEO for video and audio are tremendous.
As Tom put it, "Text is the "gulfstream" of the Internet." In other words, the Web and search are built upon indexing and categorizing text. This creates a fundamental challenge for producers of multimedia content that goes beyond simple text.
Tom is widely recognized as a leader in the field of online search and advertising, and prior to becoming CEO of EveryZing (formerly Podzinger), he held numerous leadership roles in the field, including FAST, Lycos, and NameMedia. He has also served on the IAB Search Engine Committee and holds an MBA from the Wharton School at the University of Pennsylvania.
We discussed the state of online video both for today and in the future (see Tom's guest post titled "2008 and Future Online Video Value Chain Predictions").
"Our fundamental belief is that the challenge is not to build a better search engine for videos, but rather making videos better for search engines " – Tom Wilde
I also asked Tom to tell us how EveryZing's speech recognition technology is helping partners index and gain additional exposure to their multimedia content via video SEO. (You may have seen Beet.tv's interview with Tom where he discusses how EveryZing's speech-to-text technology helps index and maximize the discoverability of Boston.com's multimedia content).
Media companies are trying to figure out how to maximize exposure and monetization of their online multimedia content. Media companies have massive amounts of multimedia content available online but making that content searchable and discoverable to Internet users is important to increase consumption and allow for increased revenue opportunities.
EveryZing is the only company that I am aware of that offers media companies a white-label solution to provide sophisticated content analysis and navigation within video and audio content. With EveryZing's technology, media companies can create a more robust set of metadata surrounding content, which ultimately leads to increased discoverability via on-page video SEO. More details and background on EveryZing here.
Let's use BostonTV as an example to demonstrate the power of EveryZing's technology with regard to video SEO:
While I will not go into depth regarding on-page SEO efforts (topic density, freshness, relevancy, good page structure, etc….), as that would take too long, keep this concept in mind with regard to relevant on-page SEO.
Increased Discoverability and Indexing:
Go to Boston.com and conduct a site search for "Kevin Garnett" (Celtics player). You will see the following results displayed in the multimedia column of the search results page:
If you click on "View all 90 results," you will be taken to a page with results for multimedia content relevant to "Kevin Garnett.
"How is this significant?
Look at the results carefully, and you will notice on the first results page that only 4 of the 10 videos displayed actually have "Garnett" in the video title. In fact, only 1 of the 10 videos on this page carries the phrase, "Kevin Garnett" in the video title. If you were looking for videos relevant to "Kevin Garnett," you would likely be interested in any video that discusses Kevin Garnett. Imagine the effect on traffic and monetization when you can provide users with many more results that contain relevant content to their search query within video and audio.
Further more, when you click on a given result, you will notice that the video begins playing at the exact time at which Kevin Garnett is mentioned. In fact, you will see on some videos where there is more than one mentioned, a flag to show exactly where the phrase can be found on the timeline. EveryZing creates an enhanced navigational experience with its patent pending "snippitization" and jump-to functionality.
To demonstrate the power this has with regard to indexing within search engines, according to Google, there are 8290 pages of multimedia content indexed – site:http://multimedia.boston.com. According to Tom, more than 5000 pages were indexed within Google immediately (within 1 day) after EverZing was launched on Boston.com.
Related and Relevant Content:
When a user conducts a search for "Kevin Garnett," I feel that it is a safe assumption that they would be pleased to find all results that are relevant to the query, not just those that are an exact match. For example, the same user may be interested in any video about the Celtics. That is the power of EveryZing's speech-to-text recognition technology, instead of returning only videos that contain original metadata with "Kevin Garnett," Boston.com can now show users all multimedia content that even mentions Kevin Garnett, or related terms, like "Celtics," "Basketball," etc….Because EveryZing analyzes each word within the text transcripts, they also have the ability to tag videos with semantically related terms using EveryZing's natural language processing technology. As a result of this, they are able to easily create multimedia category pages centered around key related terms. On the results page for the search conducted above, you will find the following categories off to the right of the page facilitating greater interaction with users and guiding users to additional content that may be of interest.
On-Page SEO Enhancements:
If you read any guide on Video SEO (ours is here), you will see one common tip is to surround the page html containing the video with relevant metadata and text. It is important to have your keywords within the title, description, tags, and within the page itself that contains the video. Why is this so necessary? As Tom reminds us, one of the biggest misses these days with regard to online video is the prevalence and dominance of Flash for video players. As most search engine optimization specialists know, search engines are not yet able to read Flash content, and therefore, the only way to decipher the content of a video, is through looking at other text on the page, as well as off-page factors like anchor text in inbound links.What if you could include relevant text and metadata automatically using the full-text transcripts from a video? In fact, what if you could also include the entire text of the video transcript? What if you could also then include links to other relevant content, including videos?
Question: Which version do you think would be indexed more easily for the phrase "Kevin Garnett”?
A) Old Version – no text on page at all, flash player with text description inside flash file.
Title Tag: "http://www.boston.com – Video – NECN.com”
Meta Description: "”
Meta Robots: content=”noarchive”
B) New Version with EveryZing – Text in title of page, video title, video description, url, related content links, full text transcript, etc…
Title Tag: "Garnett, Allen on the garden's atmosphere Audio and Video brought to you by boston.com "
Meta Description: "Watch and Listen to Garnett, Allen on the garden's atmosphere videos and audio clips on boston.com”
Meta Robots: None
Answer) If you guessed Version B, you are right and you probably know something about search engine optimization.
How is this facilitated?
EveryZing's speech recognition technology is able to create a complete text transcript of each video or audio file. I am told that this recognition technology is 85% accurate which is incredibly accurate for speech recognition these days.When a video or audio file is processed by EveryZing, each word contained in the speech transcript is time stamped to match where the word was mentioned. In this way, users can navigate within multimedia content for the first time. Users can go directly to the time frame of the video in which the word or phrase they are looking for was mentioned.Rather than explain this in detail, there is a great video available from BeetTV where Tom discusses how this is accomplished.
More About EveryZing:
PodZinger changed its name to EveryZing in June 2007. With the change, it transformed itself into a media merchandising platform that helps content producers and Web publishers to dynamically increase the volume of consumable online content at the same time that its enhancing its monetary value. Based off of BBN's world class speech-to-text technology, EveryZing has the ability to extract the full text from audio and video Web files, creating a new pattern in the consumption of online multimedia content. The technology also brings targeted advertising to the world of multimedia, and for the first time, allows users to navigate within multimedia files. EveryZing is able to perform multimedia searches in both English and Spanish. EveryZing has its corporate headquarters in Cambridge, Mass.
EveryZing highlights the particular segment of an audio or video file by creating text snippets around the searched terms. The user has the option to click anywhere in the snippet to listen or watch just that segment, download the file or subscribe to a RSS feed. EveryZing indexes new content as it is added, sometime through an RSS feed that automatically delivers new content from a Web site. Following a user's search query, EveryZing provides related alternative keywords through its natural language analysis technology. At present, EveryZing is addressing three market segments. The first segment is the major content producers, such as the large television networks, who can monetize their content with ads. In such cases EveryZing offers a white label service and charges a fixed licensing fee. The second segment is the radio stations that want to put their content on a Web site, where the content provider is unsure of the amount of ad revenue. The third category is smaller content providers that want to publicize their content. Such customers can also have a search box on their Web site, with a revenue-sharing model.
Recent additions to the customer base of EveryZing include Boston.com; WEEI, WAAF, and WRKO; 890 ESPN; and boston.tv. Audio search expands the life of multimedia material. The ability to search the multimedia files helps to increase the viewer/listener base. EveryZing will support advertisements for some of the company's top advertising sponsors. EveryZing captures consumers' intent, classifies searched content and aggregates consumer usage. This information provides the essential elements for designing advertisements specifically tailored to consumers' individual interests, and in turn, increases revenue for the advertiser and content creator.