The cable industry’s “TV Anywhere” strategy is much in the news this month, as a response to the rapid adoption by consumers of sites and applications such as Hulu and Boxee. The cable industry is loath to make the mistake the record companies did, which was to misinterpret consumer downloading of music as a desire to steal content instead of simply a desire to have a more useful form factor for their music. Indeed, iTunes emphatically showed the market that consumers are quite willing to pay to have their music, whether in the form of the billions of dollars in hardware they purchase annually (I think there are now a dozen ipod’s in my household) or the download fees iTunes charges with a simple one-click method.

The current state of internet television is one of frustration. Hulu’s growth is in direct proportion to this frustration, and the cable companies and cable networks need to react with a product that meets consumer needs.

TV Anywhere attempts to do just that. Much of the commentary around TV Anywhere are rants about having to pay the cable company for programming online, and risks that the web becomes “channelized” rather than open. ESPN360 is another example of TV Anywhere, which defies the common wisdom that content should be ad-supported and “free,” or that the content providers should rush to have direct relationships with consumers. These approaches miss two fundamental realities. First, not all content can be supported through ads. Premium producers like HBO couldn’t support their business with ads, in that there continues to be relentless downward pressure on ad prices as inventory expands across traditional, mobile and web platforms. Second, it ignores the fact that delivery of the content requires massive infrastructure, primarily in the form of the physical plant of broadband ISPs. If they don’t have a financial incentive to continue to build out their infrastructure, then innovation and deployment will stall.

The challenge, however, is that the real opportunity to drive additional consumption and to create perceived value for the consumer isn’t being addressed. Consumers’ negative reaction to TVAnywhere is primarily that they fear it is a path for the cable companies to charge incremental fees in the future (it is) without an understanding of the incremental value.

On the other hand , the current path of TVAnywhere and IPTV in general poses some risks to the content producers. In the current linear world, a cable channel owns a defined piece of real estate on the cable system “Channel XYZ.” The ability to cross promote other shows, the network and related properties is a powerful aspect of having control over this “linear” piece of real estate. Indeed, this was the justification that Lifetime used recently for paying a 40% premium for Project Runway. If Project Runway can attract a major audience segment, it can be a springboard for success for other Lifetime shows and indeed the whole network. However, IPTV and TVAnywhere pose a real risk to this reliable strategy. Internet television explodes the traditional “linear” real estate into atomic elements in a non-linear experience. As a result, the cross promotion effects of a hit show are substantially undermined. At the same time, this “infinite” shelf space creates frustration for the consumer, as finding relevant content moves from reviewing a linear program guide to understanding non-intuitive and sometimes arbitrary menu structures. This is not unlike the early days of the web. Remember Yahoo or the Open Directory Project’s attempts to neatly classify every website?

In the new world of TVAnywhere, the burden on the cable channels and content producers in general means the mindset needs to shift from that of “programmers” in the traditional TV sense to that of “merchandisers” of content. This means that the content needs to be wrapped with features that will help the end user more deeply experience the content. This would include some of the following:

• Improved search: Not only search across a video catalog, but also search within a video (characters, dialogues, faces, products) and search for related content (profiles, blogs, fansites, etc)
• Show buzz: Mining the web for tweets, Facebook posts, blog comments, etc.
• Content mashups: Allowing viewers to clip, share, comment within, and generally “mashup” the show content as a viral catalyst.

Search in particular becomes an important “tether,” that can connect an atomic piece of content to a large context. A well formed stored-search associated with a particular episode allows all related content to be delivered to the end user dynamically and contextually. This content “merchandising” will convey the added value that interactivity delivers and web users expect. This in turn enhances the perceived value and creates new revenue streams into the future for the content owners.

As has anyone in the online industry, I have closely followed the YouTube saga since its acquisition by Google.  Undoubtedly YouTube is one of the greatest success stories of the Internet era, adding users and usage faster than any site ever launched on the web, including MySpace, Facebook, Twitter, and Google.  What is also often missed is that YouTube is the third largest search site on the Web in terms of query volume, behind only Google and Yahoo.  Despite these impressive metrics, YouTube has failed to deliver meaningful revenue and loses approximately $1MM/day due to its huge bandwidth costs.

YouTube’s more recent strategy has been to pursue premium content as a means to generate revenue.  All video metrics point to the reality that premium content and professional content is the only real moneymaking video opportunity on the web, as Hulu has demonstrated most recently.  YouTube’s original premium content strategy was to approach major broadcasters and extol the virtues of its huge audience combined with the Google advertising prowess.  The pitch was essentially “give us all your content and we will send you a big check each month.”  The problem was that major media companies had already witnessed the loss of channel power from Google web search, and sat by frustrated while Google made billions from its content aggregation model in the form of Google web search.  These lessons wouldn’t be forgotten and Google’s pitch fell on deaf ears.

YouTube has had recent success signing premium content providers, most significantly Disney, but with three huge concessions to its previous business model: 1) Disney, and more specifically ESPN, will be allowed to monetize its content being viewed on YouTube;  2) ESPN will be allowed to embed its own video player on YouTube; and 3) Disney will be given a branded “walled-garden” to display its content within YouTube.  This deal solves the three key concerns media companies have about working with YouTube. 

Media companies are in business to sell advertising, and their trusted brands can deliver the dollars.  Second, they care deeply about their brands, and the ability to control the look and feel of the player, as well as protecting the brand from unfiltered user-generated content is a critical requirement that has historically kept most brands off of YouTube.  On the surface this seems like a win for YouTube, but I would argue it’s the first crack in the strategy that has made YouTube so successful, and that is the ubiquity of its brand and the ubiquity of its user experience.  A few months ago I argued that user-generated video is like instant messaging in that everyone uses it and no one makes money on it.  How then can YouTube find a business model?

Clayton Christensen, the noted Harvard Business School strategy professor, wrote the seminal work on the disruption of the incumbent’s business model called “The Innovators Dilemma.”  Essentially the book outlines the challenges a company faces when a young upstart creates a disruptive new business model.  Ironically, despite its own young age, in this case YouTube faces disruption from Hulu.  In just a short time, Hulu has executed brilliantly on its premium long-form content strategy and delivered early success for its owners.  Amazingly, it has YouTube turning itself inside-out to figure out how to compete.  This is precisely the wrong tactic for YouTube to take.  History has shown again and again that companies cannot fundamentally redefine themselves in order to respond to a competitive threat.   Everything that has made YouTube successful will work against it in trying to compete head on with Hulu.

Instead, YouTube must exploit its assets to build a sustainable business compatible with its original strategy.  In YouTube’s case, their most valuable asset by far is their search volume.  However, what continues to amaze me is that despite having the most talented search technologists on the planet, YouTube’s search and related content algorithms are shockingly poor.  With that as context, here is my prescription for YouTube to finding its own financial success.

What Google needs to do to fix YouTube

Fix the search algorithm. I  mean really? Really?  Have you really looked at YouTube’s or Google Video’s search results recently?  Can anyone in the search industry unpack their algorithm for me?  I’ve been in the search business for over a decade and I cannot figure out what they are ranking on, and certainly can’t imagine they look at these results and feel good about them.  Part of the challenge here is that videos are fundamentally different than text documents.  First, there is a paucity of content to index.  Most videos have just a generic title and description and some user generated tags.  I can tell you that we have looked at the user tags of over 2 million YouTube videos, and the same spamming problems that exist on the Web exist within YouTube. 

What’s needed is a more robust approach to creating high quality metadata for proper indexing.  Second, a better popularity and quality score needs to be integrated, because videos are also lacking the link data that made PageRank so successful for ranking web documents.  Also, how about some search syntax specific to video.  It would be very useful to have operators similar to what exists on Google web search.  For example, a syntax such as “title:keyword” for searching just within the title or “transcript:keyword” for searching just within the transcript or “source:SampleSource keyword” for searching within a certain source like FoxNews etc.

Merge the Google video search into the YouTube search box. When’s the last time you used Google Video search?  Its not even available from the home page of Google, and represents one of the strangest user experiences in all of search.  The page layout is impossible to decipher, and it seems unable to decide what master its serving.  Is it driving traffic to YouTube?  Sometimes—although via an embedded YouTube player.  Is it the world’s biggest video index of the Web?  Not even close, when compared to Blinkx or Truveo.  I have no idea what this product is trying to be.

Make YouTube the web’s video search engine. As I asserted above, user generated content will never produce significant revenue for YouTube.  By its very nature it is virtually impossible to monetize directly.  However, it is one of the Web’s most valuable loss-leaders.  This content has enabled YouTube to become the web’s 3rd biggest search engine, and its time to exploit that asset.  To do this, YouTube needs to build a content submission API that gives content owners the option to submit the entire video or just a deep link to a video on the content owner’s website.  It needs to include a comprehensive set of all possible metadata to finally give YouTube the raw material it needs to build a proper search algorithm.  Then its needs to redesign its search results page to help the user decide what kind of content they wish to view, be it UGC on YouTube, premium content on YouTube, or links to content off YouTube.

Create a business model that works. To me, this answer has always seemed obvious in that there is a perfect precedent for premium content publishers.  For decades, premium content television producers have paid cable operators huge fees to carry their channels.  Why do they do this?  To create the consumption they need to have an advertising business model, they need to have audience.  Carriage fees gives them a predictable audience they can monetize.  Merging this model with YouTube’s search asset is the path to YouTube revenue. 

If I were YouTube, I would deploy a two-part business model.  Premium publishers would purchase shelf space each month depending on the number of clips submitted for indexing.  This shelf space would buy them a set number of “clicks” on their content, and those clicks could either resolve to content uploaded to YouTube or back to the publisher’s own web site.  In addition, publishers would be allowed to “bid” for clicks in excess of their included amount to drive additional views.  Obviously the CPC on YouTube would be a fraction of what it is on web search (remember—a $.10 CPC is equivalent to a $100 CPM).  This means that average CPCs would probably hover around a penny or two, creating the need to allow for tenths of a cent bidding increments.  As ad dollars continue to flow online, publishers will be in a position to pay more over time.  In the short term, the carriage fees allows them to build a more predictable audience and for YouTube to build a recurring revenue model.

YouTube’s current gyrations chasing Hulu are the best thing to happen to Hulu.  Hulu’s DNA is about premium long form content.  Everything they do is geared towards this mission, and they do it extremely well.  Nothing in YouTube’s DNA gives them any advantage for this market.  YouTube needs to exploit its core assets and advantages to find its own business model.

Nielsen Research came out with interesting statistics on overall video consumption last week indicating that total TV consumption reached record levels when considering traditional broadcast video, time shifted video, web video and mobile video. This seems to fly in the face of the predictions that online video is cannibalizing the traditional business model. The web and mobile “screens” as incremental is a critical piece of the landscape if big media is to weather the transition from traditional consumption models to new models.

The Nielsen report echoes what ESPN also found in a study they commissioned with the Center for Media Design at Ball State University which found that there was a strong correlation between the time spent online with a media brand and time spent consuming traditional media, suggesting that multiple screens strengthened the relationship with the viewer. However, digging deeper into these reports indicates that there is also a marked difference in viewing behavior which is age-dependent. Not surprisingly, younger viewers skew sharply towards online viewing, which suggests increased pressure in the future on traditional business models including DVD sales, broadcast television, and movies.

Elsewhere on the web in the past few weeks, online TV and movie provider Hulu announced that it was pulling its content from Boxee due to pressure from its content partners. There is growing concern that the younger generations are “cord cutting” and relying solely on a fast Internet connection and online content, which of course Hulu (via Boxee in this case) is a big enabler.

Interestingly, scrolling through the comments on the Hulu blogs one finds a high level of outrage at this decision, with many voicing bewilderment. Yet at the same time the comments carry some confirmation of the very problem the content owners face, in that many commenters are lamenting the fact that they have cancelled their cable subscriptions in lieu of Boxee and are now left without their content.

The big win for professional broadcasters is the clear evidence that high quality content can translate into revenue. Hulu represents a small fraction of YouTube in terms of users and streams, but monetizes its audience at a far superior level. The challenge then for traditional broadcasters continues to be how to ensure that while the revenue per episode is likely to fall, a growth in total consumption can offset a large part of the risk to the shift in viewer habits. How to achieve this?

One critical need for the new multi-screen reality is a new metaphor for programming, and this is where search can play a vital role. In a world where users are increasingly in control of the content they consume, search is the perfect approach to delivering individualized content discovery and programming which will extend users’ consumption and contribute to a successful transition in business models for professional media producers. To achieve this, publishers need to be acutely aware of how their content is discovered and organized across the three screen reality (four if you count IPTV and time shifted television viewing).

This discovery begins with having a rich set of metadata around each content object that is produced, and having an adequate set of transcoded versions of each object to ensure ubiquitous consumption capabilities. With structured data surrounding unstructured audio and video clips, search engines are able to index the content more successfully, and directories are able to effectively categorize (i.e. “channelize”) the content. This in turn means the end user can find what they are looking for quickly and therefore consume more. Compare the search experience on Hulu for professional content with the search experience on YouTube and you can quickly see the power of good search as a “virtual” programming tool.

Robust content indexing also allows for features such as automatic and dynamic playlisting, so users are not only discovering content today but describing what content they want to discover in the future. There is simply no analog for this in the traditional broadcast environment and suggest one of many ways that search will yield more total consumption of media. In addition, proper metadata and indexing approaches mean that users are in a position not only to discover their desired content quickly but also consume only the portions of the content that they are interested in. Finally, great search means users are more likely to share content they have consumed with their social graph via clipping and commenting. The Web has shown that the more users are in control of their content experience, the more they will consume.

As we move into the new year, the “Web 3.0” references appear to be picking up speed, and most define Web 3.0 closely along the lines of the social graph, with Facebook and Twitter as the poster children.  Its not so much social media though, that is powering a fundamental shift on the web, but rather a change in how content is created, organized, indexed, and consumed.

Perhaps the biggest shift is embodied in how we all think about content.  If you think back just a few years, the content landscape was dominated by a notion of “containers”.  Regardless of content type, the container drove most of the organization, commerce, and consumption of content.  Music is the easiest illustration.  While record companies had oriented themselves around “albums” and “catalogs”, they missed the consumer shift toward a focus on the individual tracks.  The arrival of the iPod was the gasoline on the fire, providing a simple-to-use device and software for purchasing and consuming content at the “object” level (a digital track) rather than the container level (an album).  This has had profound effects on the industry.  An album was the ultimate bundling strategy.  You may have wanted that favorite song, but to get it you had to pay for the other fifteen tracks on the album.

The publishing industry is undergoing a similar change.  Magazines and newspapers have for decades been defined by monthly or weekly issues or daily editions.  These containers were vital to the business model, as to read the articles or see the photos etc, you had to page through numerous ads.  By defining the scope of the business by circulation, publishers were able to dimension the advertising opportunity at the aggregate audience level.  This too is undergoing dramatic changes, as publications like the Christian Science Monitor ceasing to publish a paper edition, essentially abandoning the container driven model.

The most recent shift towards objects rather than containers is the television industry.  The explosive growth of Hulu as an online broadcast TV destination has perhaps permanently changed the television industry.  Traditionally the television network represented the “container”, and the scarcity of “containers” (i.e. the big three networks) originally created extraordinary economics for those companies.  With the arrival of cable networks, thousands of new containers were created.  The Web has now succeeded in changing the game to the extent that fewer and fewer of us think in terms of television containers, and more think about the objects themselves.  Indeed, taken to its natural conclusion, even the episode itself as a content object is undergoing change, as online technologies allow users to create a clip of whatever they find and share and embed it as they desire.

Content objects: Google at risk?

The social graph is often referred to as Web 3.0 because it is such a strong example of the coming value of the semantic web.  Web 2.0 has enabled people to assemble, comment upon, rate, and share content objects in a way never before possible.  Digg was an early example, essentially sweeping aside the importance of a publication (the container) and bringing articles (the objects) front and center to stand on their own merits.  As major publishers such as the New York Times, USAToday, and the Wall Street Journal have deployed these approaches, it’s hard to imagine publishing on the web without these features.  

In many regards Google is at risk from the shift from containers to objects.  Google’s genius was in inventing a new way to identify the relative value of the billions of web page (containers) spread across the Internet.  The links between containers have provided the answer to the frustrating challenge of search relevancy.  Unfortunately, as content objects and web pages don’t have a 1:1 relationship (think about a video and an article being on the same web page, or a list of individual products for sale), web pages as containers only reveal a portion of a content objects detail (tags as a rudimentary approach).  A web page tells little to nothing about that object’s larger value across the web, something the social graph is effective at doing. 

This leads to the classic chicken and egg problem.  If web search is optimized for discovering and organizing web pages (containers) rather than individual videos (objects) as an example, then that object risks being left out of the broader social graph if it can’t be easily found.  To date Google has done relatively little to re-orient their search products around objects.  The Google search interface is still geared primarily to containers, although integration of YouTube videos and images suggest this will change in the near future.  Strangely, Google Base had great promise in this regard but seems to have been sidelined at this point.  The risk to Google is that user behavior fundamentally shifts toward the social graph as the primary means of content discovery rather than search, putting their lucrative business model at risk.

New challenges for publishers

This fundamental shift creates new burdens and opportunities for publishers, who have to rethink how their content objects will be discovered, shared and organized in a Web 3.0 world.  The biggest requirement is the comprehensive creation and organization of an objects meta-data, as the real currency in Web 3.0.  This doesn’t just mean a handful of tags, but rather a rich markup featuring tags, entities, geographic coordinates, and transcripts.  This enhanced meta-data drives discovery in a Web 3.0 world, which in turn sparks that object’s presence in the social graph, where user data can be further used to fill out an object’s markup, such as popularity, ratings, consumption metrics, and audience profile.   This hasn’t been and won’t be an easy transition.  The products and technologies to leverage this object level data into new revenue streams are only just appearing, while consumers “rejection” of the container driven business model is in full swing.

As we are in the cycle of the presidential election, there have been several high profile examples of how search is influencing and playing into the consumption formula for online video.  Two specific examples demonstrate how search either reinforces the opportunity big media has in capturing their share of the explosive growth in online video, or exposes the risk it poses.

First, as background, the pain the music industry has suffered is highly illustrative of the risks of ignoring consumer demands.  After seeing the explosion in file sharing on sites such as Napster, the music industry reached the incorrect conclusion that the majority of music consumers preferred to steal rather than pay for digital music.  As it turned out, Apple showed the way with the development of iTunes and the iPod.  History has clearly demonstrated that consumers simply wanted a way to way to buy digital music one track at a time at a reasonable price.  By orienting their product around consumer behavior, Apple became the dominant force in digital music.

Big media faces the same crossroad today that the music industry faced several years ago, and the lessons remain the same.  A successful online strategy incorporates the following:

1. Makes the content “findable” in the places where they are currently looking. In the case of online video, this is increasingly the large web search engines, but also sites such as YouTube, MySpace etc. The word “video” has become of the top terms in search strings across the web, outpacing “sex”, “religion”, and the old standby, “Britney”, according to Google’s Search Insights.

2. Enables the consumer to “lean-forward” with their content experience, and in the case of video enables the ability to search within specific video clips through tags, timestamps, rating, and commenting.

3. Creates the perception that the user can make the content their own via clipping, sharing, embedding.

4. Allows the content producer to control the context, brand, and monetization of their content assets.

5. Balances the efforts to protect content assets with efforts to make the media easy to find and consume.

Two recent examples illustrate this new reality effectively.  ABC scored a major coup by securing the first interview with Vice Presidential candidate Sarah Palin.  Yet immediately following the interview, ABC failed to garner the majority of the online consumption.  As Will Richmond points out in this post from VideoNuze, ABC made it difficult to find the video, navigate within the interview, or share the video.  In addition, ABC didn’t take an aggressive stance on removing pirated clips from YouTube and other sharing sites.

In contrast, NBC has found a cult hit with the Tina Fey impersonation of Sarah Palin on Saturday Night Live.  Not surprisingly, NBC was very aggressive about removing infringing copies from YouTube and other video sharing sites.  However, they simultaneously ensured that the clip was easily discoverable on their approved distribution sites such as NBC.com and Hulu.com, and as a result, they were able to capture the majority of consumption and therefore the advertising opportunities.  As Silicon Alley Insider describes here, the metrics tell the story:

• Charlie Gibson Interview on ABCNews.com: 600,000
• SNL Tina Fey clip on NBC.com 1,700,000

NBC’s approach suggests a template for big media online.  By combining their efforts to protect their assets with a content experience that responds to consumer demand, NBC was able to generate consumption that they could monetize, while protecting their brand as well.  As users increasingly adopt search techniques to discover and navigate online video, ensuring content can be easily discovered on the major search engines with video seo techniques, and making the videos readily searchable on the media companies’ own websites are key ingredients to generating consumption where it can be readily monetized and protected.

Tom Wilde is CEO of EveryZing, a Cambridge-based company specializing in next-generation Universal Search and video search engine optimization (video SEO). The Video Search column appears on Thursdays at Search Engine Land.

One of the most popular vertical search features on the web is image search index. What’s really remarkable, however, is how little has changed in the core technology approach to the indexing of multimedia over the last decade. When I was the head of product at FAST back in 1999, we launched the web’s biggest image search on Lycos with over 50 million images (which seemed like a lot at the time!). The service included many leading edge features including black and white and color image filtering, size filtering, and filetype filtering. The main differentiator today continues to be index size and freshness, and companies with the strongest technology in web content discovery have the most advantage. Not surprisingly, Google leads here, as their web crawling capability is far superior to anyone’s on the web. What is surprising, however, is how little has changed over the last decade. Audio and video isn’t significantly different in this regard. The majority of multimedia indexing today relies on the classic “titles and tags” approach, and Google Video is perhaps Google’s most underwhelming search product because of this limitation.

While text-based keyword search continues to dominate web navigation, one can see a future where the search input has multiple formats. I’ve seen many demos where you can provide a particular image of a mountain scene and get back remarkable similar images. The same is true of video. The problems with these approaches today are twofold. First, image and video processing is still incredibly resource intensive, although with Moore’s Law and cloud computing capabilities this problem seems only temporary. The bigger challenge is the lack of a “visual dictionary” on the web.

What is the visual dictionary?

The state of the art in multimedia processing today, specifically around images, is to use a pixel mapping process to find similar images visually. The challenge with this approach is that the pixels still don’t convey the “aboutness” of the image it is processing. Said another way, the computer knows it looks like a mountain based on the original image provided, but can’t tell the user it’s a mountain scene. Facial recognition has a similar problem. Facial recognition can find a similar face to one provided, and has improved to the point where it can actually find the same face rather than a similar face. But again, it doesn’t know the name of the person it has discovered. What’s needed is an approach I’ll call the visual dictionary. The visual dictionary would be a master meta data collection that would have tagged all of the pixel representations of an object for its “aboutness.” This would enable many exciting possibilities:

• Automatic tagging of new images: The moment an image is loaded to the web, it would have a set of “best fit” tags from the visual dictionary that would describe it.

• Finding similar: The visual dictionary would aid in the discovery of similar images online, either from a user presenting a keyword or an image as the “query.”
• Classification: Multimedia files could be automatically dropped into taxonomies and ontologies, which for the most part rely on text-based Boolean rules.

Google has recognized this problem and has put forward a human generated approach, similar to Amazon’s Mechanical Turk. With this approach, two anonymous people are paired together to “tag” an image. The pairing helps cut down on spam and maximizes tag coverage. While this approach is likely to yield high quality results, human tagged approaches have not scaled particularly well in search environments. The challenge with video is even greater. At 30 frames per second, a three minute clip creates 5400 “images!” In the case of a video news clip, the three minutes likely cover several entirely distinct concepts, people, and places, and therefore require granular tagging to really provide robust indexing.

The combination of computing power and natural language processing seem poised to create real innovation in an area that has seen little over the last decade.

Tom Wilde is the CEO of EveryZing, a Cambridge-based company specializing in next-generation Universal Search and video search engine optimization (video SEO). The Video Search column appears on Thursdays at Search Engine Land.

We’ve all seen the stats on video advertising growth. eMarketer suggests $4 billion by 2011 and Jupiter goes as high as $7 billion. As is often the case with these numbers, they are largely a black box, in that unpacking the elements that actually get us there from where we are today are unclear. The core drivers of any online ad business are really threefold: 1) the size of the audience; 2) the engagement of the audience; 3) the quality of the audience. In the case of video advertising, these three pieces are in flux. Let’s break them down:

Audience size: Video consumption has clearly gone mainstream, and by 2011 it is predicted to double to 165 million US uniques, and reaching 90% of the total audience, according to eMarketer. Those numbers resemble email and search adoption rates. In other words, video consumption becomes a ubiquitous part of the online experience. Key growth drivers: broadband penetration, mobile video, volume of video content available online.

Engagement: Growing dramatically, but across slices of inventory of varying quality in terms of advertising suitability. According to comScore, the average unique video user now consumes 70 video clips per month. Local television broadcasters only put roughly 10% of their content online today. Key growth drivers: Improved video search and navigation, higher quality content coming online.

Audience quality: As more high quality broadcast video content moves online, the audience demographic is changing to reflect a more typical cross section. Where the YouTube phenomenon has really skewed younger (if you have kids in your house, you have definitely watched the dancing German gummi bear, among other gems), sites like Hulu and the increasing amounts of video on sites like MarketWatch and the Wall Street Journal are attracting an older audience more in line with the demographics media buyers understand in the offline world. Big growth drivers: quality and quantity of broadcast content moving online.

This brings us to the core question: Is video advertising the opportunity that finally moves branded ad dollars online, or is it the next big direct response vehicle? To put this in perspective, less than 3% of television advertising dollars are spent online today, and the growth in the online video audience far outstrips the dollars being spent. Google and Overture proved that the web is an extraordinary platform for direct response, but the question remains as to how/when branded dollars move online, specifically around video.

As an illustration, at a recent conference I asked one of the big online insurance comparison sites if they had tried video. They quickly answered that they had and that it didn’t meet their hurdle rates for ROI. I was a bit surprised by this as they had no inclination to assign any brand value to the impressions they had run, the primary reason being that the dollars they were allocated to spend had only a single dimension of value—the direct response ROI. This is true today across the board. Search dollars are spent and managed through search agencies. Brand dollars are spent and managed through more traditional agencies. Both are trying video but neither have yet found that it meets their needs. This is as much a structural problem in the way the inventory is purchased as it is the efficacy of the inventory itself.

There are signs that this can be solved. The key to direct response is relevancy, meaning how relevant the ad is to what the user is looking at. The ability to better understand the content of a video or its topic through behavioral and contextual mining will continue to tighten the targetability of the inventory, and should begin to move it towards the hurdle rate the direct response folks need. The increase in quality broadcast content coming online combined with the improved discovery of that content means we should start to approach the kind of critical mass of reach that the brand buyers desire This will also meet their need to advertise against quality content that doesn’t put their brands at risk. It’s still early, so predicting which model adopts video will be one to watch.

Tom Wilde is the CEO of EveryZing and has been a player in the search industry for over a decade, including heading marketing at FAST, managing the global search & publishing business at Terra Lycos, and heading the North American business for Miva. The Video Search column appears on Thursdays at Search Engine Land.