The session was packed; with ‘how to’ questions targeted at the agency panelists. The SES Multimedia optimization track was pretty packed, which is heartening. Search engines themselves are trying to provide support for developers and content creators. But there is a lot of shooting in the dark still. As one of the panelists said, ‘Multimedia search is where search was in 1999’. Measurement is still expensive, and its pretty clear that with all the Multimedia, its still a text world for search engines -which explains a lot of the ‘transcribe!!!’ recommendations.

Makes sense. My search on ‘Comcast Bill Slowsky Ads’ didn’t return results, but ‘Bill Slowsky’ did. As the panelists said, transcribe!!

Meanwhile, Grayboxx, still in stealth mode, is shooting for the other side of the local search experience, removing the ‘reviews’ component that made sites like Yelp so popular. Will keep an eye out for these guys when they come out of beta.

I frequently use Google trends as well, just to be able to understand the International use of search, especially for global brands. Check the map for iPhone, Wii and Harry Potter. I find this particularly interesting, not only because I come from India and would love to know what people there seem to be looking for, but to understand the scope for International search marketing. We will be publishing a report on it soon, if you know of any interesting international search campaigns, mail me at ssatagopan at jupiterresearch dot com.

I can see why Hakia’s biggest claim is that of relevance—contextual relevance is a strong attribute in how the results are organized. The ‘iPhone’ results are organized into news, profile, ratings and reviews, accessories, comparison and deals, pictures, blogs and quotes –about 13 categories totally. How many of those categories will I actually need? Perhaps not all. But in this case, I am looking for reviews and that’s pretty neatly categorized already. ‘Razr’ fails to retrieve equally organized results, however.

The ‘better’ search experience is a tough nut to crack—challenged by habituation and intent. Search engines continue to try, with ‘personalization’ thrown into the mix. My recent visit to the Google homepage [I use the toolbar about 80 percent of the time] showed personalized results for video and news, based on earlier queries. Sharing a computer at home does mess this up a bit, displaying golf videos that I would usually never look for. But the search experience is getting better in terms of relevance, across most search engines. I like Hakia’s thought behind the search experience, even if not consistent across the searches I conducted; admittedly they are still in Beta. I do agree, the ‘…’ on the Google search results can be slightly annoying, compared to fully displayed relevant text on the Hakia results. As Hakia progresses to a full-out launch, I will keep my eyes open to note their partnerships more than their destination site rollouts. There is always scope to cater to different perceived needs of searchers, but innovation will also lie equally in acquiring that traffic.

Speaking of relevancy, go here for a relevant Jupiter take on Search and Navigation.

Spidering and parsing content seems to be lesser of a problem among the people search sites, but finding unique and valuable spins to the content is. Zoominfo’s take on people search is more for professionals, expanding into company search and job search. Spock’s take is focused on the ‘community’, allowing the members to tag results based on relevancy and additional factors. I am not quite sure I would want my profile to be tagged by a community. And as a people search user, I would probably be worried first about what content exists about me, as the Spock team noted in the initial usage of their service.

People Search can cater to more than the directory-information seekers. Research, networking opportunities exist—as proven by LinkedIn, for example. Should I visit different sites depending on my motivation to search on people? Perhaps not. [Zoominfo does provide better results for my co-workers than Steve Jobs]. I think two things are certainly clear, however.
1) People Search is fragmented, even in its initial stages. And vendors are experimenting a lot, with multimedia, social networking and tagging.
2) There is a lot more of me that is findable than ever before. [Pipl.com also pulls out my Amazon customer profile]. Whether that’s a good thing or not, I am probably going to be extra careful about profiles created in any case.

Last week, Kosmix briefed Belis Aksoy, our auto analyst and me on their recent partnership with Autobytel. Earlier this year, Autobytel announced its launch of myride.com, an automotive vertical search site [The site still seems to be under construction]. They seem to have eventually partnered with Kosmix to power their auto search-thus tapping into Kosmix’s technology that has aggregated content including reviews and guides. Among classifieds, jobs and real estate have had some buzz on vertical search, but the auto category has revolved primarily around third-party sites [Our auto analyst covers trends here]

Focusing on Kosmix a bit, the vertical topics they aggregate content for are health, auto, travel, finance, politics and oddly enough, video games. What is particularly interesting though, is the focus on partnerships. Vertical search sites have long since stopped trying to compete with Google, focusing instead on trying to be a niche destination site, while some prioritize on powering publisher’s vertical search features. A recent briefing with Boorah.com [a restaurant review site] reiterates this trend among the tail categories as well. The site works fairly well as a destination site, but the strategy is strongly on powering newspaper sites with restaurant reviews [go here for Jupiter’s take on the online newspapers]. Boorah powers the review engine for Palo Alto Weekly and a few other bay area focused newspaper sites. I am pretty convinced that trying to build a destination site for restaurant reviews that competes with the Yelps and Zagats is difficult, although I do quite like Boorah's approach on their site. But partnerships will open doors beyond local newspaper sites.

Jupiter is looking at vertical search trends as well—watch out for reports on the evolution of vertical search sites and health search in the next few weeks.

· Universal Search-In a significant change to their search interface, Google will include relevant formats such as news, video, local, book excerpts in their results. What I found interesting was the plan to blend, rather than section the results page to accommodate these results [Click here for Jupiter’s take on Search UI and Navigation]. My own search on ‘Spiderman’ returned image and web results, with tabs for what Google calls ‘contextual navigation’ –in this case, images, groups and video. As Marissa Mayer, VP of Products admitted, this is not the most perfect layout for results, they are still testing it out. There is an element of surprise, I would think, in looking at video results as the 4th,, local results as the 5th and images as the sixth result. However, it is certainly a way to change the way searchers think about results. Google justifiably has an awful lot of faith in its relevancy to be able to deliver a seamless search experience.

· Underused Features-are being given a push. Book Search, Google Groups are all a part of the integrated rollout.

· Forthcoming features-Google plans on Keyword mapping , by drilling a bit deeper than the current ‘Did you mean…’ spell check feature. Udi Manber [VP of Engineering] gave an example of a query ‘distance from Zurich Switzerland to Lake Como Italy’ will trigger a second search ‘train Milan Italy Zurich Switzerland’, the results of which will appear below the original results. I found this particularly interesting, the attempt at digging a bit deeper into searcher intent. No doubt, the web history will play an important factor in this as well. [Go here for an interesting analysis of search sites, including those that focus on user intent]. Language translation on the fly is in the works too. [Check Zia’s post on international sites here]

Bottomline for marketers? If the Google tests with interwoven elements proves successful, natural optimization strategies, especially for multimedia content should be geared and ready. Additionally, your optimized text content will now compete with videos, news and images for relevant searches.

For marketers, the current option to be included in search results is via feeds. Like.com seem to have the regular suspects such as Amazon.com, Shop.com, etc in the vendor kitty already. Spidering the web would be fantastic, and possibly in the future plans. But spidering for visual content has it challenges-I suspect the text-based search will play a stronger role than it does now.

The road ahead includes rapid category addition, given the proximity to holiday shopping season, and other interface and accessibility developments poised for the next few months. The interface takes a wee bit of getting used to, especially that the ‘likeness search’ button that doesn’t really stand out noticeably enough [seems more like a font-size and color-contrast problem to me]. Navigation could be smoother, but with the small number of categories, it hasn’t inconvenienced me much as yet.

I am currently writing a report on optimizing content for multimedia search engines. For agencies working with retail clients, it seems to come as no surprise to find that low volume traffic seems to be sourced even from Google Images Search. While Like.com is still some time away from proving impressive traffic numbers, for marketers who own the multimedia content such as images, this should be a good time to start thinking about optimization tactics for search engines.