Web 2.0 - Search Engine Execs Chat
I guess "Search is a Platform. Where is it Going?" had to be the session that I looked forward to the most. Apart from the absence of Google, the panel was made up of some of the top people in the search industry:

Steve Berkowitz - Ask Jeeves
Udi Manber - Amazon's A9
Louis Monier - eBay
Christopher Payne - MSN
Jeff Weiner - Yahoo


Photo courtesy of Brad Hill (more photos and audio on his blog)

I'll do my best to summarize the comments made.

John Battelle provided the initial questions and asked the panel about the recent trends towards personalization.

Manber explained the current phase of the search engines. "If musical instruments had been invented 10 years ago, we would all be playing 1 string instruments right now" indicating that search is still very much in its infancy and has a long way to go. "Hi honey, you remember me?" are the words he used to explain the advent of personalization. Your search engine can now remember that search you did 4 months ago.

Battelle asked if PPC is all that we have to look forward to, or is this just the start of monetizing search.

Berkowitz offered that the more we focus on the non-monetized search the more people will use search engines - the advertsiers will naturally come to the search engines that have users. Weiner added that more people are searching intensely. Click-thru rates will go up, as matching technology improves and CPC's will increase as more people bid. Weiner also suggested that a lot more can be done with local search ads, offering that only 1.4% of the $100 billion local ad industry is currently being spent online.

Manber supported the PPC ad format saying that they add value to the search results and do not annoy users like a pop-up ad does.

Moving to issues faced by the search engines, Berkowitz commented on the challenge search engines face from search spam. Saying that "spam is one of the biggest things we face with people trying to fool the search engines."

Battelle then asked Monier if eBay would suffer if search engines became so good that consumers no longer needed to find obscure items at the auction site. Of course, Monier was not worried, citing that eBay offered a lot of things that make it successful including a community, trading platform and payment options.

Battelle ask Manber why Amazon decided to build a search engine. His reply was short, but to the point. Amazon believes that "search is important" and indentified the need for improvement within the current space.

John Battelle then turned to one of my favorite topics, asking the panel members to look at the developments they see coming in the next 5 years.

Payne believed that desktop search was "super important". Files on a PC are more difficult to find than pages on the web. "Desktop will be the first foray" into search for Microsoft. The next area of development will be personalization, re-ordering the results. "If you know the individual and there is a trust relationship, you should be able to provide accurate search results". Payne also said that "technology also needs to be able to correct when there is a mistake", referring to the ability to undo personalization filters if things go wrong.

Manber said that personalization should be more intuitive. If you ask "where is the beach?" the search engine should give you a personalized answer.

Berkowitz added that search needs to figure out how to get to the 85% of data currently not reachable, offering information behind a firewall as an example.

Weiner said that personalization, then desktop search would be priorities. Understanding the intention of the user and then able them to complete their task quickly. He suggested that personalization comes in two forms; explicit and implicit personalization. Most current types of personalization are explicit. He said we can learn a lot from Tivo and the options it provides for personalizing your recording habits. Weiner also said that the user-interface would also be important going forward.

Monier appeared frustrated when he said that he had been waiting for personalization for 10 years! He said one quick fix would be that, if an entered search had three different meanings, simply ask the searcher which one they meant.

I then had a chance to ask the panel about relevancy in search. I suggested that the past 3-4 years had seen inbound links as the main driver of relevancy. I pointed out that this system had been proven unreliable over the past few months. What did they think would be the next variable measured to determine relevancy.

Weiner said that they could potentially measure how long someone stayed on the site after they had clicked on the search result. This might tell them if the searcher had found the most relevant site for their search request (timeout: didn't Yahoo once include popularity in their algo?). Weiner said that with personalization, they could also start using data they collect to determine relevancy. Weiner did say that "linking relevancy is flawed as the sites that sit at the top of the search results, tend to get linked-to the most and end up staying at the top." I wish Google had been on the panel to hear that comment!

Berkowitz replied that Teoma has been looking for the authorities on the web which could be combined with personalization to determine relevancy. He said that time spent on the site may not be a good test of relevancy.

Lastly, the panel were asked whether it was possible to re-educate searchers to use a more in-depth interface or if we should stick to a simple UI?

Berkowitz said we need to try and train searchers to give more information. "You don’t go into Best Buy and simply say 'dvd player' to them." He said the same should be true of search requests. Manber added "training searchers is not easy, but that doesn’t mean we shouldn’t do it."

So that was about all. No major announcements, no real shocks. However, it is clear that search as we know it will change over the coming years. All of the panel seemed committed to personalization with many favoring a desktop solution too!
Andy Beal | permalink | +marktd |