Keynote
Generation Next: Search In The Coming Decade - In this panel, luminaries from the major search engines discuss where they see search going as the decade unfolds. Moderators: Chris Sherman, Executive Editor, Search Engine Land Gord Hotchkiss, President and CEO, Enquiro
Speakers:
Brad Goldberg, General Manager, Search Business Group, Microsoft
Dr. Larry Heck, Vice President of Search and Advertising Sciences, Yahoo!
Doug Leeds, Senior Vice President of Product Management, Ask.com
Peter Norvig, Director of Research, Google Inc.
Chris asks the panelists: "Will Search as we know it survive?" Brad: Someone told me that search is so bad today. You have to click, go back, click, go back, etc. They asked me if there's a way to improve. It highlights a phase where consumers are getting more sophisticated around their expectations. The users refine their searches very often, but satisfaction is high. I think search will absolutely survive but as verticalization increases, search will do a better job and become more useful and become more of a utility that people use for a broader set of tasks. Peter: I think that search will become more natural interaction with data. You'll have more devices available to you and it will be more of a dialog back and forth. People will start to know things Larry: There's search and what's going on in the background: the basic matching of approaches - we're not going to throw it away; we'll build on it. That will evolve. It will become more pervasive and relevant.
Gord: We talk about the evolution of search. What does that do for the grey matter up here and the data that's down there? How do we interact with the huge store of information as humans? Larry: The way that we access the information will be different than on a PC. People will post their queries differently. Initially, the systems will not fare well. Peter: I think it will be helpful. The average query is going to be longer. On the other hand, it's going to be harder to get the results back and we all take advantage of the fact that we have something that allows us to display 10 results. People like go usually scroll down to the 3rd or 4th result. But if we have a tiny little cell phone with a 1 inch display, getting results with 10 SERP entries is increasingly frustrating.
Chris: What about voice activated queries? It's going to require a lot of changing in search marketing. Brad: The user experience will get better with time. The biggest change will really be for advertisers. Peter: People need to get more sophisticated to understand the users' intentions. On a tiny little display, where's the ad going to be? Later on, if you are driving past the store and they remember that you had an interest in a product there before, search might evolve where the product purchase is suggested it to you.
Gord: Are you using relevancy as a proxy for usefulness? What about in the ad environment? Will you merge them? Brad: There's an element of trust with the user. The user will want to know what comes back from the algorithm vs. the sponsored listing because they are becoming more sophisticated. We want purity within search. Search becomes more useful but I think we'll need to maintain separation of purity between ads and organic results. Larry: As you go through a session, it becomes clear to the search engine about the users' behaviors, but we need to make a distinction between ads and organic results.
Chris: [I have no idea what he asked. Susan Esparza from Bruce Clay Inc. sneezed and I didn't hear the question.] Peter: It's got to be separate and the user will be in more control.
Gord: Brad, you guys talk about implicit vs. explicit queries but that kind of went away. However, are you going to look at the context of the text and provide suggestions based on that? Brad: I think we will see that outside the operating system and more in the context of the application. You'll start to see search playing more and more of a role but possibly in the background. We've been seeing hot to target people at work. If people are in an intranet, we might want to treat it differently than on the extranet.
chris: Search is evolving. Do you see that coming where there's a research assistant and realization of semantic web? Brad: I'm not bullish on that. It's a great idea to be attractive to people with an element of personalized search. It sounds great but I'd be surprised to see it in a practical and useful way. Peter: We like to think of different timescales when helping the user. In terms of personalization, I think there's some room for that, but the whole point of search is to find stuff that you don't have [already]. Peter: The search engine tool is there to make it more efficient. Search in terms of coverage and memory is better than humans. Maybe for small tasks, I can see some changes there, but the users have to be intertwined with the process. With respect to personalization, a lot of times we talk about customization. It has to be looking at you as individuals over a long period of time. In the moment, the task you're trying to complete may not be identified correctly. You may be able to compare that with the searches of other people. It will feel very personalized to you.
Gord: We identified a lot of signals and patterns like geotargeting. Given these signals that show promise, how smart can search get in the next month or two? Peter: We're just getting started. We can just aim to do better at understanding the content. Brad: I wonder how much what will evolve once you enter your first query. People see more and more experiences being accessible from directly within search or search-driven. How much is going to be the technology evolving versus the user being more sophisticated? The number of users who click on some tabs for advanced searching is incredibly small. The intent will become clear to us and we'll be able to refine it. Larry: With respect to how intelligent the search engines become, the better question is about the effectiveness of the partnership between user and machine. How much more effective are we going to be as people?
Gord: The fact is that we have a similar interface and advanced search and the bar hasn't risen much with how humans interact with search. Are we expecting too much for them to use these tools available to them? Brad: Today, search is really big horizontally. 70% of the top 2,000 keywords are navigational. Most people use this as a starting point to find what they want on the web. Pageviews really exclusively lie within vertical content. I think the change needs to be really more vertical. Once that change occurs, you'll see different user models for how people are better able to access and get into that content.
Chris: Speaking of verticals, local search is one of the most promising verticals and it's so horrible. What is it going to take to improve that? How close are we to driving by and suggesting products (as Peter suggested earlier)? Peter: We need to know about the users and what's in the locations. We are starting to integrate Yellow Pages and the like. Larry: It's a fundamental shift in the technoloy's part over time. The better the content gets -- which is happening now -- we will start to see this.
Gord: I want to talk about advertising on the SERPs. We're going from text based to a richer interface. How do you see that playing out in the next few years and do ads need to be more engaging or louder to be noticed? Peter: The whole web is getting louder. Things are working more and we're seeing different types of media. Larry: I think that the right setting, right framework, right incentives, and right measurements will help us provide the most value to the users that give CTRs that convert. Search engines need to work hard to push advertisers to change that. They certainly have incentive to keep up with creatives and quality. The economics will help as well. Brad: Search has grown to be a successful marketing vehicle because it's quantifiable. The short answer: we'll see a lot of experimentation of different types of ad formats. The ingredients are there for people to experiment freely and we'll need to adapt. Gord: Are we looking at a Darwinian approach here then? Brad: I think that's fair.
Chris: Search has gotten better over the past decade but you can argue that advertising is irrelevant with the better algorithmic search. How do you reconcile this? Larry: The advertisers are providing additional information which helps. There are a lot of economic reasons to keep them there. Peter: I think everything has to provide something of value. As algorithmic results improve, you'll still see people buying ads. Brad: We talked about the difference between research mode and transaction mode. In research mode, they want product information, ratings/reviews, but other people in transaction mode want more information related to the purchase. Search will become more sophisticated and we'll see separation that is intuitive to users. If you're in research mode, you don't want an ad that says "Buy now!" You want to provide information about the product.
Gord: I want to talk about personlization and Google backed off a little about talking about it. I understand why personlization is hard but if we look at incremental improvement, how valid does it have to be to start being a very valid signal [for ads]? Peter: We need to know the types of things people are interested in. A lot of search is navigational as Brad mentioned and we'll take all this into consideration and separating that out. We can do personalization for specific ads and maybe not for other ads. Larry: We want to make it feel personalized, but we don't have to use that signal so strongly. If we can identify what the person is trying to accomplish and then leverage it, the search engine can feel very personalized. Peter: A good personalization example is horoscopes.
Chris: With personalization, it seems in a way that search marketers have less and less influence on the algorithmic side. Do you see that influence weigh even more for traditional SEO? Peter: I think it's just a bigger opportunity for them. Brad: You can see some examples of companies using viral marketing around video of ways to then take things that users may have never thought possible. From a marketing perspective, people are looking at SEO and other content too. They are new vehicles to get in the mainstream discussion for consumers. It's a question of who can exploit it the right way so that it doesn't feel like advertising. Chris: It's really more of the great thinking and not so SEO. Brad: I think we've only scratched the surface of what search can do as a marketing vehicle. There's so much more surface area and a lot of opportunity for innovation. It will be interesting to see what emerges. Larry: Opportunity for search marketers increases more and more. Search is different in advertisements in text. As SERPs become richer, it's going to move back a little more to the traditional homepage. Gord; This opens a lot of opportunities for search marketers but it may be different thinking than what you're used to.
Chris: Over the next decade, non-English content may dwarf English content. Are you going to work with that at all in processing that for English? Peter: We focused on the algorithmic side and process results across languages. Brad: Look at the scenario where someone is searching. Even for the next few years, they are probably mostly academic or research-oriented or technical material that someone is looking to access in a different language. It's probably not commercialized and advertisers may not work with that. It really depends on intent. But advertisers may also think about who they are advertising too - that opens up additional ecommerce questions.
Gord: What research area outside of search is particularly interesting to you when you look at the Internet from a human standpoint and for change? What really cool thing is being done? Peter: The mobile space is important - having information everywhere. We need to open things up in terms of bandwidth and platforms but it has to happen. Larry: One is openness in terms of leveraging all the expert pieces out there in specific areas and opening up the platform to aggregate the information and bring the best to the forefront. The other thing is intent of user - they can do it in a little box with a few words. There's an imbalance between the author and the technology. It will be interesting to understand semantics of intent. On the user side, we'll be facilitating more expressive ways of people stating what task they're trying to complete and getting better and better at it. Brad: In the car, what will happen if you look at the set of technolgy - voice, mobile, simple purpose, etc - to meet a specific need? We don't want to overcomplicate the scenario especially when people are driving. We've touched upon some themes: local improvements, mobile improvements - that's untapped right now.
Chris: What one feature would you kill to have working in 5 years? [Silence] Brad: It's a tricky question. We're wondering what the right answer is... Brad: For me, it would be how to enable people to search in a more verticalized way. That's really going to open up, whether on the homepage or the SERPs. Peter: The connection between the user and the content. Larry: Access to search everywhere. I want the same kind of access from my car than from my computer. We want the experience to be more relevant.
Gord: You were talking about convergence and more apps that will be pulled in to understand our choices in entertainment, etc. How do you see that working in the entertainment industry? Brad: Search engines need to implement it so that acknowledges that environment. Live thumbnails and videos can be developed - scan video, present user with a very condensed version that they can hover over it and get the most relevant pieces to get the essence of the video at least on the SERPs. But then when you click on it and people will get the original source with the full information. You'll see innovation there. Peter: I think that entertainment is a big opportunity.
