Will Universal Search Mean Universal Domination?

I attended Google’s Searchology announcement session for the press in Mountain View yesterday. Among the many announcements were universal search and navigation, upcoming query enhancements and a new Google Experimental tool for opting into Google’s many interface and feature tests.

With its universal search, Google is now going to offer results in web search that integrate results from all of their vertical search properties, such as image, video, maps, local, blogs, etc. The video results will include results from third party video sites like Metacafe, and not just from Google’s own YouTube and Google Video.

While some of this has been going on for a while, Google is going to be integrating all of these various component search engines so that the most relevant item shows up first. So if the most relevant item is a video, it will be the first result in the universal search results. If the most relevant item is a map, then it will be the first result, and so forth.

This universal search approach is different than existing techniques used by search engines, which clump related results together. In the past, Google would show all the vertical search results in a block somewhere on the page, perhaps even in the middle of the web results. But even so, existing systems used by search engines for doing this are arbitrary. They don’t determine the placement of the vertical results based on relevance, they simply have a place where they put it.

And that’s the basis of this innovation. With universal search, Google is potentially taking a significant leap in relevance over other search engines. But that relevance does not come easily, as Udi Manber, Google’s VP of engineering, and Marissa Mayer, Google’s VP of search products and user experience, both said at the event.

The Universal Ranking Dilemma

Once Google made the decision to fully integrate the results, the first thing that needed to be developed was a relevance scoring system that would work on the same numerical scale across all of their properties. The image search engine, for example, has a relevance scoring system of its own. So do their video search properties. But they don’t naturally have the same scale. There was no pre-existing way to tell if a particular high ranking image was more important than a web result, or a video search result.

The key thing that Google needed to do was to normalize these results, putting them all on a common scale to rank them accurately in the universal search algorithm. Further, they had to do this in a way that did not require them to suddenly send hundreds of millions of daily search queries into the smaller vertical engines. They simply did not want to have to add that many servers to each of their search properties. It would have been prohibitively expensive.

To avoid this huge expense, Google’s engineers had to find a way to extract the normalized relevance data from each of their various properties, so they could query that data without having to do a full query of the image search engine, or other vertical engine. In other words, they needed a way to get the answer without making the secondary engines take on the full burden shouldered by the Web search engine.

But once they succeeded in normalizing and extracting their relevance scoring systems, the rest was relatively easy. Some of the remaining challenges in implementing universal search have been outlined in a post on the Google Blog, Behind the scenes with universal search. They include ranking and user interface issues.

Universal Search: Increasing Relevance

With this normalized data in hand, Google can receive a user query, poll all the various sources of data they have, look at the results, and rank order them on the fly. What’s the reason for all of this investment? Relevance. Being able to do this on the fly allows them to deliver highly relevant queries through a single search box, regardless of whether or not the best result is a map, local results, images, videos, blogs, or whatever. In fact, you may see some mix of these through the universal search results.

So here is the first challenge that has emerged from all of this. Google already has a dominating market share in the search space. This presumably has emerged because of higher relevance, or higher perceived relevance. With this step, Google has made a significant move towards an even higher level of relevance (or perceived relevance).

Further, I have long believed that Google is a data company, not a search company, with highly specialized expertise in data processing and the setup and management of massive server infrastructures to deploy that data globally.

Outdistancing the Competition?

In my opinion, this innovation in universal search will be hard for others to follow, as it involves some major infrastructure changes, and Google got to this day only after years of investment (Marissa Mayer originally proposed this idea back in 2001).

The rollout of all of the new products and features that Google announced yesterday will take place over a relatively short period of time, so it should be visible to all soon. In fact, many facets of the universal search program are already live. Equally important, Google made it clear that this is simply the first component of a larger initiative, and that this is much more to come. It should be fun to see what’s coming next!

Eric Enge is the president of Stone Temple Consulting, an SEO consultancy outside of Boston. Eric is also co-founder of Moving Traffic Inc., the publisher of City Town Info and Custom Search Guide.

Search Headlines

We report the top search marketing news daily at the Search Engine Watch Blog. You’ll find more news from around the Web below.

Related reading

Graphic showing a microphone next to the Google logo, against a background of brown fading to black at the edges.

Simple Share Buttons