Mind reading with Google Chrome


All web browsers come with an address bar (1) and a search box (2). For most of them not named IE, the search box is by default Google’s. Until last week, however, no browser came with a single box for both navigation and search. Google Chrome changes that, and a lot more.

Chrome is a multi-platform, online/offline web browser with a sparse UI that claims to be faster, more secure and stable than IE, Firefox or Safari. But because it’s offered under the liberal BSD open source license, other bowsers can leverage Chrome bits (like the fast V8 JavaScript and Skia graphics engines or memory and security sandboxing) to gain parity. Google says nothing would please them more.

One box too many

There’s one Chrome feature, however, that will be very difficult for Mozilla, Apple, or even Microsoft to effectively emulate: Omnibox, the progeny of the browser address bar and the search engine box.


Traditionally, the address bar is for entering DNS information as a web address to locate a specific URI. Since long, dynamically generated URLs are not made for human consumption, users have been observed to ignore the address bar and directly enter their destinations into the search box, even in the form of a URL. This evolved practice redounds to a long list of benefits to the user, and in the process, has been a prime accelerator of Google’s $160 billion market cap. So why not merge them?


That’s a leap too high even for Mozilla and Apple that have been receiving millions in compensation from Google to place the company’s search box as default in their browsers. After all, the ubiquitous address bar has been a primary component of any browser since the web browser was invented. Wouldn’t it be great, however, if there was a single box that can handle navigation and search simultaneously…and it came from Google! Apparently, Google thought so.

Google as information broker

Google’s business model is one of information brokerage, just like an investment bank or a hedge fund. It monetizes web users’ dataflow through the use of its market-dominating search vehicle which provides advertisers with relevance. Google tracks users’ navigational history and search intent, and maps it to web’s linking/preference hierarchy. Advertisers have been more than willing to pay for that information to the tune of $5.3 billion last quarter alone.


So the mapping process whereby Google can use predictive analysis to determine what a given user might to do next is supremely important. It aids the user in avoiding unnecessary steps to get to a web destination and advertisers in getting the user to arrive at a purchasing decision, and of course Google in getting to charge for it.

One box to rule them all

For all other browsers, the entire process generally starts with a user entering a URL into the address bar or keyword(s) into the search box. From Google’s perspective having two boxes is not only inefficient but also quite limiting, because while Google can track its own search box, anything entered into the address bar is lost as an opportunity for Google to directly capture the user’s intent and manipulate it.


For example, if the user enters “pizza” into the address bar (3) and there’s no nosy ISPs redirection involved, he may be taken directly to “http://pizza.com” (4). It’s a simple matter of a DNS look-up and there can only be a single corresponding web address. This being the primary function of a web browser, there’s no money to be made here (5).

Context is intention?

But if the same “pizza” is entered into Chrome’s Omnibox, Google begins a search after each keystroke, even before the word is completed and the user hits return/enter (6). Things get very interesting at that point, because unlike the static, 1:1 DNS resolution provided by the address bar, Omnibox (6) has context. Google can discern intention and thus predict destination. After all, “pizza” can mean different things in different contexts and Google knows it better than anybody else.


On a cooking site, “pizza” likely means pizza recipes; on YouTube perhaps videos on how to make pizza; at CitySearch.com not only pizza restaurants but, because of the user’s location already known to Google, a list of pizza restaurants in a specific city; and when Omnibox comes to mobile devices with GPS, pizza restaurants directly around the block plotted on Google Maps (8, 9).

Omnibox promises to take Google Suggests a step further by elevating it from interactive keyword completion for search to destination selection for navigation, all in a single box.

Apple and Mozilla have browsers and perhaps the ability to best Chrome on all technical challenges, but they lack Google’s search prowess. Microsoft had a shot at this. Paid-for keyword navigation was pioneered by RealNames in 1998 and was offered through Microsoft’s IE to navigate MSN pages. While RealNames had significant growth, it was dropped and forced into Chapter 7 by Microsoft perhaps because it threatened Microsoft’s search business. AOL, Barefruit, Paxfire, Golog and others have also been providing variations on keyword and mis-spelled-URL originated direct navigation. But none has the reach and scope of Google. Chrome’s Omnibox is free: free to users and free to site owners. Therein lies Google’s impending dilemma.

Omnibox giveth…and taketh away?


Omnibox gives Google the ability to contextualize intent, make suggestions and directly take users to appropriate destinations (9). Imagine the utility of this in mobile devices where data entry is cumbersome and application switching is technically and cognitively costly. This is not unlike a more refined version of the Google Mobile App for the iPhone, not as a separate app, but as the sole navigation box of the device’s main web browser:


The user could practically live 24×7 in Chrome’s Omnibox. A new breed of geo-tagging, GPS-aware, social-networking, map-driven, collaborative-filtering mobile applications may find their unique advantages subsumed by Omnibox’s ability to detect and (re)direct users’ intent. Users may no longer need a slew of dedicated, often single-purpose applications when they can remain in the main web browser, enter a keyword or two into Omnibox and let Google arbitrate their existing navigation history, personal preferences, context, and compare it all to web’s link hierarchy to refine choices in real-time without any effort on their part. There’s no need to download or purchase multiple apps and painstakingly enter personal metadata. An always-available Omnibox could be a nightmare for any start-up that fancies intention brokerage at a price…an awkward position for many Android “partners.”

$64 billion questions

Google’s ability to “mind read” users’ intentions via Omnibox and funnel traffic automagically to likely destinations is a phenomenally attractive proposition to both users and advertisers. However, the promise begets its own questions:

  1. Would Omnibox redirecting impair Google’s current keyword-based advertising business?
  2. Why would advertisers continue to buy keywords if they can get traffic through the Omnibox?
  3. Would Google then open up Omnibox “suggestions” for bidding?
  4. If such suggestions are auctioned, would users trust them as much?
  5. If they don’t, would Google leave such a huge opportunity on the table?
  6. Could Google find a lucrative balance between these two business models?

What’s your prognosis?

11 thoughts on “Mind reading with Google Chrome

  1. Nicely written article, with even better illustrations, Kontra.

    I’ve come to know your desire for simple and elegant interfaces, as well as good business cases. Google’s Omnibox uses just the one user entry field to provide not just context but also capture all those revenues lost to entries made in the address bar. Capturing just those lost revenues may pay for the Chrome development, regardless of future profits from application integration.

  2. chirax: “Awesome Bar in FF does the same thing right?”

    No, because Mozilla just doesn’t have the assets I mentioned above in the backend to automatically arrive at meaningful and contextual “suggestions” Google/Omnibox can provide. Omnibox is not just about the UI, it’s about what’s behind.

    Now, FF can in fact hook into the Google empire but that’s a business decision Google has to make to allow it.

  3. Arun: “this feature is not new”

    Omnibox is a single entry point that combines search and navigation. Surely, various aspects of that proposition have been present in different capacities in a number of browsers.

    What Google can bring to the table uniquely is that seamless blending of search and navigation offered as “suggestions.” These suggestions can be the analytic end-result of local and cloud context, preferences, search and navigation history mapped against PageRank. That combination is hard to do and is not something Opera, Firefox, Safari, etc can provide.

    That is the promise of Omnibox. As I outlined at the end of the article above, there are countervailing considerations. But at the end of the day it’s Google ability to focus on it and pull it off, or not.

  4. “There’s one Chrome feature, however, that will be very difficult for Mozilla, Apple, or even Microsoft to effectively emulate: Omnibox, the progeny of the browser address bar and the search engine box.”

    I get live/google(depends on your default search engine) search results if I put in “some text” in IE address bar and press go. Same is with firefox and opera. AFAIK this feature is not new. However “google suggests” doesn’t work. But these browsers provide features like suggesting browser history/bookmarks etc. from the address bar.
    Agree with Thilo in this. Seems marketing makes a real difference!

  5. Actually the Omnibox has been always present in Epiphany GNOME default browser: http://en.wikipedia.org/wiki/Image:Epiphany_on_Ubuntu.png . I have always prefered this browser over Firefox on Linux – but funnily everybody seeemed to believe that Epiphany was indeed inferior to all other browsers. Epiphany uses the Gecko engine but will also use Webkit soon – or better it already can be compiled using that engine and the next version will use it by default. I would have wished that the Epiphany programmers would be better in marketing their own accomplishments (like Omnibox since 2002).

  6. Berend: “Google has complete control over the Omnibox and won’t hurt it’s own interest”

    A question I have is if the algorithms to determine what ads to serve on a page are substantially different for Google than to determine what suggestions to make in Omnibox. It looks to me like Omnibox could access more contextual datapoints to be more precise and more aligned with the intentions of the user, thus being more valuable.

  7. Great article! As always, there is no free lunch. How come, Google would take these huge development costs and then invite anyone else to join the party? This is a nice and interesting answer. Nice clear illustrations as well.

    As to your questions:

    1. Of course it won’t. Google has complete control over the Omnibox and won’t hurt it’s own interest.

    2. Today, with sponsored keywords advertisers have to compete against Google’s own results as well. When you have a straightforward question I think you are not so likely to click a sponsored link anyway. When you have a more open or complicated search you might prefer to view the results in a larger webpage instead of the small Omnibox, so it is not such a strange option to click “search for…” in the Omnibox.

    3. Google no doubt will experiment and see what is working and what is not.

    4. Advertiser probably are quite opportunistic. They try it out, measure the results and when those results are positive trust will follow.

    5. That’s a contradiction. When advertisers mistrust it, that’s because it proves to be not a huge opportunity.

    6. Absolutely. As said before, I think the greater sponsored link opportunities lies in those searches that are not too much focussed and that are the searches where a user likes to see them presented on a larger web page. When you already know the name of the company you are looking for and type it in the Omnibox, well, pretty much chance you will ignore the sponsored link.

  8. flo: “you know that you can switch the searchengine to yahoo or msn or so too, right?”

    User who explicitly get Chrome from Google, a company that dominates 3/4 of the search business just won’t switch to Mozilla which doesn’t even have a search business or to Microsoft which hasn’t made a dent against Google so far. No other player has the depth and breath of Google’s search and related properties, expertise and ability to tie them together.

  9. Mmmh, you know that you can switch the searchengine to yahoo or msn or so too, right? Not sure how well that actually works with the whole suggestions and stuff, but that means I don’t really see how firefox and ie should have a problem replicating the technology. They could use the exact same technique google uses, or even another search engine. Also, I dont think it would be wise for google to give the ability to get preferential treatment on the suggestion list. Contextual ads are one thing, but paid for suggestions would be like paying for getting higher up in google’s index. (which they don’t do I think)

Comments are closed.