The Evolution of Search

One of the data services I subscribe to highlighted a specific trend their survey research uncovered. They called this trend “the slow death of search”. To illustrate, they graphed the following data.

Screen Shot 2014-12-17 at 10.39.29 AM

While the Y-axis is capped at 95% rather than 100%, and the bottom at 85% but the point remains. Search is happening less on DT/NBs. Given the global trend away from PCs to mobile devices, I assume that this time next year that chart will show the bottom to keep trending downward. The question was regarding websites visited on a PC. The specific question listed a range of websites — one of them was a search engine. Thus, the resulting graphed data is how many respondents (30,000+ globally) answered that they visit a search engine at least once per month from a PC/laptop. The question asked dated back over many years so the graph shows the answer’s evolution over time. What we see is a clear trend away from desktop search as a frequent monthly task.

Clearly, search is still important. No one will say they never search the web but the interaction model for search has changed and is still changing. The idea of pulling up Google and doing a search is the fundamental engagement model in question. The graphed data likely highlights a shift from desktop/laptop search to devices like mobile and tablets. I have access to similar data sets that highlight a similar question regarding mobile search and more than 65% of respondents said they perform a search of some kind from their mobile device every week. This point shouldn’t be surprising.

Ultimately, when we understand the constructs of search, it is helpful to view it in the lens of decision making. Helping me make a decision looked like a blank text box in the past but in the future, and thanks to mobile devices enabling device and contextual awareness and intelligence, the paradigm is poised to change.

What sits at the crux of this conversation comes back to a shift in interaction models. This could be something like Google Now, or elements of Siri, which create more predictive analytics to bring things to my attention that are relevant before I have to search for them. Smarter beacons, integrated with smarter local hardware, could be another dynamic that helps shift the search interaction model. However it manifests itself, it is clear the search interaction model is changing and evolving and will likely look very different than a blank text box in five years.

As we look out at who is in a good position to capitalize on these it is those both at the platform level, but also those who are gaining data to build a comprehensive anticipation engine. Ultimately I feel it is companies like Apple, Microsoft, Google, Baidu, Tencent, even perhaps Amazon and Yahoo.

Published by

Ben Bajarin

Ben Bajarin is a Principal Analyst and the head of primary research at Creative Strategies, Inc - An industry analysis, market intelligence and research firm located in Silicon Valley. His primary focus is consumer technology and market trend research and he is responsible for studying over 30 countries. Full Bio

13 thoughts on “The Evolution of Search”

  1. This is a very important topic and I’m grateful that you bring this up.

    I think there are two sides that we need to look at.

    The first is the absolute number of searches being performed. This is going down as you describe in the article, driven by changes in the interaction model.

    The second is the intent of the searches being performed. This is what I am particularly interested in. For example, how many people come to the Techpinions website via a Google search on the term “techpinions”. That is, how many people are simply too lazy to write the full “http://techpinions.com” in the location bar, and find themselves doing a Google search, despite knowing exactly where they wanted to go. Similarly, how many people just want to go to Yahoo! Finance, but instead of using a bookmark or typing in “finance.yahoo.com”, they simply type “yahoo finance”, which will take them to Google. I’m quite sure that the number is very significant.

    My point is that the search intent is frequently very, very simple. Google’s sophisticated search algorithms are overshooting customer needs for a very significant portion of searches. This is a precarious situation from a Disruption Theory standpoint.

    I’m sure that Google fully understands this, and this is why the “Knowledge Graph” exists. They know that what users want is usually a Wikipedia article, a Yelp review or a Rotten Tomatoes page; not a list of obscure blog articles that can only be surfaced with a web-crawler based search engine.

    My opinion is that web search is on the verge of being disrupted from the low-end. Siri is one way this can happen, but their are other possibilities like the new Spotlight features which search Wikipedia directly (which actually reminds me of Sherlock).

    A low-end disruption must be more convenient. I envision, in addition to Siri, Spotlight being integrated into either the keyboard, Control Center or the app switcher. As it stands right now, it’s rather inconvenient to have to go to back to the home screen to access Spotlight.

    1. “Google’s sophisticated search algorithms are overshooting customer needs
      for a very significant portion of searches. This is a precarious
      situation from a Disruption Theory standpoint.”

      I think disruption theory does not apply in this case because 1) search is free, and 2) there’s no cognitive/UI overhead for the extra capability built into the search engine — you type in a simple search, it spits back results, the one you needed is above the fold. If and only if you need the extra power to find something obscure, it’s there.

      1. Yes, I agree with this and was going to comment on that. I’m not sure how low-end disruption plays here. What is key at a fundamental level is how the interaction model changes. Therefore the old paradigm of search (a text box) becomes obsolete and the new paradigm is the normal. We are years away from this, but that is where I believe it is going.

        Those who can tightly integrate this new paradigm are well positioned.

        1. Ben, I think what you are talking about is how to get the search query into the system. What I’m more curious about is how the results come out.

          I the traditional Google paradigm, the results were a list of web pages which may or may not contain the answer to your search intent. The new paradigm, which Google itself is pursuing through the Knowledge Graph, is to answer with your search intent itself. In the old paradigm, a search for “Larry Page” would return a list of pages. In the new paradigm, you get a photo and a description. The description typically comes from Wikipedia or another database that consists of human curated information.

          In the new paradigm, you get to the results directly. If you are on a mediocre network, you save quite a few seconds. If you are mobile, the benefits are even more significant.

          The point here is that the to imitate the Knowledge Graph, you don’t need Google’s huge machine learning system. In fact, if a better “Sherlock” could be sufficient. All you have to do is to simultaneously send out the queries to Wikipedia, Yelp, RottenTomatoes, etc. and aggregate the results. For the vast majority of searches, this would be sufficient.

          Similarly, if you want to revisit Techpinions, you don’t need to go to Google. All you have to do is to do a search against your browsing history from which you can be send directly without going to Google. This can be done locally so the results are much faster. MacOS X and iOS do this to a degree, but is not yet reliably better than Google.

          Although we may be years away from a paradigm shift in how we enter search queries, I think that the shift in how we get the results is already ongoing. Google is actually ahead of the curve on this so as long as they find a way to monetize and dominate, they will be OK. If they can’t, then they are in trouble because this paradigm has a lower barrier for entry. Traditional organic search itself though will become less and less relevant.

      2. I’ll reply to 2) in my response to Ben’s comment.

        As for 1), I’m not so sure. With Ad based business models and lot’s of subsidies in the tech market, it’s easy to go to negative pricing. All you have to do is to be a bit creative on the business model side.

        For example, iPhones are frequently negatively priced in Japan. You actually get paid to buy an iPhone (of course you need a contract).

        1. Low end disruption requires that the overserving incumbent be somehow a worse choice than the “good enough” competition — either by costing more, or by not delivering as good a UI/UX for the customer’s needs.

          If the overserving incumbent is no cheaper, and fully satisfies the customer’s needs, then it’s impossible for a “good enough” competitor to disrupt them. Currently, it seems to me that Google is not disruptable. True, they are being bypassed in some situations, but that’s not because of low-end disruption, it’s because they’ve gotten on the bad side of their former partners (eg, Apple, mozilla), and so those companies are choosing to go with other services to fulfill their Sherlock/Siri/default homepage search queries.

          1. Google has several weaknesses that are deeply engrained in their vision and how they earn money. Any company has.

            In Google’s case, they ultimately have to earn money through advertising. Currently, the vast majority of these ads are served on their search results in the form of AdWords. This means that Google needs to show search results.

            However, showing search results is not necessarily the best UI. In the many cases that I outlined, the user wants to go to the Techpinions website or they want to go to a specific Wikipedia article. The search results are not what the user wants to see. There is a UI advantage in not showing search results.

            As our devices get more intelligent about our location, our history, etc., it becomes easier for devices to pinpoint our intents so that they can directly take us to our destination website. They don’t have to give us a list of what we might be interested in; they should already know.

            This is where the current Google Search UI can be improved and is not fully satisfactory. This is more acute in mobile because you are typically more in a hurry and have a smaller screen. Hence this is where disruption may happen.

            Again, I think that what Google is doing with the Knowledge Graph and Google Now, suggest that Google itself is very aware of the danger.

            As to your question regarding whether this UI without Google search results would be better or worse. It would be worse for the more demanding customers who want to do more complex queries. It would be better for simple queries because you can get to your desired web page faster.

            Regarding cost, imagine if Baidu wanted to take the place of the Google search field on the home page of Android devices. What would they do? I would imagine they would create their own AOSP device, stick their own search field instead of Google’s, and subsidize the cost of the phone. The phones would be cheaper than pure Android. This is the strategy that the Kindle Fire started out with, but somehow lost its way.

    2. I agree that there are a significant number of mundane low-value queries taking place on PCs because it is very quick and convenient. On my phone, however, I do nearly do as much of these lookup searches because it is not nearly as convenient. My hypothesis would be that mobile-only users rely much more on home screen bookmarks, regular bookmarks and apps.

      1. That is definitely true.

        However, I think the interface could improve significantly to make these mundane queries easier on phones. In many cases, you can make them much faster without going through the network, because you could search through your browsing history, your local dictionary, etc.

        Either way, the effect is that sophisticated search engines on the Internet will be used less.

  2. I’d take those survey results with a large helping of rock salt. A decline from 90% to 85% over 5 years might just reflect improvements in browser technology more than anything else.

    First, considering that many people don’t seem to know the difference between their web browser and the internet, I think a large proportion of the population doesn’t necessarily know when they’ve used a search engine and when they haven’t. Especially with the way search engines are now integrated into the address bar of every browser, it’s entirely possible to search the web without ever seeing the search company’s branded home page and thus without necessarily being aware that you are searching. Especially if the browser defaults to some kind of “I’m feeling lucky” setting.

    Unless the survey specifically asked questions designed to evade the ignorance of the non-technical population (ie, “how often do you end up on a page that has a list of possible places you could go based on what you just typed?”), I’d be very hesitant to trust its results.

    Second, most web surfing is not “surfing” per se anymore, but rather going to a list of places that are already known and have been previously visited. Web browsers are getting much better at interpreting what you type into the address/search bar and offering you places from your browser history, which kind of bypasses the need to have your home page be a search engine in this day and age. And with browser syncing, you never need to build up a web history from scratch anymore. I still set home pages on every machine that I touch to Google.ca or .com for my American relatives but it’s more out of habit than the necessity it used to be. So extremely slight decline in searching might just reflect the reduced need/desire to go looking for what you need. I am sure that when people need something different, searching (in one form or another) is still the first thing they are going to do, whether they’re aware they are doing it or not.

    1. This is just one bit of many data points I’m seeing that searching is happening less on desktop web. The world is going mobile. And if we were to just focus on those results from countries without a PC bias, the results would be quite different.

Leave a Reply

Your email address will not be published. Required fields are marked *