Steve Rubel kicked off a furore among the SEO community during the Google Instant launch event on wednesday last week, which most people were following on Youtube. The furore was specifically caused because he blogged a classic "SEO is dead" post - somewhat of a tradition among the bloggerati. Matt Cutts and Vanessa Fox were quick to respond and steer the discussion away from hysteria. Nonetheless, to give credit to where credit is due, Rubel honed in on the question on everyone's lips, "how does Google Instant effect SEO?" Following our recent post 'Google Instant - 10 Things Marketing Teams Need to Know', in a discussion with the Search Engine Watch experts, it was deemed to be important that SEW address the question too.
1. Google's Official Position
One of the marvellous effects of Twitter is that Rubel's blog post reached audiences so quickly that the discussion had surfaced to reporters actually at the Google Instant presentation event ahead of the Q&A session. Serendipitously, Ben Parr of Mashable was able to ask the Google team sitting on the stage (among whom was Sergey Brin), and i paraphase, "what impact will Google Instant have on SEO?".
Google responded that "ranking stays the same" and fundamentally, everything remains the same. However they emphasized that behavior may change over time - the kinds of queries you do may change over time. Obviously, this could have a tremendous impact on SEO.
2. "No two people will see the same web"... But We All Live In the Same World
The cornerstone of Rubel's case is his statement above (quoted in the sub heading). He argued that SEO relied on users seeing the same results, and this caused certain phrases to be targeted as they had higher volume of traffic. Now users would be continually adjusting their suggestions as they search. The implied impact of "tweaking their searches in real-time" is that volume on high traffic phrases will now be distributed between every predicted suggestion as users.
"Once a single search would do the trick - and everyone saw the same results. That's what made search engine optimization work. Now, with this, everyone is going to start tweaking their searches in real-time. The reason this is a game changer is feedback. When you get feedback, you change your behaviors."
The statement above sounds sensible on the face of it. The likely impact of predicted searches is that users will investigate more sets of results. However, as Google said, ranking factors have not changed. So, despite streaming SERPs, results for every search term are still governed by the same algorithm - meaning that they are exactly the same regardless of how 'instant' they are.
As Vanessa Fox responded in her blog post, fundamentally there is a misunderstanding at the heart of Rubel's argument - even if personalized search meant that users would not see the same web, it does not follow that users would act on different searches.
...the biggest misunderstanding of Rubel's post is that SEO is about optimizing for a single query and that everyone saw the same results until now. In reality, searchers have been seeing different results for a really long time. Personalized search in particular has been increasing over time, causing everyone to see something different. And Google Suggest has also been around forever, so the idea of prompting refinements as the searcher types a query also isn't new.
Sure, searchers may tweak their queries in real time, but they aren't going to fundamentally change what they're looking for. If I'm looking for a restaurant in Seattle, I'm not going to see results for "relaxing vacations in Mexico" and decide to go to Cabo instead of out to dinner.
As Fox illustrates, personalization of the web does not have an impact on the demands of real life, the latter of which is what people use search to overcome. Whilst I would agree with Rubel's statement that "Google Instant... will change and personalize people's search behaviors", I would concur with Fox that most people's need for search has not changed.
Even if it were true that "no one will see the same web" this would have absolutely no impact on the everyday needs of a searcher or the way in which a searcher chose the right result. Someone looking for a holiday, is still looking for a holiday whether the web is personalized or not. Whilst new user signals introduced by Google Instant may have an impact on what the user discovers, all of the new features increase the avenues of discovery and choices available to them rather than creating limits. The user still has choice at every single step of the search to conversion process, and an increase of choices within the 'query space' of the funnel, frees up the market even more - but the fundamental direction of their choice remains aligned to the actual need and not what they see.
In summary, Google Instant makes finding the answer faster, whilst the market for the answer has not changed in any way. If the demand for search remains strong, the need for SEO remains no matter how personalized the web is.
3. "Feedback changes your behavior"... But Any Behavior Change Increases Dependency
The gold nuggets in Rubel's position however is that Google Instant is fundamentally about personalization via the feedback mechanism. Ben Gomes talked in detail of the feedback mechanism within Google instant. Built on a complicated hardware and software architecture, lesser known Google search features such as Spelling Corrections, Starred results, Trends, Hot-Trends, Google Squared and Google Suggest ultimately power the user feedback loop by organizing and incorporating real time data back into the search interface. Relatively speaking, aside from Google Suggest, these product features are still in their infancy, and whilst personalization was not specifically covered in the presentation, it's not far fetched to see how Google Instant sets an interesting new blueprint for personalized search results.
However, whilst I somewhat agree with Rubel's statement that, "Real-time feedback will change and personalize people's search behaviors," I would say that it absolutely is far fetched to believe that such a change is radical enough to affect SEO.
As has already been argued, even if search was personalized to a high degree, it would have a minimal impact on SEO because the need for search does not change with personalization. Matt Cutt's response to Rubel, develops this objection further to argue that if anything the need for search increases as the method of discovery improves.
[Q: Will Google Instant change search engine optimization? ] I think over time it might. The search results will remain the same for a query, but it's possible that people will learn to search differently over time. For example, I was recently researching a congressperson. With Google Instant, it was more visible to me that this congressperson had proposed an energy plan, so I refined my search to learn more, and quickly found myself reading a post on the congressperson's blog that had been on page 2 of the search results.
As Matt Cutt's shows, the feedback loop created new avenues to discover related topics which led him to conduct more and more queries. The crucial innovation to watch regarding Google Instant is whether feedback feeds search - to generate more queries per user.
Form To Function
Other bloggers have also chimed in on the question of whether Google Instant kills SEO to develop some of the common sense points raised by Fox and Cutts with more concrete examples. What emerges from their objections is that there is a fundamental disconnect between the form to function of Google Instant. It's designed to make search "faster than the speed of type" yet in effect the UI change restricts what is actually possible. Firstly, the personalization of Google Instant is simulated. Secondly, Google Instant changes the value of SERP real estate. Thirdly, Google Instant could lead users to the dark side of nonsense.
4. Only Search Suggestions Are Personalized - Instant Results Pages Are NOT
Rae Hoffman of Outspoken media, in her typical fashion, stops Rubel's argument dead in the water with the point that Google Instant results are ultimately the same for everybody.
The "Instant" results may vary from the traditional Google ones... but everyone using Instant, from what we can tell, gets the same "Instant results." And SEO (and by that I mean both technical SEO and marketing SEO) is still how you get your website to the top of both.
In fact, the search results pages are no more personalized than they have been for at least 12 months, via features such as starred results, web history, and social circle results. The only aspect of Google Instant that is personalized is the terms it suggests and this again is a misnomer. As noted in our previous post, the suggestions are localized based on your IP address. So a search for 'carpet cleaners' will suggest local search terms that approximate to a service nearby.
5. Lack of SERP Real Estate Leads to a Price War Which Benefits SEO
Aaron Wall from SEObook investigates the page coverage of a single results page based on browser size and concludes that "when Google includes 4 AdWords ads only 50% of web browsers get to see the full 2nd organic listing, while only 20% get to see the full 4th organic listing."
The net effect of this will be a price war for top paid ad positions and even more competition for the top spots in the organic results. Wall concludes, "Google instant only increases the value of a well thought out SEO strategy" because:
In support of Wall's argument is Google's own eye-tracking study data, presented at the launch event. Whilst premium ad spots are not shown in this example, the screenshots below clearly show that the results immediately below the search box get the most attention, as do localized results.
5. Users Are Not Stupid
Charlie Brooker, a comedian, writer and journalist, also notes that many of the search suggestions led him down a black hole of productivity. Whilst Brooker consistently plays for laughs, I am sure many can relate to his first experience with Google Instant. I certainly can - the first thing I did on Google Instant was see how much nonsense I could get it to predict.
Back in that room, bombarded by alerts and emails, repeatedly tapping search terms into Google Instant for no good reason, playing mindless pinball with words and images, tumbling down countless little attention-vortexes, plunging into one split-second coma after another, I began to feel I was neither in control nor 100% physically present. I wasn't using the computer. The computer was using me - to keep its keys warm. (Apart from "enter", obviously. I didn't even have to press that.)
Fastcompany also note that Google Instant plays right into Microsoft's hands -- especially given the company's "Search Overload Syndrome" ad campaign. Everything about Google Instant search screams search overload--the unfortunate side-effect of sifting through millions of results regardless of relevancy.
The biggest challenge Google Instant faces is persuading people that the search suggestions are relevant all of the time. If suggestions stop being shortcuts, or become continual distractions, it may be sudden death for the streaming results UI.
6. Recognition is not Relevance
A theme that got played around with a few times, slightly uncharacteristically, at the Google event was the ability of something to be 'psychic'. Not being psychic, I was personally disappointed to discover some presentations were peppered with notions of a some sort of psychic robot, as it undermined the pure maths that makes suggested discovery workable.
A real life psychic, such as the types you find on every block of New York City, plays on their worldliness to suggest possibilities that might happen to their subject. The effectiveness of the 'reading' is totally dependent on their ability to trigger recognition of possible outcomes in their subject. Yet, it's their subject's own experience that judges the validity of any scenario. The more they recognize themselves in the reading, the more open they are to possibilities that the psychic presents the more they are inclined to believe that what is being said is relevant to them. The opposite is also true - if the subject does not recognize anything that the psychic is saying, they will deem the reading nonsense. However, the tension between truth and nonsense is so taut that we tend to see them in absolute terms - we sustain and elevate truth to a priority concept and disregard nonsense to such an extent that we drop the notion entirely.
Predicted search operates in a similar way. Based on a series of guesses and a tonne of data it Google Instant can offer a series of bets as to what you might be searching for. But that's all they are - bets based on a data set.
Every search conducted and every result retrieved and every every listing clicked on are, at best, a continuous series of guesses. The first search made was, at best, a guess as to what their specific need was in contrast to the all the worlds information. The results the search engine returned, were at best, a guess based on mathematical relationships in contrast to a set of identified keywords. The result a user clicks on was the best guess a user can make on the results offered by the search engine. With all this uncertainty at the heart of search, it's a wonder search works at all.
Yet it does work, simply because chance itself exists. Even if the odds of any particular scenario were a one in a million, a chance nonetheless means it can actually happen. One correct prediction is a satisfying experience of recognition that deepens trust and leads to a deeper inquiry into the subject. The net effect is that, just like as with a Psychic, your information needs change as your trust increases. Matt Cutts discusses the exploration of questions in his own post:
...with Google Instant I find myself digging into a query more... I can get a preview of what the results will be, add or subtract words to modify my query, and hit enter at any time. The ability to explore the query space and find out new things will inevitably lead to changes for SEO. When I was in grad school, I had a professor who mentioned that peoples' information need often change over the course of a search session. Google Instant makes that process even easier: people can dig into a topic and find out new areas to explore with very little work.
And if it doesn't happen, well, it just didn't happen. Google Instant has very little to lose, as it just guessed wrong and the user will ignore the suggestion as nonsense anyway.
The unique selling point of real time search engines such as OneRiot and Wowd is exactly the same. For example, using sophisticated topic clustering techniques from crawled data, One Riot can 'predict' trending topics based on likely outcomes from a collection of related news stories.
And get this: user behavior signals make the prediction element become more accurate, the less data there is.
7. SEO Success is Totally Dependent on the Search UI (as it ever was)
Recognition generates more clues as to where relevance might lie, but neither pertain to the other. Think about it - something you recognize is not necessarily relevant, and something that is relevant does not necessarily mean that you recognize it. A similar interplay of recognition and relevance is determined by the concept of search 'rankings'. The continual assumption is that the top results are the most relevant results.
The search marketing industry exists as a whole precisely due to user search behavior. Despite a trillion web pages of options, the vast majority of users barely scan results and rarely look beyond the first page. It's this economy of attention that everyone is trading on.
Furthermore, recognition is something that search engines and search marketers in general have been grappling with since time immemorial. Bolded results were deployed by search engines to provide stronger signal to users that matches have been found. Most SEOs build on that search behavior to re-architect websites in such a way that a high degree of visual match is shown to the user. Similarly, PPC ads are written in such a way to trigger the 'bolding' of ad text against their match types. Google Instant is no different, and the eye-tracking screenshot below taken from the presentation, shows how users are still looking for the exact same 'emboldened text' signal to determine their choice of click.
The upshot of the tension between recognition and relevance, on search engines, is that a user's need for an objective 'uncolored' view of the world persists, yet is equally tainted by a need for familiarity - and that paradox is directly expressed through search. Users want to find their own way to content, but also want to have the route signposted.
Google Instant's predictions are simply signposts for the user and an attempt at displaying the fact that it 'recognizes' the topic. As a cluster of relevant topics are discovered, the query gets more complex and predicted suggestions help users to explore that topic and think laterally by generating recognizable directions that the user can take. This has got to be a good thing for SEO as the suggestions fragment the market on high volume terms and promote more specific variations.
Branching Topics, Suggested Money Trees
To conclude, it is more likely that predicted search suggestions will increase volume on the mid-tail and longtail, because Google can effectively monetize by increasing competition on AdWords. Suggested search terms are likely to increase the amount of traffic on all of the 'predicted' search terms. The technology effectively multiplies one search into 5 possible searches. Local services stand to benefit most in the short term. Taking the 'carpet cleaners' example from the previous post, the localization of suggestions may distribute the demand further along the tail of local suppliers. This is great news for SEO and smaller businesses.
A key motive for streaming search results, aside from speedy delivery, is justified by Google's business model in general, rather than part of a wider move to eradicate spam or 'kill SEO'. More searches per user naturally increases the revenue that Google can generate via AdWords. If Google keep innovating Instant and make it's search suggestions more timely and more effectively draw lateral connections in real-time across then we should see an overall increase in usage. The key statistic we will be looking for over the next few months is to find out whether the Google Instant UI has increased the ratio of searches per user. Ultimately, that will be the signal that the feedback loop of instant results is here to stay.