Jakob Nielsen Not Optimistic about Personalized Search
I had the opportunity to chat with usability guru Jakob Nielsen about what might be happening with search in the next few years and he's not bullish on Google's move to personalization. In his words "it's incredibly difficult" and he's not convinced that it will make much of a difference to relevancy in the next 5 years or so. More to come in Just Behave this week. Contrast that against a follow up interview with Marissa Mayer where she was literally bubbling over with enthusiasm for a "minority report" version of search. Still arranging interviews with Yahoo, Microsoft, Ask and our own Danny Sullivan and Chris Sherman (who are the hardest ones to nail down, but given this launch, I'm beginning to see why).
Related Links
There are no related links for this story.
Calendar
Network
Feeds
Went Hot: July 10, 2007 - 4:25 pm
16 Comments
Comments
I agree with Jacob on the fact that it will take longer to get personalization perfect than what many people think. It is indeed very difficult to get right. Its "easy enough" if your user group and traffic is limited but on the scale that the large engines need I agree, it will still take years ... 5 is a good guess
However, I am sure personalization will become a more important factor over time. Knowing the difference between "you" and "me" when a search is done can be very valuable. The same words just don't have the same meaning to us all
Ack! Where's the URL for the chat?
Nielsen - I don't blog, therefore it's not important. I don't like graphics, therefore they're not important. I don't like designerish website, so they're not important. And now... I don't like personalized search, so... it's not important.
Grrr...
I think the easiest parts of personalization to get right are to feed people back their overt biases and perhaps understanding users by frequent queries and how that matches up with topic sensitive PageRank. But outside of feeding people their own biases I really see limited upside to the value of personalization, and for the higher level queries usually just adding a word works well to make the query more specific...plus Google already tries to mix the result sampling.
Rand, I think Gord is sharing what he's talked to about with Jakob ahead of an actual article -- that's why there's no article link. Of course, it is odd that if you click on the headline, we try to still kick you out of the site!
there was some spoof of nielsen years ago where he did something like declare the letter "V" obsolete, sadly unable to find it
Rand's comment is great! Made me laugh. So true! And, Graywolf - I've heard of that spoof before, too! Some of my usability researchers told me about it. It was the letter "C", though:
http://www.uspressnews.com/articles/1101
The letter "V" would actually make more sense, rendering the spoof less exaggerated. Can't blame you for thinking of "V" these days, though -- the current administration in the US sorta makes me want to watch "V for Vendetta" over and over and over again...
What I don't get is this.
- I use google all day - I use Gmail - I have a personal google page with stuff I care about - I use the google reader for useful feeds - I have a google bookmarks account - I use google checkout - I use google calendar - I run several google groups - I've geolocated my biz for google maps/local
So you'd think google would have PLENTY of signals from me to personalize my search results. But I have seen only a couple results with an obvious personalized flavor, and none that were helpful.
So as heavy a google user as I am, if they can't personalize MY results, whos can they? It makes me think that I've given them so info about so many topics that they can't really make any distinction as to what I care most about. If that's the case, then PS will only work for the user whos signals are more vertical, like about cars or football or food. Give google too much, and PS can't help...
One problem is data consistency but another is scalability. Just because Google may have a lot of data on you doesn't mean that it is consistent enough across users to be useful and even if it is it may be difficult to process it all fast enough.
Personalization is not easy but I am sure we will see more and more of it - in one form or another.
Personalized search just doesn't work for me because I have multiple personalities. I perform some searches on behalf of clients, some searches as a mom, my 10-year old daughter uses my computer ... there's just no way they can get a clear picture of what I'm looking for. Personally, I don't want my search results skewed by these multiple personalities and there's no way that they can know which personality I have at the moment.
I think it is way too early to make too many predictions or even judgments on personalized search.
SEOs have concerns that are obviously very specific to our point of view.
It will certainly evolve over time, and should get better.
But the real test will be how it works for the average, typical user, who probably has a fraction of the interactions in a week than those within the industry.
It would also be interesting to know what the typical user thinks about the whole thing... but I'm not sure how many typical users are aware of it... or ever will be, which of course is part of the concern.
any time i hear about personalization, i definitely take today's personalized SERPs with a grain of salt, and focus on a longer view of the issue. there's no question what's at stake from the engines' perspective in terms of serving more valuable ad inventory and raising switching costs for users.
i think back to eric schmidt's quote during his interview with danny at SES san jose las year where he said something to the effect of "if you think about it, all the world's information includes personal information," {shudder} and his frequent comments that the primary obstacle preventing google from being more "useful" is that "we don't know enough about you."
at the rate at which google is providing free services with the effective cost of surrendering user data, there's no question they'll soon have enough information for a compelling offering in personalized search. five years actually seems like a pretty conservative bet to me. jakob can be bearish on personalization all he likes, but the train's not stopping.
I believe it will lead to "Personalized PPC Results"
We already know… Google rearranges the organic results based on your historical click behavior… Why not reposition the sponsored listings based on the same personal data?
This would help to maximize profits for Google as they zero-in on the users who already prefer "Brand X". (The rich get richer)
Has anyone noticed or looked into this?
Aloha,
Dave.
Dave,
I agree with you that that this is the end goal. And you know, I wouldn't care if I was being shown personalized ads. Amazon.co does this and it's no big deal.
On the other hand, if I ever click a wrong book on Amazon.com they then spend days looking life fools in my eyes as they continue to show me books related to a subject I was never interested in in the first place. The same thing could happen with personalized PPC, greatly harming its usefulness.
As for personalized organic SERPs, I still cringe about this idea and the HUGE margin for silly error.
Miriam
Dave, yes, that's a great point.
To that end, people are funny. On the organic side, many of us have concerns, both in regards to the effect on our clients as well in our own personal usage, yet on the PPC side, not only do most of us probably have less concerns, but even consider it a positive.
Truly two very different sides to the same coin here.
Miriam, that of course is the concern... personalized but either not or no longer accurate. You'd hope that if this is done, they could quickly determine this... but awfully challenging.
You'd almost need a "yes, show me more like this - no, I don't want to see ads for this any more" type option... but then maybe it isn't the ads, but the company... hard to account for everything.
Complexity scales quickly.
The thing that worries me is that potential clients with Google accounts might lure themselves into a false sense of security thinking that they have good rankings and that they don't need an SEO.
How about dodgy SEOs getting their clients Google accounts, so they think their meta-keyword-wielding warriors are getting them 'rankings?'