Wednesday, March 08, 2006

Customization in Windows Live Search

Chris Sherman at SEW posts a detailed review of some new features and enhancements at Windows Live Search.

Continuing the brand confusion between Windows Live and MSN, it appears that Windows Live Search is not intended as a replacement for MSN Search. This is despite the fact that, as Chris says, "the differences between Windows Live Search and MSN Search appear to be largely cosmetic," and that the two services use the same underlying technology.

The cosmetic changes in Windows Live Search appear to include new widgets to control your search and the display of your search. From Chris:
Sliders allow you to resize search results, adding or removing information such as descriptions, or in the case of images, making thumbnails larger or smaller.

"Macros" [are] a feature that lets you customize and save searches to be run again ... easy to create and share.
As I described in my earlier post, "Different visions of the future of search", it appears Microsoft's web search efforts are going down the path of trying to improving search by giving users lots of knobs and controls and then hoping they will use them.

Back in October 2004, MSN Search was about to launch. They were promoting their nifty sliders as a differentiating feature. At the time, I said:
I suspect sliders would be of interest only to power users, the kind of people who already use advanced search all the time. Most users want search to "just work". This kind of control is just a novelty, not something most would actually bother to use.
Search engines typically see less than 1% of searches using their advanced search interface. I wonder if MSN Search has seen much higher usage rates of their sliders, data that might cause them to believe that they can get the majority of users to bother with these knobs and dials.


Anonymous said...

Ok, so that is fine if you expect a search engine to "just work". But what do you want to happen when it doesn't work? I'm sure you saw this post on Battelle today. User satisfaction for both Google and the now-tied-with-Google Yahoo is around 50%.

Let's loosely translate that into saying that half the time the system doesn't work. What happens then? Where do you go from there? With G and Y, you're at a dead end.

At that point, I would much prefer a retrieval system (I'm not necessarily saying Microsoft's.. I'm just saying a system..) that would give me a way of going forward. Widgets, sliders, whatever.

Folks, even average folks, are used to pressing buttons and manipulating widgets all the time.. to sort their email, to arrange playlists on their iPod, to clicking thumbs down or thumbs up on their Tivo. People give feedback all the time to computer systems. What is so wrong with expecting that folks will use a widget or two, here or there, when searching for information on the web?

Maybe one reason only 1% of users use advanced features is because they're fairly hidden and unintuitive behind the "clean" modern web search engine interface. What if, through proper design, the proper tool became available to a user at the proper time, in the proper manner? What if designing the tool was an integral part of the ranked list navigation design, ala Vivisimo as we discussed a few days ago?

I guess what I just don't understand is what you expect the alternative to be, if you don't think people will use tools.

Somehow, Amazon manages to elicit tons of user interaction.. listmaking, reviews, etc. What if a search engine were designed the same way? People would use a tool if that tool worked.

Greg Linden said...

I do think people will use tools, but I think people will always prefer a tool that requires less effort.

You say that the reason 1% of people use advanced search is because it is hidden from them. I'm not sure that's the reason. For one good counterexample, only 3% even bother with quotation marks or other query syntax, and that's not hidden.

There are ways to help people find what they need without requiring them to do more work, including suggestions on query refinements, question answering, and personalization.

But a more complicated interface? Tools that most people don't understand and will never use? That won't help grandma find her soup recipe. That won't help the mainstream.

Anonymous said...

I think we agree more than we disagree here. Especially when you say "There are ways to help people find what they need without requiring them to do more work, including suggestions on query refinements..." That's basically what I am talking about here. If there are widgets that need to appear, let them appear within the context of what the user is doing. Don't just have purposeless widgets floating around.

W/r/t the quotation marks: I have to disagree. That feature is totally hidden. I went to Google just now and looked. Nothing about it on the front page. I clicked advanced search. Nothing about it there. So I clicked the "Advanced Search Tips" link on the top right. I had to scroll down halfway through that page, and finally found a link for "Complete list of advanced operators". In the second paragraph on that page, it finally tells me about the "" quotation operators.

That's totally hidden.

Ethel Q. Grandma / Joe Q. Public does not use the quote operators.. not because they're lazy or it is extra work to type in "", but because they don't know it exists. No effort is made to help the user discover that those operators exist.

This is not just an academic example. Just the other day my wife was getting frustrated while searching for something. She's not a computer person. I sat down with her and said "honey, why don't you put quotes around these two words". She said, "What are those?" I showed her, the information she needed appeared almost immediately, and now there are 3.00001% of users that use quotation marks in their queries.

The Google interface is like a Unix command line. If you know what you are doing, it is very powerful, and not a lot of extra work to add a single extra command line option to something you are doing. So I agree with you.. if you're asking the public to learn "unix", you are going to fail. That -really is- a lot of extra work on the part of the user.

Google does offer query feedback, in the form of spelling correction. That is a very good thing. It doesn't make the decision for you; it doesn't correct your spelling automatically. It does actually require the user to do some extra work, by clicking the "did you mean" link if that's what you really did mean.

So that's all I'm talking about, for all these widgets. Do the same thing for quotation marks. When the user types [new york mayor election], have the system pop up a specific widget that says 'Did you mean ["new york" mayor election] or [new "york mayor" election]?'

If you did that, I bet you'd get a lot more usage out of the quotation operator.

One might argue that it would be best for the search engine to default to ["new york"] rather than ["york mayor", because the latter phrase is much less common. I would agree, but I also think that if you don't give the user a way of specifying which they actually mean, and ALWAYS default to "new york", then you're going to piss off a heckuva lot of people from York, England.

The point is, only 3% of people use this feature, still, not because they're lazy or won't do extra work, but because the system has not been designed in a manner to make it clear to them (1) what the feature is, and (2) what it does.

Google's philosophy is that it is going to make the right decision for you, and NOT give you the option of choosing anything other than what it thinks is the correct interpretation of your query. I'd much rather have a tool that defaulted to what the search engine thought was the right decision, but then offered me a widget to give feedback and correct the search engine's understanding of what I really meant.

If 3% of the users still only use the widget, and you've got 100 million users of your system, that is still three million people whose lives you are improving. And if it jumps to 4%, because you've designed it cleverly, well, that's an extra million people.

Greg Linden said...

Hi, Jeremy. I think you're right that we agree more than disagree here.

I think the core of our debate comes down to your last paragraph where you talk about the benefit from getting 3% of your users to use quotes up to 4% of your users.

In terms of what the search giants need to do to win, impacting 1% of their users won't make any difference. To really help a lot of people and change their market share, they need features that have immediate and obvious benefit to most or all of their users.

Anonymous said...

I think I've already exceeded my correct etiquette number of blog comment posts on this particular thread. So apologies in advance.. :-)

But just to correct: I think that if you really did design the quotation operator recommendor/feedback system, you really would see more than a 1% increase in the usage of that operator. My point in saying 3%->4% was that, even if it were only 1%, you would still be affecting a million people, which is no small number, compared to a fairly small and helpful widget.

I personally think the number would be much larger than 1%, if you made the widget appear in the same way the spelling correction widget appears: Only when it's needed. Let it be data driven. Let the system decide, when a sequence of words has high entropy, by which I mean the potential set of results is vastly different depending on whether and where quote marks are used. In those situations, make the widget appear. And let the user simply click a "this is what I mean" link, rather than have them have to program some sort of Lisp-ian nested structure: (new ((york mayor) election))

I can easily imagine it jumping to 6%. Or 9%. Or more. Check out this summary of a talk by Google's Mayer last year at PARC:

One of the biggest leap in search usage came about when they introduced their much improved spell checker giving birth to the "Did you mean..." feature. This instantly doubled their traffic, but they had some interesting discussions on how best to place that information, as most people simply tuned that out. But they discovered the placement at the bottom of the results was the most effective area.

Wow.. it doubled their traffic! So what does that translate into.. 25% of the users? 90%? How many people does it have to affect, for their traffic to double? I'm sure at least 50%.

And this is from just one little query refinement widget.

And the beautiful thing here is, at least according to Mayer, they only got this traffic doubling after the widget appeared in the correct place.. at the bottom of the first page of results.. the place most people look when they've scanned through the top 10 and realize their search has failed.

Before they put it there, most people simply tuned it out. Most people == 80%? 90%?

So we're talking about a jump from maybe 10% of users to 50% of users? Even allowing for a lot of slack in my hand-waving number estimates, that's a huge improvement. Traffic did double, which means, on average, the widget got used once for every search done!

Here is this tool, this widget, that was always available, that users could have used any time, but didn't actually use.. until it got integrated with the search in a way that was clear and intuitive. The users still did have to do extra work. They had to click/use the widget. It wasn't given to them for free. But they didn't actually use it, until the search engine made it clear to them that they could, should, and why.

And if you were to do the same with other widgets, why would they not also see huge jumps? The quotation mark operator really is buried right now. Let's dig it out, the same way spell checking was dug out. It might go from 3% to 40%.

Anyway, I've now offically overstayed my welcome :-) I'm done. Oh, BTW, while at Amazon, did you know a fellow named Matt Green? mb small world..