Just one slider, but not so simple Ryan 02 Jun 2005

34 comments Latest by Joe

Yahoo! Mindset is standard Yahoo search with one new feature, a slider with “Shopping” on one end and “Researching” on the other. The default position is dead center, and if you slide to the left or right your results are updated accordingly.

The slider

The tech works. Sliding toward “Researching” tends to bring out more reviews and more discussions. Sliding to “Shopping” throws you straight into the market. It looks simple at first, but here’s where they went wrong: The slider is 300px wide with a continous motion from one side to the other. That means there are 300 settings to choose from. And what does 27% from the left mean anyway?

I’m sure the elegant and abstract Shopping-Researching continuum appeals to the science types behind Mindset, but it won’t mean squat to searchers until A) the problem is better defined and B) the interface is redesigned to reflect that.

Based on the current interface, I’m going to guess Yahoo has defined the problem like this:

People want to decide between commercial and noncommercial results.

This is too fuzzy. Here’s one way to refocus:

Sometimes when people search, lots of results for online stores get in the way of the information they want.

A more focused problem suggests a simpler interface. Instead of 300 settings, three options would do the trick:

  • Show me all results
  • Just show me online stores
  • Show me results without online stores

Or even simpler, a single checkbox: [X] Don’t show online stores in my results.

Regardless, kudos to Yahoo for the neat research. I’m sure that as they move forward they’ll come to see that even having only one slider doesn’t necessarily make the interface simple.

34 comments so far (Jump to latest)

Eric Goodwin 02 Jun 05

I agree. Whenever I have used this new tool from Yahoo, I have alway had it all the way to the right, or all the way to the left. I haven’t found any reason to have it in between. The simple checkbox, would suffice nicely.

Rick Faaberg 02 Jun 05

I would guess that their algorithms can only do so much and there are valuable research sites that happen to hit too many of the warning signs of a shopping site (and vice versa, though probably not as often) so they get weeded out. If your results are a little too sparse it’s nice to be able to back off from the strictness of the criteria, though I realize that this may in the minority of cases.

Phil Boardman 02 Jun 05

Agreed. “I only want to go part way to the shops” is a pretty ridiculous idea. What about a “Search in stores” column and a “Reviews / Discussions” column, surely a two column layout isn’t too hard to implement, with a “More like this” link at the bottom of each to continue searching…

No check boxes, only half a page of wasted space. Sometimes people don’t know what they’re looking for until they see an example. The might not even notice the check box.

Oh, shopping results, that’s what I’m looking for - more…
Oh, review of the product, that cool too - more…

It’s like saying you can start doing something until you’ve configured it, although I guess this is the default behaviour of the control element, it probably isn’t needed if you can separate the two.

Fin 02 Jun 05

It’s interesting how Y! rates your site. Shopping or information? Perversely, mine’s not as far along the information slider as I would have hoped.

grant 02 Jun 05

Dare I suggest this is a UI gimmick to get people using the new feature, and to illustrate the concept?
It might be refined in future if/when the idea becomes popular.

Tinus 02 Jun 05

The moment this feature makes it into the ‘normal’ search function lots of SEO people will have a very bad day..

Don Schenck 02 Jun 05

What’s “SEO”?

I worked at a start-up where the head guy decided to use a slider between two extremes. One extreme was “optimize for speed”, the other “optimize for cost savings”.

I argued that we needed, perhaps, five settings. Better yet, I wanted to give the user five result sets and *then* let them choose.

I lost the argument. I left. They folded.

Moral: I’m always right! *cough*

Too-finely-tuned controls are too hard to use.

kevin 02 Jun 05

The cool part is not the UI, which I’ll admit needs a bit of work - it doesn’t take a rocket scientist to offer suggestions for simplifying the experience. What I think would be fascinating to see evolve is the machine learning that’s integrated into the classification. In the end, I doubt you’d be able to get as interesting classification results from a polar selection of ‘show/don’t show online stores.’ After all, this is from Yahoo! research.

cjs. 02 Jun 05

I agree with Don, as he _is_ always right.

Adding spaced ‘clicks’ to the slider might make it easier to use.

(“SEO” = Search Engine Optimization, just in case you weren’t being facetious.)

Josh Jarmin 02 Jun 05

I would like to see something like a tabbed system at the top. The default tab would be all results, on the left have a tab for shopping related stuff, and a tab on the right for research (should be labled information) related stuff. If they use AJAX, you could switch between the tabs without refreshing. And it wouldnt get in the way of regular users who wouldnt probably notice the tabs anyway.

Adam Bardsley 03 Jun 05

Actually I think the is a very good reason for a continuous slider. It would be better if it was truelly analog in my opinion. Why?

Well actually you slide the bar in teh direction until the search results match your needs. Say I’m buying a Sony VAIO Pocket. I want to read reviews and decide wheteher to buy it as I slide right It removes people who are only offering to sell it but keeps those who maybe sell it but offer reviews, I keep sliding until I get rid of the people who are (in my mind) selling and offering reviews until I reach those offering reviews (who may still be selling on the side..)

It’s a very fuzzy line and I think everyone will choose it at a different point. I expect after a bit of analysis Yahoo will be able to see if there are patterns and we all select certain points along the gradient or not. I’m betting we don’t

Olly 03 Jun 05

Hmm. I think the slider is a good thing. Shopping can include research - for instance if you’re comparing prices across a number of stores.

Secondly, some onlne shops carry customer reviews and suchlike. They can be biased (i.e. edited/censored to help sell the product) but they’re still helpful.

I’m not convinced it needs 300 settings though.

Nice Paul 03 Jun 05

The slider would be a pain if you were trying to find a specific result by repeating a search you did earlier in the day. There’s no way of telling exactly how far along you had the slider at the time of your original search.

Douglas Livingstone 03 Jun 05

There aren’t 300 settings, there are two settings.

One is a string with a billion options (OK, more than that) and the other is a number with 300 options.

The billion options gets labeled the ultimate in simplicity but the 300 options get you billed as too complicated??

Come on guys…

Unearthed Ruminator 03 Jun 05

I like the slider idea myself (I’m with Rick on this).

Joshua Porter 03 Jun 05

Thanks for the pointer, Ryan. Though your critique may be right on, isn’t the bigger story here that Yahoo has entered the Ajax fray with a brand new control (albeit similar to the Google Maps zoom), one that we haven’t seen before that could be potentially useful (although it’s current use is debatable)?

This is big news…another leading company advocating Ajax use.

DXO 03 Jun 05

The slider is good. Why? Because it’s fun and it works. People like fun things that work. The slider gives the user the three options you name as essential and some fun stuff in between. Nothing wrong with that.

Dan Boland 03 Jun 05

I agree with DXO. The way I see it is that the slider falls under the “bells and whistles” category. Sure, you could set up the slider with three or five predetermined stopping points (like how the Junk Mail Filter slider is set up in Entourage for Mac), but then you’ve turned it into being just a different way to represent radio buttons. The way Yahoo! does it makes better use of the slider as a mechanism for changing parameters. It’s a cool gimmick that works.

Jordan T. Cox 03 Jun 05

Wow. This is a _great_ feature. Assuming it works, it might even manage to draw me away from google. As a programmer, not just a web programmer, I’m continuously searching for resources - and coming up with course sales and book sales. Wonderful idea Yahoo!.

(I make the web programmer distinction because it seems much easier to find good online resources for web programming; not because I consider it a lesser programming field)

Adam Michela 03 Jun 05

It’s there for a very good reason. Not all sites can be extreme left or extreme right.

This is clearly indicated by the “scale” to the left of each result.

I’ve found use for it at all settings, but I agree with those who say perhaps there should be a few stops on the slider.

In the end though, like grant said, the slider is 1-part marketing gimmick.

Had Yahoo “simply” included a checkbox saying “Exclude Stores” or something of the like, it would have barely got a mention.

Mindset works, and it works well. Quick, accurate, and IMO… easy. Kudos Yahoo!

andy 03 Jun 05

I think both concepts have good merits, but I’d have to really see actual case studies with both interfaces before concluding which one “worked better.”

My guess is that people would prefer more clearly defined options like Adam outlines, but I would want a fourth option added: “Show me results from both, but only include online stores if the results are reviews.”

Of course… I guess I could search for “widget reviews.” and include both. Hm… never mind… I argued myself out…

..ak 03 Jun 05

This reminds me of the “results ranking” sliders for MSN Search. (go to MSNSearch.com, click on Search Builder, last “tab” is results ranking).

I agree with DXO: it’s a fun toy. It doesn’t take away from the experience.

My problem was the sponsored results pushed the search results to the bottom, leaving only one search result to be seen. I had to move the slider then scroll down to see what changes happened.

Search for “cruzer mini”

Jonny Roader 03 Jun 05

Nice idea, but still very much beta.

A search for ‘laptops’ seemed to work rather well, but a search for ‘css’ much less so.

Josh Jarmin 03 Jun 05

A search for Adobe seemed to be a bit wacky. When the slider is dead center adobe.com is first, but as you slide to the shopping side, you get all of the adobe.co.jp and adobe.no. Why did adboe.com (a major seller of all things Adobe) drop from the shopping side?

Douglas 04 Jun 05

> Why did adobe.com (a major seller of all things Adobe) drop from the shopping side?

.co.jp and .no focus more on selling than .com does, in Y!’s opinion. Clearly it is more than an on/off scale ;)

Tobias 06 Jun 05

All in all, I agree.

“Or even simpler, a single checkbox: [X] Don�t show online stores in my results.”

But IMO a 3-set of radio-button would be more clear:
O no filter, O more shopping, O more research
Otherwise I would ask myself how to deactivate it completelly.

DaleV 06 Jun 05

I like how it dynamically revises as you tweak the slider. IMO, THIS is simpler than all the ideas you guys are posting about tabs, and 3-5 pre-stettings and such … I like that you can just toss the slider over a bit more and get a revised set of listings (albeit slightly).

I showed this to a couple of non-techies this weekend and they thought it was very useful. I think this, or something like this, will become common. Nice job, Yahoo.

Bryan C 06 Jun 05

A slider with 300 potential settings is probably too fine-grained, but I don’t think the binary option you suggest would work very well. How do you define “online store” for the purposes of a search? If I review a book or CD and include a referral link to Amazon, would I be in or out of the results? How about cafepress merchandise or inline ads? Is it any site that sells things? Any site that sells the item you’re searching for?

I suspect that Yahoo had these same questions and realized the one set of answers wouldn’t always fit every users’ expectations. Thus the tuneable search results with an “analog” indicator.

Personally, I’d be happy with a filter that chopped out the reams of sites that republish the same identical afilliate reviews. And those annoying autogenerated pages that promise that I’ll “Learn more about [insert search term here]!!” Those really annoy me.

hcblue 08 Jun 05

I think that a slider may, in fact, be helpful in providing the customers fine-grained control over the search results—keeping in mind that ten stop-points may be enough. However, the problem is that in the current slider bar of “Research” versus “Shopping”, they are trying to define two things in one parameter.

Research and non-research would work (though it would negate the need of shopping, or at least shopping may need redefining) since they are talking about two sides of the same coin: research. Comparing research and shopping would be trying to combine two coins into one coin; four sides into two sides. This is when you would lose clarity of what is in question.

So… would being equidistance from research and shopping mean:

  1. Neither research nor shopping

  2. Both research and shopping

  3. Equal weight of research and shopping (whatever “weight” is defined to be)

dhulis 13 Jun 05

Imagine this site going live/commercial. Would the user manipulating the slider be subjected to an equally dynamic set of ads…which boils down to shopping anyways!

Mark Dawson 14 Jun 05


Sorry to reiterate what has already been said: I would just like to throw my opinion out there because I have always found the 37signals posts to be very insightful and was surprised to find myself in strong disagreement with this particular post.

Shopping v. research is a paramater that cannot be accurately described by a binary. Informed shopping involves research to varying degrees and research requires commercial undertakings as well. Thus, the idea that the best way to handle this problem would be with a simple checkbox seems incorrect to me.

Even setting aside the search-specific aspect of this research project, I think its commendable that Yahoo! is doing something different. It may not be the next killer app but becoming complacent with existing frameworks to sort and describe more and more complicated data is dangerous. At very least, this project has shown us that there are ways outside the old-school form to manipulate and use data intuitively. That may be all it does and that seems awesome to me. It is not in wide release and thus speculation on whether or not it is the best way to improve search as a whole seems overly speculative.

In sum, I have always enjoyed 37signals posts and hope not to sound overly critical. However, I was taken aback by the “this is just plain wrong” answer, especially given the complex and nebulous nature of the question at hand.