Jared Spool: Welcome everyone to another episode of the SpoolCast. Today, an amazing treat. We have Luke Wroblewski, who is with his own company now, Luke Wroblewski, Ideation and Design. Formerly of Yahoo, he was in charge of everything they did that was really great and not in charge of any thing that wasn't great.
But he is now coming to our User Interface 15 Conference where he's going to talk about web form design and beyond. He is the author of the book "Web Form Design" and he is with us today. Hello Luke, how are you?
Luke Wroblewski: Hey, how're you doing Jared? I just wanted to point out that, that actually was on my business card. All the good stuff and none the bad stuff.
Jared: That's right! That's right! Because they had people for the bad stuff.
Luke: That's correct.
Jared: They had a whole army for that, but just one guy for all the good stuff. I thought today, we would talk about this new thing. This week in the news, for those of you listening it'll be some weeks back, but Google introduced this really interesting way to use their servers called Google Instant.
Where as you start typing into the Google box on the screen, the moment you start typing anything it goes off and starts doing searches and displays the results. Luke, have you played with this?
Luke: No. I actually played with this in 2005 at Yahoo.
Jared: Ah! So they just copied you guys. Was this one of the things that you did or one of the things that people who do the bad stuff did?
Luke: I think it was a combination.
Luke: This is kind of an interesting story in of itself, Steven Hood who is a product manager over at Yahoo posted on his blog kind of the details behind that experiment that Yahoo ran and what they learned and why ultimately it didn't go to market back in 2005. But Yahoo had a product called "Live Search."
Jared: I remember that.
Luke: They ended up putting it on alltheweb.com instead of putting it on yahoo.com because they were mostly concerned about server load and just the capacity of the system to support this sort of instantaneous behavior. The team that was working on it and I was sort of peripheral to that team. I was on the search team but I wasn't directly on that project.
When they did the testing of that interface the code name of that project was Eureka. Peoples' eyes sort of lit up when they did usability testing. And they got really, really excited. There was this great wow factor that as I type, things are coming back to me, real time suggestions, the results are updating right away.
More than a few people actually used the word I believe, "magic," to describe what was going on. Because it was such an instantaneous response to what they wanted from the web. And I'm assuming that Google saw some of the same things. Sounds like they figured out some of the scaling and server issues because that's sort of their bread and butter, and they put this thing out there.
Jared: Now, they're probably getting four or five hits to the server per query, where they use to get just one. So just the connection load now, has to have gone up by almost an order of magnitude I would think.
Luke: Yeah. That was a challenge we ran into, right? Except you're returning things as people are doing these exploratory searches, you can be running up to ten searches for what previously was a single one when somebody hit that search button.
Jared: Right. So what's interesting to me is that the Google home page is a form. It's a very simple form. It used to be that you would type something and press one of two buttons.
Then they changed it so that you'd type something in and it would show you some completion suggestions for what you were typing in and then you'd click on that, and that would do the trick. Maybe you'd hit a button or enter or maybe you wouldn't. But it would have a definitive endpoint. Now you just start typing and magic just starts flickering all over the page.
I'm wondering if this means we're going to go back and rethink the nature of forms again or if in fact this is just sort of a natural evolution to stuff that was already happening?
Luke: Looking back and when Yahoo had that product and testing in 2005. There, like I mentioned, the reaction was really kind of one of surprise and delight. I wonder if in today's web where people have become more and more used to things sort of happening in real time. Updating as they're even on the page and getting instantaneous feedback through the interactions we've built up over the web over the past five years.
I wonder if it's becoming more of an expectation that ... of course, well, as I put something in why wouldn't it respond right away. Why wouldn't it? People are on Facebook, people are on Twitter all the time and there, it's kind of the real time feed is the norm.
If all of a sudden there's a delay between when you put something into the system and when the system responds and I don't understood what you're doing that strikes you now as an issue as opposed to a delight factor.
Jared: Definitely. I mean, things that were once delightful become expectations and eventually become just sort of basic performance stuff that aren't delightful at all unless they're missing. And it's interesting that I wonder if in fact the wow factor from this will wear off faster than it might have in 2005?
Luke: It's definitely not as dramatic, right? Because you mentioned before, Google had these suggestions happening in line. And now they're loading the results in line as you type. And is that really revolutionary to people that when they are on their Twitter page they see new tweets come in?
I think it's great, don't get me wrong. But the expectations are higher than they were five years ago.
Jared: They are! I mean, I can go to a site like Linked In and I can start to type in somebody's name. It will show me not just the names that match what I've typed so far, but their pictures and where they work and where they live. That all sort of pops up right there hanging off of the search box.
That's been there for a while now. And that's no longer delightful. That's just, OK. I get that. That works. I like it but I don't really think about it any more.
Luke: Yes. And if you don't have that experience elsewhere it starts to become frustrating, right? So all the sites that have search integrated but don't have say the technology or the design thinking to put that stuff together, all of a sudden they're a little bit at a disadvantage because people are saying, well, why don't I see results from these people on site X? I get it everywhere else.
Jared: Right, right, right. I'm thinking of an example that you've shown me where someone is typing in I think it was a user name. You typed in the letter L, I guess for Luke because you were going in that direction when you were thinking about it. And the system instantly came up with an error message. It said, "L is not a valid user name."
Luke: Not a valid email address.
Jared: That's right. Right. You hadn't quite gotten there yet. You were not done. And I'm wondering how much of that is actually going to be distracting with this Google Instant.
If I'm typing in a phrase and I just get the first word out and it matches to something completely irrelevant because it couldn't possibly know what I was shooting for in my head, is all that input suddenly distracting to me? Do I actually have to cognitively block it off to be able to continue working unimpeded to get my query out.
It's sort of like trying to talk to someone who try's to finish your sentences but doesn't have any idea where you're going and constantly says the wrong thing?
Luke: Yes. Having seen many of these kinds of automatically-complete features come to life at Yahoo, I can tell you that the details is what really matter. About not only what are you showing, but how are you showing it, the timing under what you're showing it. There's a fine line between relevance and, as you say, distraction.
I don't know what the details or the calculations that Google is running. But I do know, for example, that the search assist has stopped that. We put together, at Yahoo, we'd actually go and scan the results that came back on the page for your query. We extract the meaningful terms associated with those pages, combine them up to this more concise list, and only show you that.
So it was doing a lot of calculation on the fly, based on what you actually saw. This is in order to give you things up there that were relevant. It comes back to, when a small company or when a small site tries to do this, they really lack a lot of the resources to put that kind of intelligence in there.
The solution can end up being kind of clunky. You're describing the email thing where we're just kind of throw in error at you when you entered the first character. That's a great example of the nuance of this real time feedback systems not being thought through or not being developed or even not being designed properly.
Jared: That brings up an interesting question. So, I'm going to expect what's going to happen over the next few weeks. There's going to be a lot of conversations in the hallways of companies that do Web stuff. They are going to go something like this, snior manager or executive, they'll say, "Hey, I just played with that Google thing that they got, and that's really cool. Could we do that?"
What should do the answer be when people are sort of faced with that question of technically, we could do that. But, is it the right thing? How do you think someone should tackle that? I bet you run into that a lot when you were at Yahoo. Brilliant ideas coming left and right and having to pick out the ones that actually could work.
Luke: I think the ideas are really the easy part. It's the actual implementation and details of making that idea come to life that's hard part. Anyone of these kind of off hand comments, "We should do that" or "Why don't we do it like those guys?" Let's take a look at what that actually entails. Lots of things, sound good in words and look good on paper. But when you actually get in there and start using it, it's a whole other ball game. So. what we actually did, this is a team effort that was led by Tom Chi. They did the searches of this product over at Yahoo.
Jared: I know Tom.
Luke: Of OK/Cancel fame.
Jared: Yes. Tom's a great guy.
Luke:Yes. He was kind of the lead driver of search assist back in October of, I want to say, 2007. When Yahoo put those features out there. The way he got that project going was he actually went and recruited an engineer or two and they started working on a prototype.
They were trying to use as much real data as possible and model out the interactions as much as possible. They would kind of show it to anybody who would listen. Including some of the folks that were running the search business over there. It took a few times before that idea actually got traction.
By the time the idea got traction, Tom had worked through a lot of the data problems, the content problems and the interaction problems. This is through this process of constantly changing the prototype. So, he wasn't running around and pitching, "Oh, we should do something like Google." He was saying, hey, we've been playing with this.
Here's a bunch of things that we learned. This could enhance the experience in this and this way. Try it. When we had the SVP of search actually go and try it, the second time. The first time, in the cafeteria, he wasn't too impressed. The second time, he tried it in a meeting. His eyes sort of lit up and said, "Wow! This is actually really impressive and important." I guess I sort of went off on a little bit of tangent through my personal anecdotes here.
Jared: That's OK.
Luke: I'm just really weary of anybody only making a decision based off of words when it comes to things like this. There's the data you have your hands on and the kind of technical implementation you can actually put together, the amount of time you're going to devote to getting the design details right and the interaction is right. We're going to determine whether it's a crappy experience or it's a good experience. It looks easy but it's really hard. Right?
Jared: I would think the context makes a huge difference, too. Because looking through search terms when you have billions of pages of content to work with, it is different than, for instance, if you were to try and do this when you were an ecommerce site and you only have 400 products. Maybe the person knows what you call the product and maybe they don't.
Luke: I think the generalizable pattern of this real time search results feedback can apply multiple cases. I don't know if you've seen the Apple site's integration of type ahead search.
Jared:It's very cool. I have seen that.
Luke: Yeah. It actually shows the image of the products. It's kind of doing what you're talking about here. They have a limited product catalog. What they're doing is sort of matching those things directly to products with images and a little bit of a description and kind of the headline. They're not giving you this type ahead, you're dealing with a billion different documents, and it could mean anything. So their solution is much more constrained, if you will, but appropriate for that environment.
Jared: Right. They also go beyond products. They get in to support and other places on the site where they talk about whatever you're searching for. So they do a really good job. So this is when you go to Apple.com and you just type something in to the search box at the very top. Like iPod Nano or something like that. It does some really neat stuff.
Luke: Same design pattern and different context. You can sort of see how the two things differ based on the real data that they're working with. This is kind of, what I'm talking about here. If Apple just decided to go and copy Google's search suggest, when they put their actual catalog of 200 products in there, they'd really quickly realized, well, this is kind of stupid. Maybe we should do something different. Putting that real data into something that you're interacting with will reveal that pretty darn quickly.
Jared: No I can see that completely. So this sort of instant results when you start typing is one thing that a lot of people are experimenting with. What other sort of Web form innovations have you been seeing lately?
Luke: I think that instantaneous and in line approach to input is really showing up in lots of different places. I think, more and more companies, especially newer Web companies, are more comfortable. They're saying, hey, we're just going to leave some of these older browsers behind. Frankly, you're just going to get a bad experience there. We're not going to punish the people that have more up-to-date capable Web browsers from getting a really great experience. Just because there are some people out there that are stuck on these older products.
So, you look at a site like Quora. I don't know if you, guys, have taken a look at Quora. It's a Q&A site where anybody can ask their question, anybody can answer it. But a lot of their interactions just happen dynamically in line through quick pop up windows. Everything just slots in and appears on the page. It feels very, very responsive and quick. This is as opposed to, "OK, let's go over at this forum to ask the question. OK, let's go over this forum to provide an answer. Let's go over to this forum to flag something or follow something."
You can really get a nice sense of the contrast almost if you look at Yahoo Answers, which is a site that was created 5-6 years ago. And you look at the Quora version, their book's Q&A sites, where people can ask questions. But the Quora version is very much built on this dynamic, inline, responsive interactions.
The Yahoo version is very much, built on 5-6 years ago, and the form was sort of a page. The way you got input was you went to the page to the form, filled it in, and hit the button. It wasn't stuff that happen in context in smaller interactions, and, like I said before, in line.
Jared: So Quora, Q-U-O-R-A, they use this inline capability. I've seen that show up in a lot of different places. I think, again, that's one of these interesting innovations. Are there contexts, do you think, where that naturally lends itself more than others?
Luke: That's a good question. I think, anytime you can not take people out of the context of their task to have them contribute something that's relevant to that task. That's generally a good thing. Even if I'm dealing with something that's rather mundane or complex domain. When I have something to add to the specific thing I'm looking at, taking me off to another page, asking me to go fill out all that information, probably not a good idea, if you could capture that information right there, in context.
That said, I would hesitate to put things inline that are lengthy and almost tasks in and of themselves. I have a counterexample of this that I published on my blog not too long ago. I went to Comcast to do some form of registration, and they essentially put their entire sign-up flow inside of a dialog window that took up the whole page, and there's multiple pages in that little pop-up dynamic layer on the page.
I think, at that point, you have to sort of ask yourself, OK, well, is this going too far? It's a full page of content. In fact, it's multiple pages of full content. I can't see anything underneath it. It has nothing to do with the page I was on previously. It was a marketing page with a sign-up button. Why are they doing this stuff in-line? And in that kind of situation, I think doing things with these kind of dynamic, in-line suggestions, it's going too far if all of a sudden you're inserting an entire checkout or sign-up flow inside one of those processes.
Jared: Right. What other sort of innovations have you been seeing, both in forms themselves and in the way that we're getting input from our customers and clients and users?
Luke:I sort of have three categories of things that are happening in the desktop web. Mobile's got a whole bunch of other things going on. But on the desktop web, we're talking about rich interactions. That's a category of stuff.
Another category of things is this notion that I like to call "input can come from anywhere, " which means people are going to be spending more time off your site than on your site, no matter who you are. Even if you were Yahoo a few years ago, where they had 15 percent of all time spent online in the US, 85 percent of the time spent online was not spent on Yahoo.
That's just in the web browser. When you start thinking about email clients and chat windows and all these other places people spend an awful lot of time, there's a huge percentage of people's time where they're connected to the Internet and can provide input to you but aren't necessarily on your site and on you web forms.
So the idea that input can come from anywhere, we're seeing a lot of web applications and services take advantage of this. You can use your email client to provide input. You can use your IM client to provide input. You can use Twitter, or you can use your calendar. You can use bookmarklets or browser extensions.
The previous approach was, OK, if we want to capture information from people, we need to get them to our site, to our form, have them fill in our input fields and hit the button. Now sites like Posterous, which is a blogging service, you can write a blog post in your email, attach a photo, send it over to Posterous, and they'll essentially publish that whole thing for you, no need to ever get out of your email client. People spend countless hours, probably [laughs] per week, if not per day, inside their email inbox. So it's a great opportunity to start capturing some input there.
The third one that I was talking about in terms of innovations deals with taking advantage of work people have done elsewhere. So the first one was rich interactions. The second one was "input can come from anywhere."
And the third one is, people have invested a lot of time in providing information to services like Facebook or LinkedIn or Delicious. All these services have built out platforms that allow you, as a web application or a website, to get access to that information.
So, instead of somebody comes to a new website, they're asked to fill in a form about their profile, they're asked to fill in a photo, they're asked to put in a description, their name, their location, their gender, all these sorts of things, what they can literally do is hit a button that says "Use my Facebook account" and all that's completed for them, and more. They can get access to who they know, and they can make connections and all these sorts of things. Which, from a user-experience perspective, I think, is great, because it kills this ghost-town approach to new products.
So I'm pretty excited about how especially new things can take advantage of that stuff, because, ultimately, if you do these things correctly and you take advantage of them appropriately, you can kind of eliminate sign-up forms and profile-creation forms and "go find your friends" kind of forms, and instead just have people get a really instant-on user experience with a new service.
Jared: I think that that's got, really, a lot of potential. We were with a client. It was a client that has a product that people use to collaborate on projects. We went out and visited all these customers. And one of the things we noticed was that nobody was using the capability to put in your picture or to populate with your contact information, even though this was basically about professional collaboration for remote teams, where, when you have a picture and when you have the contact information there, it makes that collaboration more rich.
But nobody was filling out that initial stuff. They were just sort of jumping in and working. The system had a capability to do conversations, for example, between the various parties. And you look at the conversations, and they'd have those fake avatars that they use as sort of placeholders till you get your picture, and every conversation would have this shadowed, shoulders-and-head avatar. Whether it was a man or a woman, whether they were in India or Africa or New Jersey, they'd all have the same, exact image.
So none of that personal effect that they were trying to get was coming through, but these images dominated the screen and people learned to ignore them. The whole thing was really interesting. But they came down to the fact that uploading your picture is a real pain in the ass. And chances are, they've already done that some number of times. And if you can leverage that information, if you can go capture it from their Twitter or Facebook account: boom, you're done.
Luke: And boom, you can also grab the connections that they would have, based on their address book on Yahoo or Gmail or Facebook.
The specs I've seen, and these may be overly generalized, is that asking people to upload a profile photo has something like a 40-percent drop-off rate. So you're losing 40 percent of people as soon as you say, "Hey, give us a photo." And like you say, people have already done this multiple times, right? If you can take advantage of the fact that they have done work elsewhere, think, that's a great thing to try and do.
Jared: Absolutely. And I've even got a photo of me that I use in at least one of my things that I really like, but I have no clue where the original file is. I couldn't even use it again if I wanted to. I don't know where the hell it is. So I think that's part of where this impedance comes from, this resistance.
Luke: Here's actually a great example, to go to the recent news again. Apple launched this iTunes Store social network called Ping. When you go and create your profile on Ping, they ask you for a photo. And that photo has to be approved by Apple. So I was actually getting invitations from people who work at Apple without profile photos.
Even the people who are working at the company can't get their own picture on the profile. In fact, I actually went through the process of picking a profile photo from my desktop and uploading it, and I still don't have a profile photo on that site.
We can actually talk a lot about some lessons from Ping, I think, in terms of what actually makes an effective first-time user experience. I don't know if that's a direction you want to go in.
Jared: Yeah. No, I think that's interesting. Let's talk here a little bit about Ping, since you brought it up. Because here we have Apple, a company that's really known for their design. I mean, it's impossible to have a conversation about design without Apple sneaking in someway or other. The joke running amongst those of us who gives speeches a lot is, can you actually give a talk and never mention Apple? It turns out it's really, really hard to do.
And yet, Ping has not been the smashing success that it could have been given Apple's view of design and Apple's attention to detail. I'm curious what your thoughts are in terms of what's not working about it.
Luke: Let me start by saying I'm kind of an Apply fan boy. I use Apple products left and right. I actually think Ping is potentially the worst product they shipped since Steve Jobs came back.
Jared: Which comeback? The liver transplant comeback or the job transplant comeback?
Luke: The original. The original one.
Luke: The one before he went to play for the Vikings.
Jared: OK. [laughs]
Luke: And the reason I say that is Apple, I've always used as an example of great start up experiences. When you purchase an Apple computer, and you take it out of the packaging they've thought through how are you going to remove the labels. They've thought through when you start the computer, it actually comes up and says "Let's take a picture," and the camera's active and you're actually looking at it. You just hit a single button and it takes a picture of you for your account, right?
So it's this very personalized, very thought out unpackaging experience. And in fact, people like unpacking Apple products so much that they will take photos of every single step along the process and post it on the web and share it. That care and that attention to the out-of-box experience on Apple products is almost unparalleled to some extent, right? You don't see that much rabid devotion to opening boxes from very few other companies.
Jared: Yeah, yeah. No one gets really excited about opening up the box to their new air conditioner.
Luke: Right. And now here's a digital product. And I think the unpack aging experience there, the out-of-box experience for Ping is just really, really poor. So let me give you just a couple things that I think cause that.
One is your Ping profile seems to be totally ignorant of anything you've ever done in iTunes. It doesn't know what your favorite genres are. It doesn't know what music you listen to, right? It starts with this completely blank slate and uses none of the information. If you're an iPod user, you've been providing iTunes for years by rating and listening to and adding songs. Complete blank slate.
Then to top it off, you try to kind of put in your photo. Well, that has to get approved by Apple. OK, then you try and put in your name. Well, it turns out your name actually has to be associated with the billing account you have in iTunes. So you can't put in a nickname, it has to be the name that's on your credit card. That's how I refer to myself, right? [laughs]
Luke: My credit card name is my name, damn it.
Luke: Then it asks you to pick genres of music. And you can pick up to three. So if you happen to like more than three genres of music, you're out of luck. If you want to go and share music you actually like, you have to go to the iTunes Store. And there, actually write a review or like a song or record. You can't do it inside of your music library, which is where you spend all your time. It's all these things that really start to add up, and I think create a very poor experience, like I said, out of the box.
I was actually super-excited when I heard about Ping. I said "Wow. I've been building up an iTunes library for years and it knows more about what I like in music and what I listen to than anything else, probably in the world." Including my wife, right? So I'm like "This is great. I'll be able to share and find out what other people like and listen to and it'll be fantastic."
And it was just a really huge disappointment that it turned out to be a social network for the iTunes Store. Not for iTunes.
Jared: It's just crazy that they did not pay attention to these details that they are so good at paying attention to.
Luke: We can theorize all we want about why it matters, but I think you know, culturally Apple is a pretty insular culture. Are very centrally focused and some things that are, by their nature, outwardly driven like social network applications, all about sharing and showing off and building better relationships with people other than yourself. We could theorize that maybe some of this stuff just isn't in their DNA. But, I don't know. It'll be interesting to see what they do with the V2.
Jared: Yes. Yes. For everything that Apple gets credit for, they do have a way of looking at where they've succeeded and where they haven't and figuring out new ways to approach the problem that are new and novel. And sometimes it takes a little while. It's not the same sort of phenomena that you see with Microsoft, where you have to sort of wait till version three of the product before there's actually a functionality that you'll care about. But there is something to it.
They do have a way of sort of incrementally rethinking ideas. You see them going through that now with Apple TV. They certainly did it with iTunes in the early days. There is that sort of culture in how they approach things that makes it interesting.
Luke: Yeah well definitely, and Apple TV I think is almost a counter-example here because during the latest announcements, Steve Jobs got up on stage and basically listed out, like just super spot-on, I thought, market insights. You know, he's like "We've been having this product on the market for a while and here's what we learned that people actually want." And he just went down this bulleted list where I was like "Uh-huh, uh-huh, uh-huh, uh-huh." [laughs]
So really deep market insights about how actually people want to do this. And it feels to me for Ping, all that stuff was really missing. Now maybe that means they're just testing this out and trying to get those market insights and they'll make the changes. Or it means they just really didn't think about those sorts of things because this is a new area for them and it's not something they have a lot of experience or domain expertise with.
Jared: Right. Yeah, I completely get that.
Luke: That said, I used this analogy recently. I think a lot of people are going to get into the system, but the way I describe it, is maybe too harsh. I'm sorry, Steve. I still love your products. But there's 160 million iTunes users, and within 48 hours they announce that one million of them were on Ping. So the metaphor I used there was, well if 160 million people are driving down a road, and I put a pile of dog poo on that road, chances are one million will drive through it.
Jared: Yeah. I think that's pretty much it.
Luke: That's a metaphor I think you can use in many places where people have a really big customer base. And they make a change to their product and they go and flash all these big numbers. You know, "Two million people adopted the new Facebook home page." Well, they got 500 million people, and you just kind of put it there. Just the numbers alone aren't really an indicator of whether or not you got a great product.
Jared: No, I mean, yeah. One million sounds really big, but not compared to 160 million.
Luke: Right. And 160 million where this is the thing that's on the front page of iTunes for everybody that goes and looks.
Luke: Anyway, rant off.
Jared: There you go. On a different note, you're going to be coming to Boston in just a few weeks for the User Interface 15 Conference to speak to your adoring fans. And you're going to do a full day workshop there. Of the stuff we talked about today, what sort of things are you going to get into there?
Luke: I will definitely get into these kind of first time setup experiences, right? How we make the unpackaging experience not a series of forms that are frustrating and that limit you, but actually that really get you…
I like the term "gradual engagement." They let you see what the service has to offer and why you should care and get you up and running through a series of just lightweight interactions. That scenario I will definitely be spending a lot of time on.
I'll talk about a bunch of research we've done with Rich Interactions. In fact, I got an article coming out on a list of parts towards the end of September that talks about inline accordion forms and what we learned testing those. So we'll talk about stuff like that. We'll talk about our input on Google devices, and a whole lot more.
Jared:Well that sounds excellent. Well Luke, looking forward to seeing you in November. And I want to encourage other people who are interested in Luke's workshop to check it out at the User Interface 15 Conference website, uiconf.com. Thank you very much for sharing all your thoughts with us today.
Luke: Thank you, thank you.
Jared: I want to thank everybody listening, for once again encouraging our behavior. Take care, we'll talk to you later. So long.