Nate Koechly and Matthew Darby: YouTube's UX Director and Director of PM on redesigning one of the world's most-used apps

19:13Matthew Darby:Yes. Yeah, critical user journeys, customer user journeys. Yeah, like what's the flow through the app? Where's Main Street? Where are the cul-de-sacs? A bunch of different metaphors you can layer on. But yeah, if you're trying to design for the user first and foremost, you have to understand what journey they're on, what their motivations are, what their context is. And I think that's excitingly so much more powerful. There's so much more potential there sort of emerging now, it feels like, as we can continue to tailor the experience for individuals. But yeah, we do focus a lot on that. And I think the fundamental viewer user journey is, like I said before, immerse, dive deeper, move on, find content.

19:51Matthew Darby:And then there's subsets of those. The list gets really long really quickly, but it also can get that clear. But yeah, we talk a lot about user journeys and the intent of the user at a given moment and how to help them achieve their goals.

20:02Nate Koechly:Yeah, we spend a lot of time thinking about those journeys, these three goals of watching. You want to be immersed in the video, and then you want to dive deeper and find out more stuff about the video. we spent several years investing in the video player and we actually launched like a ton of features like video zoom where you can zoom into the player chapters, a feature we call speed master where you can hold down on the player and have it play at double speed like extra speeds for premium users that whole suite of features that work together as a system we spent many years like sweating the details on those and that's really in service of this be immersed user journey that you really want to find exactly the right point in the video Heat Seeker, which is like the little heat map that you get as you sort of scroll around and you can find interesting points in the video that other people are watching.

20:47Nate Koechly:And that was a multi-year journey that was just to make that one critical user journey that much better. My favorite feature in that suite is we call it internally Double Tap to Seek, where you like double tap on either side of the player, so like skip forwards and backwards five seconds. We launched that several years ago now, actually, and it's been copied a lot. I think most good video players have it. But A, I think it's a killer feature because it was really difficult to seek inside the video before. It's super tedious to use the scrubber bar, which is the little bar at the bottom. And B, it was actually a nice demonstration of when the bottom-up culture is a real benefit to doing product development.

21:22Nate Koechly:Because it was one of our designers, she came up with the original idea for using the double tap on either side. And the insight was that it was on different sides at different speeds. Seems obvious in retrospect. But we were actually very constrained. We didn't want to use swipe because we had another idea for swipe where we already had tap. So she had these kind of leftover gestures and managed to repurpose them. And then somebody else came up. There's another designer on the team that came up with, well, there's a product manager actually who came up with the idea that you should be able to mash, you know, and so that you can just keep tapping and it would just advance a long way.

21:51Nate Koechly:And then another designer came up with the idea that you'd have to have, like, the indicator to show you how many seconds you're advancing. And then there was an engineer that reached out to me as we were getting near launched. And he said, how close does it have to be? Does it have to be exactly 10 seconds? And I was like, well, not necessarily. it's more you're trying to get back to what was just said, like I missed something, what was that moment that was just said? And so he introduced this idea that we would snap to the nearest keyframe, and this is a detail about how video encoding works.

22:18Nate Koechly:And every 10 seconds we send a keyframe, and the keyframes are just a full representation of the video frame, and they're much faster for the client to decode and put on the screen, whereas the intermediate frames you have to do some math to figure out what the differences are. And so it just speeded it up by a few milliseconds, It just made it a little tiny bit snappier by snapping to the keyframe. And actually, if you use the feature, you'll find that you sometimes jump five seconds, sometimes you'll jump 15 seconds. And so anyway, it was a nice synthesis of all of these different disciplines and different ideas from people across the team.

22:50Nate Koechly:And it went on to be a gangbusters feature that gets used a ton in the app. What are the challenges of redesigning and launching new features on a platform that is servicing billions of people every day? It's an enormous challenge. Every feature that we launch, we launch in an A-B test. I mean, it's the only way that you can launch at scale. So even the smallest thing, even if it's a bug fix sometimes, we'll just test it in an A-B test to see what happens. Sometimes we even do things like we have the code available for the feature, and we'll just launch the code without even switching the feature and run an A-B test on that just to see what effect putting the code in the client had.

23:28Nate Koechly:We're always trying to measure the effect on the ecosystem. And also there's a lot of tragedy of the commons dynamics that can work out if you don't pay attention to them. You can slowly, slowly make the crash rate get worse and worse and worse with lots and lots of features. So lots of system health metrics, for instance, we track in the client to make sure that the client's still in a healthy state. And yes, there's like thousands and thousands of metrics, way, way too many metrics. And it's putting a new, something brand new into that dynamic nature of the ecosystem and seeing how things change inside the ecosystem.

24:01Nate Koechly:Quite often there are these zero-sum trades. If you get some engagement here, it comes at the expense of some other surface over here. And so it's almost always a pretty delicate balance. And there's a lot of analysis that we do and a lot of metrics that we look at.

24:17Nate Koechly:We'll return to the conversation after this quick break.

24:24Nate Koechly:Design Better is supported by Masterclass. I've been fascinated by the research from Cornell University showing that we get far more lasting satisfaction from experiences than from material things. They call it the experiential advantage. The idea is that while a new gadget eventually becomes just another thing, the wisdom and skills you gain through learning actually become part of your identity. That's why I've made Masterclass such a consistent part of my routine. It's an investment and an experience that pays dividends for years. And I'm not alone in that. 88 % of surveyed members feel that Masterclass has made a positive impact on their lives.

25:04Nate Koechly:I use the platform to stay curious across the board. I've been using James Clear's class to refine my daily habits. And I'm looking forward to Michael Lewis's class to find better ways to tell the stories that we share right here on Design Better. Whether I'm in the car using audio mode or at my desk, it's a way to ensure I'm actively becoming a better version of myself. With Masterclass, you can learn from the best to become your best. With plans starting at just$10 a month billed annually, you get unlimited access to over 200 classes taught by the world's top minds in design, business, and leadership.

25:42Nate Koechly:Plus, there's no risk. Every new membership comes with a 30-day money-back guarantee. Right now, our listeners get an additional 15 % off any annual membership at Masterclass.com slash Design Better. That's 15 % off at Masterclass.com slash Design Better. Masterclass.com slash Design Better. Design Better is supported by Laundry Sauce. I have two teenage boys. Don't tell them I said this, but they stink. My wife and I have been teaching them to do their own laundry each weekend, but sometimes they totally botch the process. They put in too much detergent, not enough. So we ended up getting them laundry sauce to lower the learning curve.

26:27Nate Koechly:Laundry sauce is like a bath bomb for your laundry. You drop in a high-performance detergent pod in with your dirtiest of clothes, and they come out smelling and looking fresh. Each pod is enriched with a bioenzyme blend that tackles the toughest stains and revives fabrics for a brilliant clean. And I love the scents of Laundry Sauce that they've created with the help of the world's top perfumers. They have scents like French saffron, Australian sandalwood, Egyptian rose, Himalayan cashmere, Italian bergamot. It is a treat to get dressed in the morning smelling this good. And I love hugging my kids when they smell like bergamot and sandalwood.

27:11Nate Koechly:Sure beats the alternative. Make laundry day the best day of the week. Get 20 % off your entire order at LaundrySauce.com slash Design Better. Use the code Design Better at checkout to get 20 % off. LaundrySauce.com slash Design Better.

27:31Nate Koechly:And now, back to the show.

27:38Matthew Darby:So we're just about half an hour in, and we have not yet mentioned AI, which is probably refreshing for a lot of people. But I'll break our streak there and just ask, curious mostly about how AI is affecting search, and since YouTube's the second largest search platform at the moment, it's obviously disrupting all kinds of search. But I wonder how you're thinking about it from a product standpoint and a discovery standpoint for your users, and what you kind of see that trajectory like? Yeah, super interesting times. I zoom out a little bit first and think about the progression of sort of the internet eras through them all.

28:15Matthew Darby:It's about getting more capabilities for users and about technology that gets you closer to true user intent and true understanding of what a user's trying to accomplish. And so specifically, language really reduces that gap. You know, in the Web 1.0 days, everything was done through form elements. And then in Web 2.0, you could directly manipulate things. And then in mobile, you were freed from tables and grid-based layouts. And you could have sort of bespoke designs that more closely matched the user's expectations and mental models. And I think now with language and with presumably reduced cost of sort of interface development over time, it enables us to have a better picture of what the user's actually trying to accomplish.

28:57Matthew Darby:And then hopefully the product design capabilities to meet that even more closely. The world learned how to talk in a pretty optimized way with short queries. And now that language is more accessible to more people, we're seeing more expressive queries and people bringing new questions forward to us. So I think that's really interesting. What we can do with that is we'll see. But I think when users have higher expectations of us, it just gives us new moments to meet. And so maybe when you're trying to find the best thing to do on a trip or the best way to fix this thing. It just lets us be more efficient, we hope, where we can identify more clearly what the users want and then find that information in new places.

29:41Matthew Darby:So it might not just be from a single video. We might be able to bring a set of videos to a user in a new way and explain it. So that's one thought. The second thought maybe that goes with it is, what is video good at and how can we make that work better? Sometimes I think about the history of text and all the tools that have developed around text and all the things you can do with text. You can snip it and you can skim it and you can morph it in different ways that are more challenging today with video. And so what is a good snippet? What is a good narration? How do you help explain the context around the videos to a user in new ways?

30:16Matthew Darby:And so I think it's pretty early days, but to have so many new raw materials available, I think it's pretty exciting. I think as a result, we also do see users bringing more queries to YouTube and more query diversity and users come to YouTube for entertainment, but they also come for a lot of information and a lot of connection. And we're starting to see that. And I think with more expressive design and with the ability to sort of craft a better response to each query, we're super excited about it.

30:44Nate Koechly:How are shorts changing the way you think about the product?

30:49Matthew Darby:Well, one thing they've told us is that users value for some journeys, efficiency is really important. And getting to the essence of the answer or getting to the essence of the joke or the story or the insight. And so I think shorts and short-form video in general have increased our thinking around content efficiency and getting people to the heart of what they're looking for even faster. So that's one dimension. I think another dimension is the amount of participants which continue to lower the bar to be a creator, to be creative, to put something out in the world. It can be a big project to put together a long-form video.

31:28Matthew Darby:And so more people are getting to experiment with new ways to be creative and to communicate thanks to short-form video. So I think that's a dynamic that will play out for a long time, presumably.

31:40Nate Koechly:Right now, they're sort of squished together where you get a block of shorts and then a block of long-form videos. Do you think about those as different? Is that a fork in the road where maybe there's two different pathways for users?

31:52Matthew Darby:Well, I think there's a lot of different user intents. And so sometimes you're looking for a specific video, and that's sort of one path. And other times you're really looking for a set of videos or just for a scrolling sort of content experience itself more than an individual piece of content, particularly. So I think that is a pretty distinct, or it can be a pretty distinct sort of user intention. But it's not the only one. I mean, sometimes people want to lean in, sometimes they want to lean out. Sometimes they're looking to be really active and guide a research session and poke around and curate and find a lot of things.

32:26Matthew Darby:And other times, they're looking for a different experience. So I don't think that's the only fork in the road, but I do think that is a distinct user journey at times.

32:35Nate Koechly:We have done quite a bit of experimentation with mixed feeds as well, where we mix what we call VODs, non-shorts and shorts together. And they can actually, in some circumstances, perform quite well. It's definitely something we're looking at. it's not completely straightforward, but definitely something that we're working on and we're aware of.

32:53Matthew Darby:How do you think about prioritizing the different user groups that make use of the platform? Obviously, there's probably some, definitely some business decisions that are made there,

33:02Nate Koechly:but also just from your own perspective as product and design leaders, what does that prioritization look like? I was thinking about this actually when Aaron was asking earlier about all of the different sort of niche uses and niche user journeys. And Nate and my teams, we really think of the broad set of users. So we're trying to build features that are just generally useful across the platform. And most of the teams that are adjacent to us are trying to keep the health of the ecosystem at large in really good shape. And that allows these different communities to flourish, as Nate was saying.

33:34Nate Koechly:There's a team that just looks at podcasts and they're building podcast features. And yes, we have to figure out how to sort of fit that in such that it feels connected and integrated into the app. That can be a challenge. Same with music features, same with premium members have their own sets of features. So that's an ongoing discussion. It's not the most straightforward thing, actually, to sort of bring those in at the right times. But that's what we aspire to do.

33:56Matthew Darby:I guess I would just add that, well, I mean, there's a lot of people on YouTube. And so we look at lots of slices of the user group. And we do, you know, focused user research into different segments, you know, from time to time. And a lot of that is just to track the health and to try to identify any blind spots or any areas where we're not serving people as well as we could. It is important, I think, to be diligent and look at different segments and look at different sort of usage behaviors and really use that as a way to sort of triangulate what's working and what isn't and where sort of the opportunity is.

34:27Matthew Darby:It's less about prioritizing specific groups and it's more about, I think, prioritizing the aggregate health. It's also interesting to see different segments teach us different things. And so, you know, the way podcast consumers consume content is a little bit different than the way sports people do. And all of that can be inspiring. Maybe one more interesting dynamic, I think, is looking around the world. And the Internet and YouTube plays a different role in different markets. In the same way we saw sort of mobile have different characteristics on different continents. and different, you know, it's sort of when different markets matured and what technology was sort of most prevalent right then.

35:04Matthew Darby:So it is interesting, you know, we see interesting information-seeking journeys in India that we don't see as much in some other countries, for example. And so that just reminds us to just to keep looking and to keep trying to understand more. At this scale, there's just so much diversity that on some sense it aggregates out or averages out, but that's a risk. And so it's important to keep drilling in everywhere and understanding.

35:27Nate Koechly:Yeah, one thing I'd say as well that might be surprising is in general, like Nate and my teams, I mean, we measure the business impact, but that's not a goal for our teams. Like we don't have a revenue goal. We're just thinking about how can we deliver value to viewers? And that's very intentional. There's a separation of church and state. Now we work with the ads teams quite closely and mostly our interests are like totally aligned, right? We want to do the right thing for viewers. But in general, we want to just deliver as much value for viewers that creates a healthy ecosystem from the viewer point of view.

35:56Nate Koechly:And that makes a bigger opportunity for the teams to monetize in different ways. I'm curious how you think about the algorithm because the algorithm in YouTube is quite good. I discover so many wonderful things through the algorithm. But there are also times where it's like I'm stuck in a loop and I want to kind of break out of that. There's a button up at the top, new to you. And I've often thought like, I wish there was like, I could dial it up to 11 where I could really just scramble the whole thing. How do you think about the algorithm and the experience it creates for people?

36:32Matthew Darby:One way to think about it is that the algorithm sort of is the audience. Its job, when it does it well, is to represent what that user wants. And, you know, I think we believe that there's content on YouTube for everybody and it's our job to help find it. So at its basic level, it's that. But it's interesting to try to design for these probabilistic and things that aren't sort of deterministic. It's not just one in, one out. And so, one, what are all the right signals? And what are we sort of optimizing for? You know, we want long-term satisfaction. We want long-term value, long-term satisfaction.

37:07Matthew Darby:And so our understanding of that has changed a lot over the years. Early on, we were measuring clicks and then watch time and then satisfied watch time as a subset of overall watch time. And we keep trying to figure that out. And then we try to figure out the things that detract from that. So content that might be totally fine and beloved by one user might be off-putting for another user. So we have to figure out how to model things like that. And then users have other characteristics. They want freshness. They want novelty. They want tried and true. And so how do you manage all of those things together when you really just have like a flat list to present by default at the end?

37:46Matthew Darby:Can I give you a use case? Yeah, please.

37:48Nate Koechly:So lots of people use YouTube to do research, especially on products, researching a car that you might buy or some other thing. And then you've made that purchase, you are no longer researching, and yet you just get the endless feed of, here are all the cars. Those sorts of scenarios where what I wanted has changed and now it's totally different. Are those parts of your conversation around the algorithm?

38:13Matthew Darby:That's definitely part of the conversation and the time of day and the device you're on. You probably want a slightly different piece of content on Friday night on your couch in the living room than you do on the road. And it's really tricky to identify those sessions. And so it's an interesting blend between explicit and implicit. How much can we just predict and how carefully can we pay attention and respond versus when is the right answer just to ask the user and to listen to that. We also see the dynamic where what people say isn't always what they do. And so you have to balance those two things out.

38:47Matthew Darby:But in general, we're trying to give more agency and more responsiveness based on increasingly nuanced understandings of what satisfies a particular journey. And so in the case of music, maybe the priority is to give you five songs in a row without you having to change the channel. Whereas that wouldn't be the goal at all on like a research journey where we want to give you more diversity and let you sort of hone in on it. And so it's tricky to come up with UIs that are simple enough to get out of the way, but still enable you to sort of get what you want. And we have a long way to go. That's an endless quest.

39:23Nate Koechly:That's something we hear from users all the time, actually, is that they would like the ability to tune their algorithm more precisely. It is something that we're working on in sort of two ways, giving viewers insight into what we know about them, the inputs into the algorithm, the things that we know that they're interested in. We are also exploring ways where you could tune the inputs. It's really not straightforward. As Nate said, quite often viewers, they want one thing and actually you can show empirically that that's actually bad for their usage on the platform. So that's a big dynamic that we have to navigate.

39:55Nate Koechly:But it's definitely something that we hear. It's definitely something that we're interested in and working on.

39:59Matthew Darby:And it's an area, I think, where we have renewed interest. The sensitivity to the signals, like the things that we can measure, is increasing. But more importantly, I hope, and we're starting to see some early signs, that user expectations are changing. And so the ability to tell us how to change your feed with natural language instead of with a bunch of bells and whistles is interesting, certainly for some users at least. And so it does feel like the tools we have to approach this problem are changing really rapidly right now.

40:30Nate Koechly:Do you guys remember that episode of Seinfeld where Kramer is pretending to be movie phone? And George and Jerry, they would call in and try to get answers. And finally, it was just like, why don't you just tell me what you want?

40:44Matthew Darby:That's basically what you're describing. That's right. Yeah, yeah. And it's interesting, you know, users do so much on YouTube that it does reflect them in some ways or can be a nice reflection. And so I think that's an interesting dynamic also. What's your favorite YouTube channel that you've run into lately? And if it's not a full channel, it can just be a video or podcast or music too.

41:05Nate Koechly:Mark Lewis, actually. It's very obscure. He's like a middle-aged guy doing fitness stuff and building his home gym out in the country in England. And so, I don't know, has a lot of resonances for a middle-aged guy who was also working on his fitness. And he has a very dry sense of humor. so I enjoy his content.

41:21Matthew Darby:Number one is animagraphs. This guy does 3D modeling, like he just did the Golden Gate Bridge. He does like three or four videos a year, works on them for months and months and months, and then models it and then flies around and describes it all. And so I drive across the Golden Gate Bridge a lot. I learned a ton. It's like an hour and a half video, and then he did another hour behind the scenes, and at the end of it, he just looks at the camera and just thanks the audience for enabling him to be so passionate about something and find these so many people that love the same thing. So to have a platform where that level of passion and craft and nicheness can find such success.

41:58Matthew Darby:And that's a different way of YouTubing than you think of the weekly vlog or whatever. And so I love that one. One other one is called The Lookout. For those in Northern California, it's a person in the forest service that knows everything about forest fires. And it's an example to me of, again, somebody's depth of expertise being there for the world. And also just the quality of the information. Like I've learned more from them about the climate around here than I have from any more traditional source. Case after case, when you give people a voice, the things they say are just inspiring. And those are two, I think, just sort of only on YouTube examples that I watch quite a bit.

42:36Nate Koechly:And you sound so much more wholesome and educational than mine. You're going to have to give me those channels, Nate.

42:42Matthew Darby:There's more where that came from. There's something out there for everybody.

42:45Nate Koechly:I want to add one to your list. Have you guys seen Kurzgesagt? Yeah, yeah, that's a favorite, yeah. Yeah, I love that one. So good. Yeah, so well done. Yeah. That's the difficult thing, you know, having cast my mind back to sort of 2008 and 2009, to imagine that a channel like that could exist with that level of polish and production value. It's astonishing, really. Well, Nate and Matt, congratulations on the release. Thank you for making YouTube. It's just a wonderful resource. I love it so much. That's fine. It was just me and Nate, by the way, I hasten to add. So you're definitely thanking the right guys.

43:17Nate Koechly:Yeah, but it's nice when people say that, actually. One of the things that has made me stick around at YouTube, as well as it being a consumer product that I love, and it's such a privilege to work on a consumer product that you love yourself. But yes, it's common when people ask you what you do. It's very common for people to say, I love YouTube and get a lot out of it. And that's pretty special, actually. That's great. Thanks for being on Design Better. Thanks for having us.

43:43Nate Koechly:This episode was produced by Eli Woolery and me, Aaron Walter, with engineering and production support from Brian Paik of Pacific Audio. If you found this episode useful, we hope that you'll leave us a review on Apple Podcasts, Spotify, or wherever you listen to finer shows. Or simply drop a link to the show in your team's Slack channel, designbetterpodcast.com. It'll really help others discover the show. Until next time.