Katarina Batina - Making big bets with design

00:00:00Initial discussions around some of theprototypes that we had created weremaking a lot of product managers anddata scientists a little nervous sayinglike wow this really thwarts what weunderstand best breast practices to looklike. People come to do great craft workbut they leave really great productthinkers and [music] that to me is oneof the most enduring skills in this era.Think about a place about a formerproject or a project you're in now whereyou feel like you're being blocked andit inherently feels like a differentcraft's responsibility that's theblocker and say actually there's nothingstopping me from like jumping the pondhere and like embedding myself in thatcraft or in that skill set or learningit so that I can come to my ownconclusions about this thing.>> Welcome to Dive Club. My name is Rid andthis is where designers never stoplearning. Today's episode is all aboutwhat it takes to make big bets withdesign because we get to talk toKatarina Betatina, who is the designdirector behind the shop app, which inmy opinion is one of the premiumconsumer experiences available today.They did things pretty differently andit worked. So, we're going to get intothe weeds about how they explore newideas, how designers are making the mostout of AI, what it takes to consistentlyhit this bar for craft, and a

00:01:15lot more.So, let's dive right in.>> Shop I talk about as a love child of tworeally cool internal projects atShopify. [music]One was called Arrive for the OGs in thecrowd. like they'll know that that waskind of the origin story of the app andit was to solve a merchant problem.Actually getting a ton of supporttickets into their local cues saying heywhere's my package and us as Shopifybeing extremely merchant wanting to saylike hey that's a problem we can solve.The other connection point for Shop wassomething called Shopify Pay at thetime, which was this sneaky littlecheckbox that just said, "Hey, next timeyou come through a Shopify backed onlinestore with the Shopify checkout, yourstuff will be here." Six or seven yearsago now. Um, brought those two productstogether underneath the banner of shop.What an audacious brand name, by theway. Like, how who has the guts to dothat besides Shopify? I I love it so somuch. It creates some funny brandingproblems sometimes because it's like theverb, the noun, the app, the everything.Bringing those things together, therewas this third pillar of like we've gotthis payments network thing. We'rebecoming a trusted source for people astheir kind of shopping identity. Andthen we have package tracking which issuch a beloved feature even today. Butwe knew that there was this hugeopportunity obviously to become afirstparty

00:02:30platform that merchants couldreally trust. Toby has always had anincredibly strong vision for what amerchant first, merchant-led shoppingdestination can look like that buyerscan fall in love with. I really creditlike where we are now in our vision tothat really audacious prompt. You know,most people who work in commerce orreally any modern consumer app are soused to being like they found their oneflywheel and they are just milking it.super datadriven, extremely microoptimization focused. And what's beencrazy about our journey in shop is we'vebeen able to paint with very broadstrokes against very bold ideas. And oneof those bold ideas has been taking acompletely different tact with thebrowsing experience, saying things like,"We want this to be a place that youcome when you're bored and you want toshop." You think about shopping kind ofin a more romantic way in the offlineworld. Shopping is a form ofentertainment. It's a way to engage withculture and so many of the brands thatlaunch on Shopify represent the best ofwhat that world looks like, right? Andwhen I think about the interface andwhen we think about the experience inshop, we think about on purpose how todo things a little differently. How dowe celebrate those brand stories? How dowe bring them to life? And how do wemake them extremely personalized and

00:03:46engaging? And through severaliterations, we've arrived at what wecall today our super feed. There waslike feed version one. The super feedbecame something that was extremelycontentdriven and focused and highlypersonalized and we're working on ournext iteration right now. [music]>> Real quick message and then we can jumpback into it. Okay, so get this.Jitter's latest release lets you buildyour own creative tools inside of Jitterto make custom animations and effects.Just describe anything you want and thenJitter will give you custom parametersthat you can use to tune every singledetail, whether it's 3D control,dithering, or even an old VHS filter.And then once you get the effect whereyou want it, you can save it, share itwith your team, and easily reuse itacross projects. And it really is assimple as just typing out your ideas,which is perfect for someone like methat doesn't have this rich backgroundin animation. And sometimes I even liketo have Claude help me with the prompts.This feature has blown the roof off ofmy creativity, and I cannot [music] waitto see what you make. Just head todive.club/jitterto get started. When you're prototypingwith AI, if you're spending time [music]recreating styles, components, andlayouts from your actual product, I gotnews for you. You're probably

00:05:01wastingtime and definitely not using the righttool [music] either. That's why I loverecommending Desen to Teams. In oneclick, [music] you and everyone on yourteam can prototype directly in yourcodebase without [music] ever opening anIDE. Destin extracts your designlanguage and gives you the perfectsandbox to explore without any of thetechnical hurdles. It's a pretty bigdeal and you can connect your codebaseand start prototyping today. Just headto [music] dive.club/desen.That's dsn.Now on to the episode. I want to talk alittle bit about those iterationsbecause you talked about shopping asentertainment and I feel that like Iknow that I've said this to you offrecord. I'm going to say it again justso that it is on record. like the shopapp is the only app that I will browsewithout clear intent which is a littlebit dangerous you know it's like the Ican't fall asleep app for me which>> I know I feel get blamed a lot forpeople shopping more recently>> I know yeah my wife doesn't love itwe'll put it that way but I I want toknow like what are some of thoseiterations that you've made like how didyou get this to the point where I amjust kind of laying in bed scrollingthrough this app>> absolutely well one of the biggesttension points was around the thethought of density Right. So if we thinkabout what makes modern consumptioninterfaces really

00:06:16successful, it'sdensity of information, collectingintent, you know, are people likingproducts, how many products can you getinto one view scroll, you know, all ofthese best practices that have led to alot of the commerce interfaces we're allpretty used to, but we really wanted tomake a quite a different decision tosay, how do we make this feel more likewindow shopping, right? So each viewportprovides an opportunity to kind ofportal into a particular brand story, aparticular product recommendation,whatever it may be. And trust me,initial discussions around some of theprototypes that we had created weremaking a lot of product managers anddata scientists a little nervous sayinglike, wow, this really thwarts um whatwe understand best practices to looklike. And I feel truly lucky to work inan organization that's just like willingto build the thing and take the bet. Andcertainly in our first couple ofiterations of this, it was not at all aslam dunk. Our recommendations algorithmreally needed to be on point and able toserve that kind of consumption model. Ithink one of the other huge unlocks hasbeen how we've been able to verythoughtfully and gently deploy AI tobring more of this content to life. Ourfirst couple of iterations in, you know,what we call today the super feed werevery static. It's great to have theserich cards, but ultimately they can feel

00:07:31a little flat at times. And one of thethings we've been working on in shippingmore recently is bringing a little bitof video generation to just create alittle bit of life and interest in thefeed. And that's gone a really long way.The other thing that we've recentlyintroduced is merchant post. Somerchants actually be able to syndicatecontent from the other services likeInstagram and Tik Tok where they'realready producing and showing content.Bringing that into a commerce nativeenvironment where they can tag theproducts. those products can bepurchased right there and it's reallyreally bringing our feed to life whichis awesome.>> You talked about big bets. One of themfor sure like the window shopping singlemerchant at a time that feels obvious asone. Are there other ones that come tomind that we could maybe talk about anduse as a window into your process andhow you even evaluate these ideas.>> Another huge bet [laughter] that we'veslowly been teasing out and it's kind ofdownstream from the feed is really themanifestation of what we call the shopstore. So today, the swan song of anymerchants's experience using Shopify isgetting to build their online store. Itis their piece of real estate on theinternet to really tell their story. Thegreat thing about the online store isthat it is endlessly configurable. Youcan go headless off of our main frameand build whatever you want. The themesare endlessly extensible. That's thewhole point.

00:08:46It's to maximize merchantcreativity. The tension that we've hadcoming into shop is wanting to have auniform enough experience that there'scohesive browsing for buyers, but alsonot to limit merchant expression andstorytelling. And one of the thingsthat's been really cool, this issomething that Luke Deon has beenworking on. He helped to kind ofgenerate the first version of this shopstore or I should really say the secondversion where suddenly the stores manymight have noticed like last kind ofholiday shopping season. Um the storesuddenly felt a lot richer. Thestorytelling was there. There was a lotof video. You know we allowed brand wordmarks to bust out of the like trapped inlittle avatar and like let them sit asreal SVGs that feel rich and feel owned.The next step though and like a funchallenge for us from a data perspectiveis actually getting all of the onlinestore content to feel native within theshop store. And the prototyping thatLuke has been spearheading here has beenhonestly incredible. There's beenlongstanding data structures and the waywe conceived about data within theonline store that initially made it hardto imagine just translating that easilyto shop. What's been so cool about thisproject and this bet in particular iswatching Luke and other designers justchallenge

00:10:02whatever we thought waspossible and it's and it's been becauseof AI wholeheartedly, right? His abilityto develop the set of prototypes thatare able to express this idea butactually get into the real data and notjust, you know, handwave his way throughthis has been truly remarkable.>> Can we go another level deeper there?Like what are those prototypesconsisting of? How are you even givingdesigners access to this data to thecodebase? Like what does thatprototyping look like in practice?Because I know a lot of people arenodding along as they're listening.We're like, "Yes, that's what we want toinstill in our org." But there's a lotof little hoops and hurdles to getthrough along the way.>> The favorite setup right now isdesigners working with a harness like Piusing something like Cloud Code,plugging directly into production datasets, being able to pull those down anddevelop whatever kind of experiencesthey want. And so in the example ofLuke's prototype, a bunch of work hasbeen done uh thankfully to our in ourdeveloper experience team to make itsafe for us to be playing with this datawithout all the kinds of unfortunateaccidents that could come. And so hisprocess has really been talking a lot toClaude about really philosophically theconcept of what he hopes to build,developing a plan and then going back inand actually developing.

00:11:17You know, hisprototype is really fascinating becauseit's not just one endto-end feature. Ithink a lot of times when we're talkingabout prototyping, it's I've got, ifwe're talking about shop, I've got thesimulator up and I'm going to build anew feature end to end, which we'redoing too and is awesome. What he'sreally doing is creating a whole worldto conceptualize of a new productdirection. So when you get in there,you'll see like a whole grid ofdifferent merchant storefronts. You'reable to click in configure the differentdata blocks that are actually being fedby the online store data. And he didthis exclusively using cloud code andhosting our prototypes internally insomething we call quick, which allows usto like get something up very quicklyand send each other links. So we get outof that problem where everything isliving locally. Um, everything's hostedand can be shared internally supereasily, which I'm very grateful for.>> I want to get into the AI piece, butfirst I want to return to a thing thatwe've we've kind of been touching onfrom different angles, which is thisrelationship between like metrics andthe the quality bar and the core visionof what you want to create with the shopapp because I'm sure sometimes these bigbets just works. You know, the graph goup, great, everybody claps.>> But it's probably not that easy, I wouldimagine. And so youwrestle with that tension

00:12:32because youknow that Luke has this amazing vision,this amazing prototype. Everybody getssuper excited, but then the resultsaren't exactly what you want.>> It's a really important part of my job Ithink and part of it setting the groundslike having an intimate understanding ofwhat our business goals are and havingan intimate understanding of what ourperformance looks like for us. So it'snot to say that I or any of mycolleagues live in La Wandland. Um it'sactually having an intimate awareness ofthese things that allow us to moveforward. And one of the keys is on theonset of a project or a provocation,understanding what our tolerance forrisk is and resetting the baseline. Oneof the things that Toby's really good atis making sure that we're not locallymaximizing at any given point in ourproduct. So even when we find thingsthat work, making sure that we're notjust totally locally stuck at everhitting the ceiling. Being an incumbentand expecting that you're going to dothat on the first shot is an impossibleplace to put any product leader in. youknow, be it from design, product, data,engineering, that's kind of an insaneassumption, right? Any team who's had tooptimize an onboarding flow and that onetime you decided to redesign it, like italmost always loses, right? But it'shaving the confidence and the agreementto begin, like we're actually going outto beat the incumbent and we're going toreset a baseline that inherently mightbe a little

00:13:48lower than what status quois today. There's a great example ofthis that was led by Jess Ericson, ourgrowth team in Shopify where like theShopify signup page was just thishardened thing. It's the lifeline ofgrowth for Shopify and so touching it,it was just such a sensitive thing andthere was this belief that like oh someof the designs that the team wereproducing were absolutely beautiful,super brand forward and there was justassumptions being thrown at this thinglike it's not going to perform as well.It's not the right time. We're not sure.Sure enough, we like set the grounds tobe able to ship that and it was not justnet neutral, it was net positive. And itwas such a great like we've been usingthat story internally a lot to say likeyou need to create the grounds to takesome risk knowing that you're going tocommit just as much iteration into thenext thing as you did the the theincumbent product.>> Can we talk about the interaction toopen up the cart with like the drawerunderneath it? That's a super novelidea, right? Like I I don't think that Ihad seen that in a shopping app before.It makes total sense. I think I'veactually seen it in other apps sincethen. Did that have a metric storyaround it? Like how do you take that bigof a bet? Because it's like gosh, ifyou're not going to break one thing in ashopping app, it's like the UX of thecart, right?

00:15:03>> I know that was a John Rendle special.One of the things, and I'll beinterested to see if this sticks with usas a process going forward, but one ofthe things that has underpinneddeveloping shop over time, is this funnything we do called shoplifting, which isabout once a year, I want to say itdoesn't just come from me, but sometimesit really does where I just have thispencils down moment. I'm just scrollingshop and I'm like, we've lost our touch.We've shipped a lot of stuff. The roughedges are starting to show. The seamsare starting to show. It's time for usto have this moment, these two to fourweeks where design just goes ham onevery surface horizontally. A smallgroup of designers saying, "What wouldcould we do to make our product feelthat much more special, that much moredelightful and fun to use?" And out ofone of those shoplift sprints cameJohn's undercart, as we call it. Andit's been such a signature of the appand like as romantic as that is to say,especially in this era, I think it'ssomething that's become like a verybranded part of our product. It's beenwild to watch, you know, going back tothis thing about being super data drivenand standard commerce experiences. I've

00:16:18been so floored to watch other muchlarger marketplaces, if I'm totallyhonest with you, start to get inspiredby what we're doing in shop. And mostpeople would be annoyed by that. I thinkit's fantastic largely because I'm like,well, their current scale is like quitea bit larger than ours. And so ifthey're deploying these ideas, thatactually also means that they work. Imean, take for example, if you hold asidebyside up of um where we landed withour very content driven feed and whereAmazon took their core discovery feedwith these cards. um target as well.Clearly, they see something in the morestorytelling narrative and I that'ssomething I'm really proud of honestlythat we're able to take the risk andthat that's it's inspiring the rest ofthe industry to make shopping moreinteresting for everyone honestly.>> All right, let's talk about the two tofour week periods we have the pencilsdown moment and and maybe just even thethe broader philosophy behind thatbecause again I just keep coming back tohow measurable every single thing in ashopping app is, right? Like metrics areking in so many ways>> and if you let metrics you know run wildyou probably get to Amazon right likethat's that's where all roads lead. So>> how does the emphasis on visual appealshape the way that you as a teamoperate? Can we go a little bit deeperinto that part of

00:17:33the way that you work?>> It's something we earn through havingreally trusted relationships withproduct and engineering. It would be avery hard thing to accomplish if we justsat in our cool little, you know, designivory tower and like tossed down the nenext version of shop that we justinherently believe is right. That's nothow this works. It still happens withinextremely small, very talented,crossunctional teams. The expectationthat's set though is that we're alwaystrying to reach for a new bar thatcontinually starts to rise. andsomething that I'm very grateful myproduct counterpart and particularly myengineering counterpart feel aligned inwalking into this journey together andfrankly also it really helps to be at afounder company where the founders kindof set the stage right some of the mostparticular directives I get from Tobyare like Katarina I want to open up shopand see something delightful that's nothow a lot of like CEOs talk uh you knowlike it's typically uh you know I I hopeto see this KPI go up I hope to see thislevel of user growth We all hope forthose things, but there's just aninherent sense, well, not an inherentsense, like a real real directive tocreate things of quality, and itpermeates everything we do at Shopify.I'm not

00:18:48saying that it's always a smoothconversation. There's negotiations thathappen every day about where are we onthe build, is it ready, how do we makesure that we're making the righttradeoffs. Much to the sugarness of mycolleagues, I'm very much a have a cakeand eat it too kind of gal. like I thinkit can be performant, beautiful, uh, andthat we can ship quickly.>> Let's say Toby asked you to bringsomething delightful into the feed. Likewhat are the types of things that youwould even reach for then? Where doesdelight manifest inside of a productlike this?>> Yeah. Well, the things we're reachingfor to realize these ideas are wildlydifferent right now. I mean, whether itbe our shoplift sprints or, you know,every December we used to sit down andsay it's time to generate like a visionand a moonshot for the year for shop.That's something that we've done everywinter closeout period. It's kind of thething we do before the holidays. I thinkthat is changing entirely and I can talkabout that in a sec. But typically whatwe've thought about is just reallydreaming up things. And I I would tellpeople in those moments like block offyour calendar, go to your like nearestshopping neighborhood, go visit some ofour merchants in their real retailspaces where they've created theseworlds. I mean, if you've ever been to aSkiims retail location or Tacobas orOutdoor Voices, there's a whole worldthat they build. And like what I've beendeeply passionate about is trying tocreate recreate

00:20:04the joy and theexcitement of those worlds in software.Not easy to do, but to really pushpeople to find inspiration in all matterof sorts of places. And so it starts,you know, outside of the app and reallyreally inspiring people to go findinspiration outside of where we where weusually work. I think today what arepeople reaching for to render thoseideas obviously is changing greatly youknow talked a little bit about Luke'sprototype talked a little bit about howoddly enough and I think it's debate uhthat is reragging for a lot of designersin organizations that are buildingmobile products is whether to stick withum react native or go back to swift andcotland right I think all the colleaguesI'm DMing right now who are still areact house I'm like no if react nativereally makes much sense anymore you knowswift and cotlin and are becomingthey're much more legible languages.There's no layer of abstractions. agentschoose to write in those things anddesign is actually now responsible forkind of pushing us in the direction ofnow reconsidering our commitment toReact Native and and [clears throat]working directly in Swift and theprototyping that has come out of thatand the tool building has beenabsolutely phenomenal and it's unlockingcreativity in all manner of sorts ofways which has been so cool

00:21:19to see>> and that push is just coming from thefact that designers are more empoweredto raise the ceiling with Swift and isthat something that you're like seeinglike how how common is it for designersat shop right now to be tinkering withSwift versus simply doing like you knowstatic mocks more traditional design>> it's changed so much so quickly one ofthe things that I I think a lot ofpeople who working in mobile design aregoing through this transition butFigma's always been a mainstay as thecanvas but this transition or thistension into other tools we've felt fora long time as many of the talenteddesigners who came to shop introducedorigami into the organization. I kneworigami from 12 years ago working atFacebook. Like that had been somethingthat came into the organization. Not atool everybody was reaching for. Thelearning curve felt high. Origami is avery particular piece of software. Butour dreamiest ideas usually came throughthe subset of designers working inorigami. And a lot of people were askingthemselves other designers in shopshould I be learning origami? And it'sat that exact moment that like we havethis explosion of tools and our abilityto work directly in code that peoplestarted to say, I mean, I could learnorigami, but I'm not sure that makes aton of sense>> or I could not>> or could not or I could just be workingdirectly in the code. And one of thefirst

00:22:35as Shopify was building up itskind of prototyping capabilities. It wasreally easy for us to spin up that worldfor the admin. We now have adminplayground which is the mainstay of howmost the designers work now. might sortout a few things in the canvas and Figmaand then go directly into the adminplayground, work with cloud code anddevelop their ideas. We didn't quitehave the same environment built up. Andso designers hit a bit of a wall tryingto get into production code working withReact Native. It just having a couple oflayers of abstraction and the way thatthe codebase was structured, it wasn'tsuper legible. And so some of ourdesigners started to basically do atransition and build a version of shopinside of Swift that's a lot easier towork with. And we've had incred likejust an explosion of people now justfeeling like they can jump in and workdirectly there. And so kind of had thislike hop and skip over the the previousleader and tool in Origami and nowpeople are just working directly inSwift. That has then led to thediscussion, well, if we want more peoplecontributing to the code, shouldn't webe working in the most legible codingframeworks that are native to theplatform, whether it be React for web,Swift for iOS, and Cotlin for forAndroid? And that seems to be winningout. And at least in

00:23:50the case ofShopify, that's really been a design-ledpush unintentionally just based on theprototyping tools that we've picked. andit's forcing us to come to the tablewith, you know, our CTO and and Toby andsaying like I think it's time to toreconsider the the question aroundnative and seemingly we're all bought inand it's as you said just has completelyraised the ceiling on designercreativity which has been really cool tosee.>> The fact that that level of technicaldecision can actually be influenced bysimply wanting to get more non-engineersinto the code is really interesting.Like that to me speaks quite loudlyabout where the future of work isheaded.>> Absolutely. And I think it's it's it's avery agitating force within theorganization, right? It's it's noteverybody's like, "Cool, sounds great."Posted this the other day, but it's likeCleo keeps saying the future belongs todesigners who can build and I'm like,"Yes, and those who exist inorganizations that have philosophiesabout tech stacks and how they work."like part of the our role now is talkingabout the user experience of buildingwhich is super interesting. I don't it'sa funny byproduct of us being designerswho are trying to build in this new waybut I kind of love it.>> It's something that I think about a lotlike I have a few friends that work invery

00:25:05legacy like banking infrastructurehealthcare style companies where they'reso far from what you would consider likea modern web tech stack. And as aresult, there's this splintering of whatthe discipline of design even lookslike. Like my day-to-day looks soincredibly different than somebody whofrankly it's not even their fault.There's just no way that they canactually work with the real code becausethere's so many hoops and hurdles tojump through. You know, it's just it'sjust not set up for that. And if you areworking either like in Swift or mobile,but definitely doing the you know,typical React Next, Tailwind, invitethat kind of a thing. It's a runawayfreight train right now in terms of whatdesigners can do where the very shape ofour roles are just moving so much morequickly than people who don't have theability to work in those types of techstacks.>> Yeah. And my advice is and this ishappening at Shopify, but I think it'shappening a little bit everywhere. It'seasy to think about doing this insideprojects or think about doing this atstartups where like there is no basis.You know, Shopify is over 20 years old.There is a lot of legacy code. There isa lot of things to consider. There are alot of philosophies about front ends,backends, all this stuff. I find somedesigners are like exhausting themselveslike fighting those battles headto headon the principle

00:26:21trying to debate withengineers directly about theirtechnology choices. The advice I've beengiving more recently is like tell youwhat refocus on what makes you as adesigner differentiated which isdesigning what's desirable. Don't worryabout feasibility. Pick the tools thatare going to allow you to render theidea in the most believable way. If thatmeans you're going to spin up a webproject in React, even though you knowyour team's using web components orwhatever, whatever they may be using,don't let that stop your vision, gobuild the thing. The fact that you can'tship it to production doesn't mean thatyou've failed. I think Max from uhNotion was just talking about this theother day like not really caring ifdesigners are shipping to productioncode that it's more important thatthey're using the medium to be able torender the fullness of their ideas andto work with real data. That to me iswhat's most valuable. And so, you know,don't worry about whether or not yourthing is going to ship to productionright now. That problem will start tosort itself out. And I can't waitbecause who doesn't want more peoplecontributing to code safely and andperformantly? I think everybody wouldwant that. Focus on on rendering youridea because it's literally in the spanof time been like a total of 2 hourshistorically that designers couldproductively ship to code if they didn'tknow code in the first place. Andsuddenly

00:27:36we went from you know myresponsibility is to render idea to myresponsibility is to like build thething end to end. And that's where a lotof designers are actually getting stuckright where they're like oh my gosh Ihave to like my tests are failing as Itry to merge this PR. And I get to apoint where I'm like, first of all, askyourself, is it the right scope thing tobe trying to ship to production? Are youtrying to ship a full feature or youjust trying to fix some stuff? If you'retrying to prototype a full feature,don't go for the moonshot of end toending this and shipping it toproduction. Just render the whole ideaand then use that as the thing you talkto engineering with. And if they can useit, amazing. If they can teach you howto build it more usably next time, evenbetter. But that's not the expectationI'm holding for designers right now. Ijust want to see the best ideas, moreideas. Like that's how we build greatthings.>> Yeah. A lot of my work starts right nowwith uh asking Claude to create a routethat starts with SLP playground slash swhatever it is. Like>> unfortunately I duplicate a lot of, youknow, components and and break themapart. So maybe there's a little bit oflost token use there. But uh>> it, you know, it's quickly become astaple of my process. I think of that asthe funny like signature workbench. It'slike the old how organized was yourFigma? Did you name your layers? Is likehow we reason

00:28:52about where our stuff isstored and why is kind of the new messwe've all created in this work which isfun.>> I was one of the very OCD Figma people,you know, like I literally taught acourse about how to organize Figmafiles. I cared and I had a lot of ideasabout what the So now I'm the kind ofperson who like I have a whole branchthat I actually am going to merge inprobably in the next couple days andit's literally like>> the organization layer for all of myplayground files and it's all like datedwith like AI generated synopsis of likewhat I was exploring and it's like thisbig timeline and I'm like>> a new designer anyagram test like you'reprobably also like a clean desk workerwhere>> oh so clean so clean my desk right nowyou'd stream.>> There's one question that I can't stopasking myself. What if companies appliedto talk to you rather than the other wayaround? And that question is thefoundation for the all-new dive talentnetwork. And it's working. Like rightnow, I'm helping many of the mostexciting startups that I know to hirethe designers and builders who listen tothis show. So, if you're curious whatmight be out there and maybe you want toget on my list, or maybe you're evenlooking for your next design hire, headto dive.club/talentto

00:30:07join today. I want to push back onone thing that you said. I'm going toput on my design Twitter skeptic hat fora second here. [music] And you talkedabout how designers really kind of needto ignore feasibility by, you know, justmake the thing. Don't worry aboutactually getting the thing into prod.Which I'm like, yes, that makes sense.But if we take that one level further, Ithink you could also make the argumentthat the purest form of ignoringfeasibility is actually not working incode. It's just throwing the quicksketch of the crazy idea that startswith what if and you have no idea if iteven makes sense to build or if it'seven possible to build. And that byembracing code as a medium, designersactually are putting themselves in alittle bit more of a feasibility box andpotentially capping some of the crazyideas simply because they might be alittle bit more challenging to makereal. I'm curious if you have a a takeon that.>> I believe this was like popularized byIDO a million years ago, but it was liketheir sweet spot for innovation, right?And they talked about three things thatyou need to innovate. desiraability, youknow, is this a thing people want?Feasibility, can we build it? Viability,is this something that if built can helpa business grow and sustain? One of thethings that I've talked about in thepast is that

00:31:22you could actually mapthose responsibilities to our crafts,like designers are responsible formaking it desirable, engineersdetermining feasibility, productmanagers figuring out if it's viable.Right? Rough roughly speaking, there'sthings that are no longer true aboutthat and things that are still true.things that are no longer true aboutthat is that like those don't cutcleanly across crafts anymore. And Ithink that's very exciting. I think whatis still true about it is that designshould be the tip of the spear fordesiraability. And I think that'sprobably the genesis of like if youextrapolate that comment out to liketaste, editing, deciding what actuallymatters, that's a skill set that hasbeen naturally a home in the designpractice since before computers wereinvented. So that has just been along-standing thing that design hasdone. I think more types of people, moretypes of builders will begin to do this.But I tend to agree with you. I thinkyou were challenging me, but I tend toagree with you that the thing that Ihate to see are is when constraintsstart driving the design process,especially now. The advice and feedbackI'm typically giving to our more juniordesigners is they'll they'll show up toa critique or a demo and say, "Here'swhat I thought about." And beforethey're telling me a story of what theirvision is or walking through theirprototype, they're like, I talked

00:32:37tothis engineer, this data isn'tavailable. I talked to, you know, andwe're like into a grocery shopping listof why something can happen or why whatI'm being presented is like a verycompromised version of the idea. That'salways been an issue and alwayssomething I'm trying to coach out ofpeople. The thing I worry about with anynew kind of tool explosion as designersare learning is that they'll startdesigning to their local skill set. AndI think it's really important in theoryright working with tools like you know aharness like pi or cloud code you shouldbe able to break beyond thoseconstraints but it's not a format thatevery designer is super comfortablewith. There's a certain level ofprompting you have to understand. Youhave to get your environment set up. Andthe thing that I keep stressing withdesigners is like don't lose your idea.Like do not lose the full fidelity ofyour idea. If that means right now foryou the first thing you need to do toget your brain to map it out is gettingsome screens and Figma reasoning aboutyour idea before you try to create aprototype, great. There's nothing wrongwith that. But do not let constraints ofyour own skill sets in this momentconstrain what's possible. because thenI think we're just missing out on thegood ideas and that's what's importantright now.>> I like that a lot. It reminds me of thephrase taste execution

00:33:52gap that wealmost used as a way to define whatdesign engineers brought to the table.And>> yes,>> even for myself, it's hard because everytime that I try to build something and Ipush myself a little bit further, I canfeel the muscle growing of, you know,front end interaction, all this kind ofstuff. And yet every once in a while Iwill get blocked on executing againstsome piece of an idea because it's somestupid state that I just can't quitefigure out. You know, it's like soengineering and not design and you kindof almost have a little bit ofconviction where I'm like, "Ah man, didI just did I just go too far in the codepath here where actually I could havegot enough of the nugget of an idea toget what I need to take the next set ofsteps and it could have just been likedoing things a little bit moretraditionally in Figma for instance."That's why the tool you had Marvin on acouple weeks ago and he talked aboutartifact, this new tool that we built tokind of be the new place to share what'shappening in design. It's thismultiartifact approach that actuallybecame a very important driving forcefor why this thing needed to exist sothat in a workspace you can have anysort of asset, right? You can have aquick origami prototype that gets themotion transition just right. Like wouldJohn Rendle have like prompted theundercart? I don't know. but heorigamied it and it's beautiful and it'sbecome a signature in our app. I'm

00:35:07thrilled about it and we built it. Youknow, whether it's Figma, a webprototype, a Swift prototype, a videocapture of an origami, whatever it isthat they can all be collected in onethought space canvas of all these assetshas been incredibly helpful for thisexact thing so that we don't get stuckthen trying to build our idea and kindof getting stuck in the in the deathloop if it were.>> All right, so we have a bunch ofprototypes flying around. We haveartifact. We also have Toby who'stalking about how, you know, he wants tosee more delight in the app. We have abunch of talented designers who are justexperimenting and tinkering and buildingtheir own tools. So my question then ishow do you make sure that you don'tovercook this app and stuff it full of abunch of things that in isolation areamazing and yet together you have thisoverwhelming experience.>> Editing has been an important part ofthe job as of recently. um reallyfocusing in on what is the most criticalaspects of our experience and gettingreally really disciplined about that. Ithink certainly that's actually aconversation we're having right now aswe think about shop coming into a stageof maturity and what it is and what itcan deliver. That's been a really bigpiece of it. What are the primitivesthat really drive the most successfulsessions for users? What are they comingback to? Package tracking just stillreally being

00:36:22a mainstay. the feed beinga mainstay, the new innovations we'retrying to drive in more of an agenticsearch, like those kind of being thethree major things that we want to begreat at, and then when we think aboutmore broadly in shop, what it needs tobe good at from a network perspective,we obviously have our other suite ofproducts that people often don't thinkabout when they think about shop, whichis Shop Pay and our identity layer,getting to sign in and kind of being theidentity center of Shop and reallyeverything circling around that andmaking sure that we're not crossing overinto that red line of doing things forthe sake of doing things and making surethat we're really disciplined. It meansthat like all of the fun things thatwe're doing, we're generating way moreideas now than ever than we ever plannedto ship. Whereas before we had very fewvery exciting ideas that we put all ourwood behind kind of one arrow. And Ithink that's one of the things that'schanging and getting designerscomfortable with the fact that like>> we're not going to ship everything. Butall of these things play a role ininspiring either us or other parts ofShopify, which is super exciting.>> Is that what ties back to your ideaabout no longer doing the moonshot betsat the end of the year? Just the purespeed and the amount of ideas that aregoing out. Can you talk a little bitabout how that's shaped the way that youthink about roadmap

00:37:38sequencing, thepractice of product strategy even?>> Absolutely. Yeah. the the idea that welike sit down in January and decide whatthe app's going to be next year is iskind of hysterical now given [laughter]how quickly we can generate ideas. Whatwould end up happening typically is wewould make that design vision and thefirst quarter road map would fall out ofthat and we'd kind of say which of theseideas should really shape our road map.What are you know the backend systemsthat would need to exist in order toserve that kind of experience. Okay,we're going to make those investments inQ1. We've really been always designledin how we shape our road map. What'shappening now is that that's justhappening in micro doses week to weekwhere new prototypes are beinggenerated. The one I spoke of earlierfrom Luke like he made it he was kind oflike designer Frazzled of Lost in theWoods like I'm just cooking on all theseideas and I'm like this is veryexciting. Can you tell a coherent storyin a quick async video that we can passaround and see if people are excitedabout it? Sure enough people's like hairlit on fire was like oh my gosh we'rereally on to something here. we zoomedhim in a couple of engineers and likethinking about existing team structuresand squads, it's like no, let's justlike chase heat on this idea and see ifit has legs and see if we can build itout. And so to your point, yeah, it'schanged immensely and our

00:38:53ability tojust validate ideas is uh now happeningso much more regularly, so much faster.>> Your comment on the video made me wantto go down a different rabbit hole,which is obviously something like thebottom drawer is just a good idea,right? like that worked and there wasbuyin because it's a good idea thatexists on a spectrum, right? Not everyidea is a 10 out of 10, but there areplenty of good ideas that are worthshipping, but there's now this challengeof like, well, you got to get kind ofyou got to get people excited about it.You got to get there's got to be somelike buyins, some support, some heat asthe word that you used. So, my questionthen is let's say a designer joins theshop team. They are now a year into it.What do you expect them to have learnedabout building excitement for the ideas?And what are some of the like tactics orstrategies that you see of the bestdesigners that are consistently able torally support or even just have ideascirculate internally?>> I think the answer in a previous worldof building wouldn't have been thatcontroversial, but it was like make areally compelling prototype and tell astory. I think that stands now. I justthink that your ability to do that, it'shonestly much easier.>> Yeah, it's democratized.>> It's it's extremely democratized. It's alot easier, I think, for designers whostart, you know, where they

00:40:08are fromwhen they start to 6 months and then 12months in. Um, I have the story rightnow with a, you know, this predated ourapprenticeship program, but we had uhLiam Fenel join us as like designer andresidence. It was like we didn't have aninternship program, so that's what Icalled it. and he started with us andfresh out of school computer sciencedegree looking to really go deep indesign and the starting fundamentalswere like start small and small featuresbut make them absolutely beautiful rightand so like getting down the craftfundamentals as he's grown incrediblyquickly in deepening his skill sets as adesigner working alongside other greatdesigners and then adopting the toolshe's landed at the logical place I thinkreally great designers always land whichis a hunger to shape the strategy ofwhere shop is going. As much as there'sa really myopic focus on like designersin code, which is important and true,and I am totally signed on on, we'remissing the other part of what this erais unlocking for designers, which isaccess to data, access to allegibilityin the business that you didn't havebefore. You know, one of the tools Ilean on most often in my day as I'mtalking to my agent at work is the dataportal. I used to say like, "Oh, I'dnever run product because I don't knowSQL and I can't query my own tables."Well, that went out the door. Like now Iget

00:41:23I can ask all matter of dataquestions about what our users aredoing, what's working, what's not. Andthat is one of the most helpful piecesto informing a holistic view on what weshould be doing and what we should bedesigning. And so, you know, back toLiam, he's 8 months in now or so, andhe's talking to me about the fact thathe really wants to be in more strategicconversations, understand why we'remaking certain choices, why we thinksomething should ship versus not,because I think he has an understandingthat's going to overall make him abetter designer. And so, that's the bigshift change is people come to do greatcraft work, but they leave really greatproduct thinkers. And that to me is oneof the most enduring skills in this eraas we move out of like our central valuebeing execution into being people whocan shape great products that peoplelove to use.>> Okay. So then let's talk to a designleader who is not at as innovative of acompany as Shopify. there's a little bitof a lagging effect. And they're lookingfor, you know, what are the north starsthat we should point at organizationallyto help us keep up with everythingthat's happening, all of the newcapabilities that AI is unlocking. Twothings that have stood out to me is youhave Artifact, which I'm a big believer.I mean, there's the whole reason I'mbuilding a startup around it. Like, I Ilove Artifact. The

00:42:38second thing is thedata portal and everything that thatunlocks and democratizing this access todata. Is there anything else about howyou've seen the practice of the designshift internally that you think othercompanies could learn from?>> Making uh internal decision-m legible.We transcribe all of our meetings. Wemake those accessible. We make ourentire internal development processlegible through a homegrown tool calledVault where all decisions made on allprojects can be easily documented. Wehave>> Oh, very cool. G GSD get done.That's like our main way of saying likehere's the proposal, here's theprototype we've decided to, you know,build and we've decided to release. Andone of the things that we're iteratingon right now is like are there fewersteps? Probably based on how the processhas changed, but it gives us a reallyclear way both from a decision-mstandpoint for Toby to underwritedecision makers and say like you'regoing to be in the queue to just look ata couple of things coming through andsay like yes, this should go, no, itshouldn't. this should be a proposal,this shouldn't. And our ability to makethose decisions is so much betterbecause we have real highfidelityprototypes that we're looking at. Butall of that data then has an MCP hookedup to it that as

00:43:53I'm looking to makedecisions as a product leader, I canjust ask my agent and it will query andfind all of these previous decisions,we've actually worked on that before.Here's where it failed. And so making asmuch internal decision making legiblefeels critically critically important. Ithink too just no matter whatorganization you're in, great designleaders of this era just have to have adeeply intense relationship with whatthe business is there to do and pairthat intensity with a deep relationshipwith what makes great design. Like thisis the combo that has up until thispoint served me pretty well. Like I justfeel like it's so critically importantto bring a deep passion and intensity tothe work that's happening that you arealso really deepened because that isultimately going to be your leverage.When designers tell you that they'rerunning against a wall of gettingsomething implemented because of anengineering constraint or something inthe data, you know, says otherwise, youas a design leader need to be able tozoom into those conversations and bereally productive. And it's funny for meto think about that because I thinkabout a lot of the advice I've gottenover my career as a design leader that Ipurposely chose not to listen to. Iended up in a place where like a lot ofthat ended

00:45:09up being probably the rightintuition about staying close to thework, staying close to the craft, reallyhaving a deep understanding of thebusiness, making you like a reallywell-rounded decision maker that cansupport your team and support thebusiness outcomes. from like amanagement standpoint, if we keep on theleadership trail for a second here,something that I'm interested in is howyour philosophy around what you gooddesign leadership looks like has shiftedespecially given all of the AI inducedchanges where in many ways like IC's arehaving outsiz impact like the ability toshape strategy or the shape the way thata team operates. I mean there's been somuch power afforded to IC's even in thelast four months. How what is that likeas a leader?>> I think it's a question on a lot ofleaders minds and it's also has peopleleaving leadership to go back to beingan IC if they nurtured their craft allthis time. I certainly always describemyself as a designer first and foremost.You know when anybody asks me what I'mdoing I design and then you know if afriend's around or my wife she's likeyou like manage x amount of people. I'mlike, [laughter] it's not the certainlynot the part that animates me in my job.What I'm animated by is trying to be agreat designer and I'm very fortunatethat I get to do that alongside

00:46:24reallyreally talented people. All of them aremore talented than me, let me be clear.But coming from the craft is just reallyimportant. I don't think that part'schanged for me, but the industrycertainly caught up. I would say 10years ago is getting advice to say likeas if there was like some hardened pathlike oh you'll have to like step awayfrom the craft. You'll need to focus onthe way you run a one-on-one. You'llneed to really develop this entire likemanagerial set of skill sets that feltcompletely isolated from moving theproduct forward or like helpingdesigners design better. That alwaysfelt really foreign to me. We're at areally interesting time 10 years laterwhere you know Brian Cheski is talkingabout this and Spiegel a lot of peopleare talking about the fact that likeactually no staying really close to thecraft is incredibly important. I find itreally interesting because like the theguidance is like fly at the higheraltitude, work horizontally, all thesethings that aren't wrong necessarily,but I think one of the reasons whydesigners in shop tend to stick around along time is is I think they feel thatthrough the the culture we've built andthrough the way I lead that they'rebecoming better designers every day bygetting all to work together. And Ithink it's really

00:47:39hard to generate thatculture if you're not deeply passionateabout the craft. I think it'll be areally tough go in the long run for likethe professional manager whose calendaris really driven by one-on-one timetalking about career development. Like II just don't see that being the the mainstay. That's important and it can happenat a certain cadence, but one-on-onetime should be about the work. Let's getinto the work. Let's open up yourprototype. Let's have live critique andand you know move this work forward andhelp designers figure out how to makebetter decisions. I think right nowthere's a lot of discussion around likeCam talked about this in a couple ofweeks ago ditching critique for demosand and demos and people are buildingthat's the thing that we should belooking at. I think that's definitelyimportant and something we've adopted.But I think designer's role is stillfiguring out like the editing, right?The critique. Why is something betterthan another solution? Why is everybodyon this call suddenly like, "Ah, this isso cool." versus cocking their head toside and not really having the words toarticulate it, but like no, it's notright. It's really important that wevocalize these things and like have theopportunities to give critique inwhatever we call those formats andhowever they happen, async or sync. Whyis that important? Well, our materiallanguage for building

00:48:54is now becomingwords. We now have to give critique toour agents. Most the time we have topoint at the thing and say, "This isn'tright, and here's why." The higherfidelity that designers can have in thelanguage they use to make their workbetter, the better off they're going tobe in an agentic era where they can tellagents what they want to have fixed. Thebroader your language, they're like,"This doesn't feel right." You'reprobably not going to oneshot something.giving an agent that feedback andcertainly designers around you aren'tgoing to get any better. So creating thegrounds for that level of feedback to behappening feels critically important inthis moment.>> It's so obvious in retrospect, but Ihaven't actually thought about it thatway. You're right. Like I give so muchmore design feedback every day now thanI previously would have. And it stillfeels like I'm designing, but actuallylike no, I'm looking at things that arenot good and trying to figure out whydoes that not look good and what needsto happen for it to be better versushaving, you know, me just being the onethat's moving things around. And youhave way less points where you take kindof take a step back and evaluate yourown work where now I'm doing it likeevery 90 seconds.>> Um, and we're so much more willing tobecause we didn't do the hard labor ofpulling the thing to make the rectangleor whatever it is that we were doingbefore. We're so much more willing andit's actually been really funny. I've

00:50:09been having some conversations with somefolks who are like whether it's valuedor not. Like for me, I had a classicdesign education and studied graphicdesign at university and I still talk ina pretty traditional way. I'll talkabout hierarchy, balance, symmetry, allof these like core tenants of gooddesign that we used to talk about, youknow, 30 years ago. And I was talking tosomebody recently who's like gosh, likewho's an incredibly talented designer,mind you, like top top designer, is likeI didn't come from that background and Ifeel at a loss for words. these days andthe way you describe what can beimproved in design like what should I goread how do I like re-educate myself insome of these concepts you know and I Ijust kind of giggled so many softwaredesigners now are selftaught didn't comefrom any background in design which iscool and amazing but at this time wherelike language is becoming sofundamentally important in our processthe lack of words people have issomething that they're reaching for thecool thing though is this whole thing isrecursive because I'm Like I canrecommend you a bunch of books and talkto you about my design education, butyou can also just ask Claude or yourchosen model about how to betterdescribe the work that you're doing andthese concepts so that you be can becomemore articulate as you give feedback.One more thing that I want to ask youabout before you go

00:51:24and I kind of justwant to talk to the person who isfeeling inspired to, you know, expandtheir tempo with AI and do things thatmaybe are outside of the traditional boxof responsibilities that they've heldfor the last, you know, four or fiveyears of their career or whatever, butit's not super clear. Are there anyother examples at shop where we couldpoint to to help people get a bettersense of what it looks like to reallycapitalize on this moment in time withAI and to really push things forward orexperiment or move needles that justweren't possible even a couple years agothat other people could kind of use asmore tangible inspiration for their ownroles. Part of the reason why I'm atShopify and have stayed, it'll be almostsix years in the fall, is because it wasa place that never put me in a box. Mycareer prior to this was marked by beingan IC designer, then being the personwho would hire other designers andeventually manage them, and then begiven the responsibility of running theproduct and then spinning out of myfirst startup thinking I needed to go bea product manager and like went intointerview loops as a product leader andthen was like, "These aren't my people.What am I doing?" and having a totalidentity crisis and then kind of sayingto myself, you know what,

00:52:40I'm going toreshape my expectations here. I am onlygoing to go work at places that allow meto exercise the full spread of mycapabilities to help build the rightproduct and I'm only going to go work atplaces that allow me to do that. AndShopify is one of those places. What'sfunny now obviously is like there's allof these discussions about how roles aremerging or not merging. Unfortunately,there's like a very zero- sum sillydiscourse that's happening now is likewho's taking whose job? And I'm like,"Oh my god, who cares?" Like, this isnot this is totally not the the takeawaythat people should have. The takeawayshould be it's never been easier toextend what your capabilities are as aproduct leader. And so, you know, theexercise that invite any designer intois to say, what's a constraint you'verun up against that just bothered thehell out of you? Like it could have beenan engineering constraint. It could havebeen prior data here or testing says Xor Y. Keep in mind data can be twisted.It can have a story behind it. Notintentionally, but like data itself isobjective. The reading of it is neverobjective. It is always subjective. Andso you have an ability now to query thatdata, pull it back up, and say, I'mgoing to look at this. I'm going to talkto, you know, Claude or

00:53:55whatever andactually do my own analysis and come tomy own understanding of what this testsaid and determine, did we misssomething here? Are we biased in the waythat we're looking at it? Think about aplace about a former project or aproject you're in now where you feellike you're being blocked and itinherently feels like a differentcraft's responsibility that's theblocker. and say actually there'snothing stopping me from like jumpingthe pond here and like embedding myselfin that craft or in that skill set orlearning it so that I can come to my ownconclusions about this thing. I hopethat everybody's working at a voice likeI am where that's actually encouragedand Biden and not seen as a you know asa threat or stepping on toes like weshould all be running in the samedirection in our organizations thatthey're healthy and you should embracethat opportunity to be able to teachyourself or get to your own context andunderstanding. Toby has a pretty simpleadage that I feel like I'm saying moreand more these days, which is like>> figure out what's true and then figureout what to do about it. And that thatis like the baseline under how everyoneshould be operating. And I just love itcuz it's really simple and I think itcan apply to almost what everybody doesin their everyday work.>> I can't think of a more inspirationalending point. I appreciate you comingon, Katarina. is super fun and I'm justa big fan of honestly like the years ofyour work that you've put out there andand

00:55:10shop app and everything in the team.Uh just always a joy to get a littleglimpse into how you all operate andwhat you all think about. So Iappreciate [music] you coming on today.>> Thanks. It's great to be with you.