Unwritten Rules for Better Web Design w/ Vitaly Friedman

are there any like tips you can give someone who's like okay you know what I don't have the time or the resources to run multiple tests with that many people is there a shorthand version where like you know never do this always do this and sometimes do this fortunately there is a the ex Community just incredibly like incredible on LinkedIn so there are plenty like there is no shortage in templates we always need to take templates with a bit of you know grain of salt because again it needs to be kind of customized for whatever your needs are I think but in general you know there is this almost like Unwritten Rule and I think that this is something that we often forget just take five people five users and speak with them about whatever they are going with if they encounter any issues is everything clear just try to find any problems any issues any pain points that they have just five people you don't need a lot of time I think that very often we think that this is going to take an incredible amount of time this going to be extremely expensive it's going to delay the project it's going to derail our commit ments not necessarily you just bring five people in the room or in a zoom room you have a conversation and you will find incredible insights already so just start with that then once you have that in mind what you can do is just you know run a couple of you know one-onone is ability testing where you start giving people tasks to see how they performing those tasks and as they do you can bring up your questions and try to figure out what exactly was the issue but for that you need to of course understand what your environment is and what your product is what your service is and so on what things are important you might want to break it down by different groups of people who are using the service or using the site usually the good layer or good level that you want to achieve is when you have this kind of endtoend testing somebody just lands on the product they go for onboarding eventually they have to do some task to get things done and then eventually they maybe upgrade right this could be like endtoend testing but then if you have let's say something very specific that you've been working on last quarter like search well then you go in and you start looking okay let me come up with 10 12 tasks related specifically for search different types of searches with filters without filters with sorting without sorting and then you see if this is working this doesn't have to take a lot of time I know five people already you will get incredible insights in web especially maybe with design and branding it's a little harder to measure but how do we make it so that we can measure if it's effective if it's intelligent if it's efficient or whatever how do we measure this stuff when it comes to designning ux when whenever I'm designing anything I need to guarantee even before I start designing a single Pixel on the screen I want to guarantee that it is going to work better than what we have right now I do not want to waste anybody's time be in a position where I have to get into defensive mod saying oh well we tried our best but we failed right especially when you're dealing with a large organization a large company so for that what we actually do is before we even start designing we think about okay let's create these kind of Concepts and we can do round some concept testing or also we can also take an existing website and create a baseline or status quo first so for that what we do is this we take a website or product that we have and then we basically ask ourselves okay what are some representative tasks so-called top tasks that people are going to perform there so if it's like a banking application they're probably going to send money to someone maybe they're going to withraw money or add money to the bank account so we create something like 10 to 12 representative depending on the environment in which we are and then usually you need at least 15 people 18 is better but at least 15 after that you actually get to the stable results in terms of like statistically reliable data what you get is you bring 18 people uh ideally and you give them this tasks all 12 of them and you spend maybe one hour of them some of my colleagues do it kind of unmoderated testing where they're not even present they just send this tasks and there is a basically a tool like you know there are plenty of tools they can just record Rec it for you and then they send you the result of that recording but I like to be spending the time with that person not like in a physical room but usually on Zoom so I basically give them task I tell them I'm not going to speak I'm just going to observe I'm not going to tell you what to do we're not testing you of course we're testing the website at this point feel free to stop feel free to fail if you fail I learn from that that's very valuable and they go and what we do is we measure two things first we measure how much time they need to complete those tasks and we measure if they're successful with it or not you get 18 people right completing those tasks you get to the point where you actually get an average time per task right for every task you will get an average time that they need right and also get the success rate typically what you would aim to achieve is at least 80% success rate and then if you actually run it once and then you run it with a new design 6 months later right five months later after Out roll out you bring the same segment of people not the same people but the same segment people

18 people you give them 12 tasks you measure time you measure success rates and you basically get before and after and that's a clear indicator that something has either improved or not right so because you basically give them not just random task but representative task things that they are kind of critical for that application for that website and one thing that I think I really want to emphasize here usually you get to the point where you ask people hey tell us speak uh speak through your experience right as you are doing this tasks but that of course delays people because again when they start speaking this doesn't represent the reality so they need more time to complete task and that's just not what people do in real life so try to be as realistic as possible and then you get basically the results just in the end after a person has completed all the task we basically do a debriefing get some insights take some notes and we know what to do next time I'm curious and I'm throwing you a curveball here in the age of AI do you see a future where okay it could take something that you've done where you've labeled it a certain way you've given it certain rules and depending on who shows up it creates a version of the site just for them not just the way it looks but even the offers and the copies written differently so hyper targeted super narrow Niche what do you think about that well this is an interesting future we're talking about here I think you know what Chris I wouldn't mind that at all as long as people who are coming to that site or using the service a product can get their work done fast right and understand what is going on and it's all clear I think we can still create experiences that are pretty much accommodating for a vast majority of people I don't know if we need to go that route to really C hyper customize hyper personalize because at some point you also get to the situation where the system maybe understands the human better than the human does him himself or herself so it becomes a little bit you know freaky a little bit dangerous in a way right and I think it's it might not be necessary at times uh what is however necessary I think is simple Humane things this is also what I'm missing I mean we have conversations about Ai and of course AI is extremely important and of course this is a future and of course we need to understand as designers what to do with it and of course we probably shouldn't be afraid of it I think well we'll see what we'll be we'll know in a couple of years I'm sure but I also miss Humane touch I mean I don't know about you and everyone who's going to watch or listen this to this later sometimes I can smell CH GPT text I can just I don't know I just feel like there is something CH gpsh in it it's I cannot point the finger it's maybe like you know the bold and then you have the text and then a bit of text as well and it's a little bit too perfect and it's bit too clean in a way too sterile and so way that's not how the world is messy and complex and people make mistakes and typers and I want to see typos everywhere I mean it's not maybe in formal documents but I want people to be just respectful and kind and sometimes forget about things and sometimes remember things and sometimes be inconsistent we do not have to have 100% consistent experiences just for the sake of being consistent just make me I don't know surprised or smile or worried about that footer that is not aligned that's okay that's the web we can always interact and play and change and experiment