#116: Turning Weird User Actions into Big Wins with Gojko Adzic

What do lizards have to do with product growth? In this episode, Gojko Adzic reveals how unusual user behaviors can unlock massive opportunities for product innovation. Discover the four steps to mastering "Lizard Optimization" and learn how you can turn strange user actions into game-changing insights. Overview In this episode of the Agile Mentors Podcast, host Brian Milner chats with Gojko Adzic about his new book, Lizard Optimization. Gojko explains the concept of finding product growth signals in strange user behaviors, sharing examples where unexpected user actions led to product breakthroughs. He outlines a four-step process for optimizing products by learning, zeroing in, removing obstacles, and double-checking. Gojko also discusses helpful tools like session recorders and observability tools that can enhance product development by uncovering and addressing unique user behaviors. References and resources mentioned in the show: Gojko Adzic 50% OFF Lizard Optimization by Gojko Adzic Mismatch: How Inclusion Shapes Design by Kat Holmes Trustworthy Online Experiments by Ron Kohavi Advanced Certified Scrum Product Owner® Subscribe to the Agile Mentors Podcast Join the Agile Mentors Community Want to get involved? This show is designed for you, and we’d love your input. Enjoyed what you heard today? Please leave a rating and a review. It really helps, and we read every single one. Got an Agile subject you’d like us to discuss or a question that needs an answer? Share your thoughts with us at [email protected] This episode’s presenters are: Brian Milner is SVP of coaching and training at Mountain Goat Software. He's passionate about making a difference in people's day-to-day work, influenced by his own experience of transitioning to Scrum and seeing improvements in work/life balance, honesty, respect, and the quality of work. Gojko Adzic is an award-winning software consultant and author, specializing in agile and lean quality improvement, with expertise in impact mapping, agile testing, and behavior-driven development. A frequent speaker at global software conferences, Gojko is also a co-creator of MindMup and Narakeet, and has helped companies worldwide enhance their software delivery, from large financial institutions to innovative startups. Auto-generated Transcript: Brian (00:00) Welcome in Agile Mentors. We're back for another episode of the Agile Mentors podcast. I'm with you as always, Brian Milner. And today, very special guest we have with us. have Mr. Goiko Atshich with us. I hope I said that correctly. Did I say it correctly? Close enough. Okay. Well, welcome in, Goiko. Glad to have you here. Gojko (00:15) Close enough, close enough. Brian (00:21) Very, very, very happy to have Goiko with us. If you're not familiar with Goiko's name, you probably are familiar with some of his work. One of the things I was telling him that we teach in our advanced product owner class every time is impact mapping, which is a tool that Goiko has written about and kind of come up with on his own as well. Gojko (00:21) Thank you very much for inviting me. Brian (00:47) But today we're having him on because he has a new book coming out called Lizard Optimization, Unlock Product Growth by Engaging Long Tail Users. And I really wanted to talk to him about that and help him explain, have him explain to us a little bit about this idea, this new concept that his new book is about. So, Goiko, let's talk about it. Lizard Optimization, in a nutshell, what do you mean by that? What is it? Gojko (01:14) We're going to jump into that, but I just need to correct one of the things you said. I think it's very, very important. You said I came up with impact mapping and I didn't. I just wrote a popular book about that. And it's very important to credit people who actually came up with that. It's kind of the in -use design agency in Sweden. And I think, you know, they should get the credit for it. I literally just wrote a popular book. Brian (01:19) Okay. Gotcha. Gotcha, gotcha. Apologies for that incorrect. Thank you for making that correction. So lizard optimization. Gojko (01:44) So, lizard optimization. Good. So, lizard optimization is an idea to find signals for product ideas and product development ideas in strange user behaviors. When you meet somebody who does something you completely do not understand, why on earth somebody would do something like that? Brian (02:03) Okay. Gojko (02:11) and it looks like it's not done by humans, it looks like it's done by somebody who follows their own lizard logic, using stuff like that as signals to improve our products. Not just for lizards, but for everybody. So the idea came from a very explosive growth phase for one of the products I'm working on, where it... had lots of people doing crazy things I could never figure out why they were doing it. For example, one of the things the tool does is it helps people create videos from PowerPoints. You put some kind of your voiceover in the speaker notes, the tool creates a video by using text to speech engines to create voiceover from the speaker notes, aligns everything and it's all kind of for you. People kept creating blank videos and paying me for this. I was thinking about why on earth would somebody be creating blank videos and it must be a bug and if it's a bug then they want their money back and they'll complain. So I chased up a few of these people and I tried to kind of understand what's going on because I originally thought we have a bug in the development pipeline for the videos. So... I started asking like, you know, I'm using some, I don't know, Google slides or, you know, keynote or whatever to produce PowerPoints. Maybe there's a bug how we read that. And the person, no, no, we, know, official Microsoft PowerPoint. They said, well, can you please open the PowerPoint you uploaded? Do you see anything on the slides when you open it? And the person, no, it's blank. Right? Okay, so it's blank for you as well. I said, yeah. So. Brian (03:48) Yeah. Gojko (03:54) What's going on? so what I've done is through UX interviews and iterating with users and research, we've made it very, very easy to do advanced configuration on text -to -speech. And it was so much easier than the alternative things that people were creating blank PowerPoints just to use the text -to -speech engines so they can then extract the audio track from it. Brian (03:54) Yeah, why? Gojko (04:23) and then use that and it was this whole mess of obstacles I was putting in front of people to get the good audio. It wasn't the original intention of the tool. It wasn't the original value, but people were getting unintended value from it. And then I ended up building just a very simple screen for people to upload the Word document instead of PowerPoints. And it was much faster for users to do that. A month later, there was many audio files being built as videos. Two months later, audio... production overtook video production. then at the moment, people are building many, many more audio files than video files on the platform. So it was an incredible growth because of this kind of crazy insight of what people were doing. kind of usually, at least kind of in the products I worked on before, when you have somebody abusing the product, product management fight against it. There's a wonderful story about this in... Founders at work a book by Jessica Livingston and she talks about this kind of group of super smart people in late 90s who Came up with a very very efficient Cryptography algorithm and a way to compute the cryptography so they can run it on low -power devices like Paul pilots Paul pilots were you know like mobile phones, but in late 90s and Then they had to figure out, how do we monetize this? Why would anybody want to do this? So they came up with the idea to do money transfer pumping, Palm pilots, you know, why not? And kind of the built a website. This was the late nineties as a way of just demoing this software to people who didn't have a Palm pilot device next to them. The idea was that you'd kind of see it on the website, learn about it, then maybe download the Palm pilot app and use it in anger. People kept just using the website, they're not downloading the Palm Pilot app. So the product management really wasn't happy. And they were trying to push people from the website to the Palm Pilot app. were trying to, they were fighting against people using this for money transfer on the web and even prohibiting them from using the logo and advertising it. They had this whole thing where nobody could explain why users were using the website because it was a demo thing. It was not finished. It was not sexy. It was just silly. And Jessica kind of talks to one of these people who insists that it was totally inexplicable. Nobody could understand it. But then a bit later, they realized that the website had one and half million users and that the Pongpilot app had 12 ,000 users. So they kind of decided, well, that's where the product is really. And that's like today, people know them as PayPal. They're one of the biggest payment processes in the world because kind of, you know, they realized this is where the product is going. And I think in many, many companies, people Brian (07:03) Ha ha. Gojko (07:18) stumble upon these things as happy accidents. And I think there's a lot more to it. We can deliberately optimize products by looking for unintended usage and not fighting it, just not fighting it. just understand this is what people are getting as value. And I think for me as a solo product founder and developer and product manager on it, One of the really interesting things is when you have somebody engaging with your product in an unexpected way, most of the difficult work for that user is already done. That person knows about you, they're on your website or they're using your product, the marketing and acquisition work is done. But something's preventing them from achieving their goals or they're achieving some value that you did not really know that they're going to achieve. you know, that's something the product can do to help them and remove these obstacles to success. So that's kind of what lizard optimization is making this process more systematic rather than relying on happy accidents. And by making it more systematic, then we can help product management not fight it and skip this whole phase of trying to fight against our users and claim that users are stupid or non -technical or... They don't understand the product, but they're trying to figure out, well, that's what the real goals are. And then following that. Brian (08:47) That's awesome. So the pivot, right? The pivot from here's what we thought our problem was we were solving to now here's what we're actually solving and we should organize around this actual problem, right? Gojko (09:02) or here's what we're going to solve additionally. This is the problem we've solved, but hey, there's this problem as well. And then the product can grow by solving multiple problems for people and solving related problems and solving it for different groups of people, for example. And that's the really interesting thing because I think if you have a product that's already doing something well for your users and a subset of them are misusing it in some way, then kind of... Brian (09:04) Yeah. Gojko (09:30) The product might already be optimized for the majority of users, but there might be a new market somewhere else. So there might be a different market where we can help kind of a different group of users and then the product can grow. Brian (09:43) Yeah, I like to focus on the user. There's an exercise that we'll do in one of our product owner classes where we have a fake product that is a smart refrigerator. And one of the exercises we try to get them to brainstorm the different kinds of users that they might have for it. And one of the things that always comes out in that class is as they're going through and trying to describe the types of users, they inevitably hit to this crossroads where they start to decide Well, yes, we're thinking of this as a home product, something for people to use in their homes. But then the idea crosses their mind, well, what about commercial kitchens? What about people who might use this in another setting? And it's always an interesting conversation to say, well, now you've got a strategic choice to make, because you can target both. You can target one. You can say, we're ignoring the other and we're only going in this direction. So to me, I think that's kind of one of the interesting crossroad points is to say, how do I know when it's time to not just say, great, we have this other customer segment that we didn't know about, but actually we should start to pivot towards that customer segment and start to really target them. Gojko (11:03) Yeah, think that's a fundamental question of product development, isn't it? Do you keep true to your vision even if it's not coming out or if something else is there that's kind more important than I think? For me, there's a couple of aspects to that. One is, laser focus is really important to launch a product. You can't launch a product by targeting... the whole market and targeting a niche type, figuring out, you know, user personas, figuring out like really, really, this is the product who we think the product, this is the group who we think the product is for and giving them a hundred percent of what they need is much better than giving 2 % to everybody because then the product is irrelevant. But then to grow the product, we need to kind of grow the user base as well. And I think one of the things that... is interesting to look at and this comes from a book called Lean Analytics. It's one of my kind of favorite product management books is to look at the frequency and urgency of usage. If you have a group that's kind of using your product, a subgroup that's using your product very frequently compared to everybody else, that might be kind of the place where you want to go. The more frequently, the more urgently people reach for your product when they have this problem. the more likely they are going to be a good market for it. with kind of another product that I've launched in 2013, we originally thought it's going to be a product for professional users. And we aimed at the professional users. And then we found that a subcategory that we didn't really expect, were kind of teachers and children in schools. we're using it a lot more frequently than professional users. And then we started simplifying the user interface significantly so that it can be used by children. And it's a very, very popular tool in schools now. We are not fighting against other professional tools. We were kind of really one of the first in the education market there. And it's still a very popular tool in the education market because we figured a subgroup that's using it very frequently. Brian (13:14) Hmm. Yeah, that's awesome. How do you know when, you know, what kind of threshold do you look for to determine that, this is, because, you know, in your book, you're talking about, you know, behaviors that are not normal, right? People using your product in a way that you didn't anticipate. And what kind of threshold do you look for to that says, hey, it's worth investigating this? You know, I've got this percentage or this number of people who are using it in this strange way. At what point do you chase that down? Gojko (13:49) I think it's wrong to look at the percentages there. I think it's wrong to look at the percentages because then you get into the game of trying to justify economically helping 0 .1 % of the users. And that's never going to happen because what I like about this is an idea from Microsoft's Inclusive Design and the work of Kat Holmes who wrote a book called Mismatch on Brian (13:52) Okay. Gojko (14:17) assistive technologies and inclusive design for disabled people. And she talks about how it's never ever ever going to be economically justified to optimize a product to help certain disabilities because there's just not enough of them. And there's a lovely example from Microsoft where, Microsoft Inclusive Design Handbook where they talk about three types of, Brian (14:34) Yeah. Gojko (14:44) disabilities, one are permanent. So you have like people without an arm or something like that. And I'm going to kind of throw some numbers out now, order of magnitude stuff. I have these details in the book and there's kind of the micro -inclusive design handbook. Let's say at the moment, the 16 ,000 people in the U .S. without one arm or with a disabled arm. And then you have these kind of situational disabilities where because of an occupation like you have a bartender who needs to carry something all the time or a worker who does it, one arm is not available and they only have one arm to work on and this temporary like a mother carrying a child or something like that. So the other two groups are order of magnitude 20 -30 million. We're not, by making the software work well with one hand, we're not helping 16 ,000 people, we are helping 50 million people. But you don't know that you're helping 50 million people if you're just thinking about like 16 ,000. I think they have this kind of, one of the key ideas of inclusive design is solve for one, kind of help, design for one, but solve for many. So we are actually helping many, many people there. So think when you figure out that somebody is doing something really strange with your product, you're not helping just that one person. Brian (15:45) Right, right. Hmm. Gojko (16:13) you're helping a whole class of your users by making the software better, removing the obstacles to success. this is where I, you know, going back to the PowerPoint thing I mentioned, once we started removing obstacles for people to build the audios quickly, lots of other people started using the product and people started using the product in a different way. And I think this is a lovely example of what Bruce Torazzini talks about is the complexity paradox because He's a famous UX designer and he talks about how once you give people a product, their behavior changes as a result of having the product. So the UX research we've done before there is a product or there is a feature is not completely relevant, but it's a changed context because he talks about people have a certain amount of time to do a task. And then when they have a tool to complete the task faster, they can take on a more complicated task or they can take on an additional task or do something else. I think removing obstacles to use a success is really important. Not because we're helping 0 .1 % of people who we don't understand, but because we can then improve the product for everybody. And I think that's kind of the magic of lizard optimization in a sense, where if we find these things where somebody's really getting stuck. but if we help them not get stuck, then other people will use the product in a much better way. And I think this is, know, the name lizard optimization comes from this article by Scott Alexander, who talks about the lizard man's constant in research. And the article talks about his experiences with a survey that combined some demographic and psychological data. So they were looking at where you live and what your nationality is and what gender you are and then how you respond to certain psychological questions. he said, like there's about 4 % of the answers they could not account for. And one person wrote American is gender. Several people listed Martian as nationality and things like that. some of these, he says some of these things will be people who didn't really understand the question. they were distracted, they were doing something else, or they understood the question but they filled in the wrong box because, know, the thick thumbs and small screens, or they were kind of malicious and just, you know, wanted to see what happens. when you kind of add these people together, they're not an insignificant group. kind of, he says 4%. And if... we can help these people, at least some of these people, and say reduce churn by 1%. That can compound growth. Reducing churn, keeping people around for longer is an incredible way to kind of unlock growth. going back to what we were talking about, some people might be getting stuck because they don't understand the instructions. Some people might be getting stuck because they're using the product in a way you didn't expect. And some people might just like not have the mental capacity to use it the way you expected them to be used. But if we can help these people along, then normal users can use it much, easier. And you mentioned a smart fridge. I still remember there was this one wonderful bug report we had for my other product, which is a collaboration tool. we had a bug report a while ago. that the software doesn't work when it's loaded on a fridge. And it's like, well, it was never intended to be loaded on a fridge. I have no idea how you loaded it on a fridge. It's a mind mapping diagramming tool. It's intended to be used on large screens. Where does a fridge come in? And then we started talking to this person. This was before the whole kind of COVID and work from home disaster. The user was a busy mother and she was kind of trying to collaborate with her colleagues while making breakfast. breakfast for kids and kind of running around the kitchen she wasn't able to kind of pay attention to the laptop or a phone but her fridge had a screen so she loaded the software on the fridge and was able to kind of pay attention to collaboration there and you know we of course didn't optimize the software to run on fridges that's ridiculous but we realized that some people will be using it without a keyboard and without a mouse and then we kind of restructured the toolbar, we made it so that you can use it on devices that don't have a keyboard and then the whole tablet thing exploded and now you get completely different users that don't have keyboards and things like that. I think that's where I think is looking at percentages is a losing game because then you start saying, but 0 .1 % of people use this. But yeah, I think lizard optimization is about using these signals to improve the products for everybody. Brian (21:30) That's a great example. I love that example because you're absolutely right. You're not trying to necessarily solve that one problem because you don't anticipate there's going to be a lot of people who are going to want to run that software on a fridge. However, the takeaway you had from that of, we can do this for people who don't have a keyboard or a mouse. There's another way that they might operate this that could apply to lots of different devices and lots of different scenarios. Now we're talking about a much bigger audience. Now we're talking about opening this up to larger segments of the population. I love that. I think that's a great example. I know you talk about that there's kind of a process for this. Help us understand. You don't have to give away the whole candy story here from the book, but help us kind of understand in broad, terms what kind of process people follow to try to chase these things down. Gojko (22:26) So there's like a four step process that's crystallized for me. And the book is kind of more as a, like a proposal or a process. It's something that works for me and I'm hoping that other people will try it out like that. So it might not necessarily stay like that in a few years if we talk again. And I've narrowed it down to four steps and kind of the four steps start with letters L, Z, R and D. Lizard. And it's kind of so learn how people are misusing your products, zero in on one area, on one behavior change you want to improve, then remove obstacles to use a success and then double check that what you've done actually created the impact you expected to make. I think kind of when we look at people who follow their own logic or people who follow some lizard logic you don't really understand, by definition they're doing something strange. your idea of helping them might not necessarily be effective or it might not go all the way or it might. So double checking at the end that people are actually now doing what you expect them to do or doing something better is really, really, really important. And then using signals from that to improve the kind of feedback loop is critical. I had this one case where people were getting stuck on a payment format entering tax details and The form was reasonably well explained. There was an example in the forum how to enter your tax ID and people were constantly getting stuck. A small percentage of people was getting stuck on it. However, I don't want to lose a small percentage of people that want to pay me on the payment form. So I thought, well, how about if I remove that field from there? I speed it up for everybody and then I can guide them later into entering the tax details to generate an invoice. I thought that was a brilliant idea. tested it with a few users. Everybody loved it, so I released it. And then a week later, I realized that, yes, I've sold it for the people that were getting confused, but I've ended up confusing a totally different group of people that expects the tax fields there. So the net effect was negative. then I went back to the original form. so there's lots of these things where people don't necessarily behave the way you think they will. Brian (24:38) Hahaha. Gojko (24:48) Ron Kohavi has a wonderful book about that called Trustworthy Online Experiments. And he has data from Slack, from Microsoft, from Booking .com and... The numbers are depressive. on one hand, the numbers range from 10 to 30, 40 % success rate for people's ideas. And if leading companies like that do things that don't pan out two thirds of the time, then we have to be honest building our products and say, well, maybe this idea is going to work out, maybe not. Brian (25:03) Hahaha. Wow. Gojko (25:30) the more experimental the population is, the more risky that is. think monitoring and capturing weird user behaviors, capturing errors helps you understand that people are getting stuck. as you said, you don't want to follow everybody. There's going to be a lot of noise there. We need to extract signals from the noise. That's what the second step is about, focusing on one specific thing we want to improve. Then, try to remove obstacles and then double -checking that we've actually removed them. That's the four steps. And there's like a shorter version of all the four steps. It's easier to remember. It's listen alert, zooming, rescue them, and then double check at the end. that's again, LZRD. Brian (26:13) That's awesome. Yeah, I love the process and I love the kind of steps there. Are there tools that you recommend for this that are easier to try to determine these things or chase them down or are there tools that you find are more helpful? Gojko (26:32) So there's lots of tools today for things like A -B testing and looking at experiments and things that are very helpful to do this scale. And it's kind of especially useful for the last step. In terms of kind of focusing and things like that, the five stages of growth from the linear analytics are a good tool. Impact mapping is a good tool. Kind of any focusing product management technique that says, well, these are the business goals we're working on now, or these are the kind of user goals we're working on now. out of, know, 50 lizards we found last week, these three lizards seem to be kind of in that area. And for the first step, spotting when people are getting stuck, there's a bunch of tools that are interesting, like session recorders for web products. There's one from Microsoft called Clarity that's free. There's another called Full Story that's quite expensive. There's a couple of open source one, one is packaged within Matomo analytics application. There's a bunch of these other things. Any kind of observability or monitoring tool is also very useful for this because we can spot when people are getting stuck. One of the things I found particularly helpful is logging all user errors. When a user does something to cause an error condition in a product, the product of course tells them like, know, an error happened. But then... logging it and analyzing that information in the back is really critical. for something like that, people sometimes use web analytics tools or any kind of product analytics. I think what's going to be interesting in the next couple of years, and I think if people start doing this more, is we'll see. more like these technical exception analytics tracking tools mixed with this because most of the product analytics are showing people what they expect to see, not what they don't expect to see. And I'll just give you an example of this way. was really helpful. So I've mentioned the screen where people can upload the Word documents. Occasionally people would select weird file types. So they'll select images, they'll select, I don't know, what else. Brian (28:31) Yeah. Gojko (28:49) Sometimes I guess that's a result of, know, a fat finger press or somebody not selecting the right thing. I have a not insignificant percentage of users every day that try to upload Android package files into a text -to -speech reader. Android package files and application files, I don't know what the right way is to read out an Android application. My best guess is people are doing that. as a, you know, these things where you drop a USB in front of an office and somebody kind of mistakenly plugs it in. So maybe they're hoping that I'll know the Android application on my phone just because they've uploaded it. I don't know, but a small percentage of users was trying to upload files that had SRT and VTT extensions, which are subtitle files. And they were not supported, but Brian (29:31) Yeah. Gojko (29:45) I kept getting information that people are uploading those types of files. And then I said, well, this is interesting because it's a text to speech system. People are uploading subtitle files, there's text in, so why don't I just ignore the timestamps and read the text? I can do that. And I started supporting that. And then some people started complaining that, well, the voice is reading it slower than the subtitles. I said, well, yes, because... Brian (30:11) Ha Gojko (30:12) You know, you're uploading subtitles that were read by an actor in a movie. This is a voice that's reading it at their speed. And then we started talking and it turns out that these people were doing it for corporate educational videos where they have a video in English, they need it in French, German, Spanish and all the else, but they don't want to kind of re -edit the video. They just want an alternate audio track. Okay, I mean, I have the timestamps, we can speed up or slow down the audio, it's not a big deal. And we've done that and this was one of the most profitable features ever. Like a very small percentage of the users need it, but those that need it produce hundreds of thousands of audio files because they translate the corporate training videos. And now, you know, we're getting into that numbers game. If I said, you know, there's like 0 .1 % of people are uploading subtitle files. Brian (30:58) Yeah. Gojko (31:07) then it doesn't matter. if we start thinking about, this is potentially interesting use case, it creates growth on its own because then people find you. And I think my product was the first that was actually doing synchronous subtitles. Competitors are doing it now as well. But it opened the massive, massive market for us. And people, you know, I got there by monitoring user errors, by, you know, the fact that somebody uploaded a file that had an unsupported extension. That was our insight. Brian (31:38) Wow, that's really cool. That's a great story. This is fascinating stuff. And it makes me want to dive deeper into the book and read through it again. But I really appreciate you coming on and sharing this with us, Goiko. This is good stuff. Again, the book is called, Lizard Optimization, Unlock Product Growth by Engaging Long Tail Users. And if I'm right, we talked about this a little bit before. We're going to offer a discount to to the listeners, Gojko (32:07) Yes, we will give you a listen as a 50 % discount on the ebook. the ebook is available from Lean Pub. If you get it from the discount URL that I'll give you, then you'll get a 50 % discount immediately. Brian (32:24) Awesome. So we'll put that in our show notes. If you're interested in that, you can find the show notes. That's a great deal, 50 % off the book and it's good stuff. well, I just, I can't thank you enough. Thanks for making time and coming on and talking this through your book. Gojko (32:40) Thank you, it was lovely to chat to you.