You build Miro, used by 100 million people worldwide. Would you say you do the same in 2025?
I would definitely focus on... This is Andrey Khusid, CEO and co-founder of Miro. He took a simple whiteboard idea and turned it into an $18 billion company. But in the past few years, the way he builds has flipped completely. Why? The one thing that we didn't predict is AI. And that's changed everything. Even for him, you can't predict from my perspective more than 12 months. What do you think is going to happen in 12 months? I don't know. AI made building easy. The world is now overflowing with products. If you want to build something big, you need to move really fast. You need to understand who you are, what you are passionate about. In the new era, the rules are simple and brutal. And only those who know the main secret will survive. The rest, they'll disappear.
Andre, welcome to Silicon Valley Girl. You build Miro, used by 100 million people worldwide. Let's talk about how entrepreneurship has been changing in the past few years. So Miro is this innovation workspace. But initially it was a mind map, right? It was a whiteboard.
We started with a simple idea of bringing a whiteboard into a browser. Pretty simple idea which grew to become an almost $18 billion company.
If somebody who's starting out today has a simple idea, how do they rationalize around how big it could get?
When I started the company, I hadn't thought about how big it can be. I was just thinking about how I can solve the problem that I have. Because before this business, I was running a creative agency. We had customers who were in the same city with us and then we had customers who were remote. I saw an opportunity to have that shared space where you can collaborate with customers who are remote, and that's how we came up with this simple idea of bringing a whiteboard into a browser. At that time, the only goal I had was to get to break-even as fast as possible. So I had a team of 10 people and we were trying to build this product, and then we were rebuilding it after we figured out some of the early signals that what we originally built was not working. Everything that we were doing was just trying to build the product that would get to the break-even point. Then we saw that a lot of people were quite excited about the product and we started to think how we can scale. But again, it was not like we had this ambition to build a multi-billion dollar company.
But have you ever had that ambition, or was it just 'I just want to...'?
Not really. We were passionate about the problem we're solving, and yes, I understood that the market can be quite big because there are a lot of knowledge workers in the world who can benefit from such a product. But it was not like I was going to sleep thinking how it can be 100 million users or 1 billion users. It was more about picking up the right problem, solving it, and solving it best in class. But also, you understand that the market can be quite big if you do it right. So the motivation back in the day was to build something that people would love using, that passion about the product and the product experience and frictionless experience. But obviously, when I'm building something, I'm thinking what is the market I'm playing in and do I need to shift towards the bigger market or not, because if you are playing in a small market, that can limit the growth of the business.
Do you remember the aha moment when you were like, 'Oh my god, this is growing, this could be a really big company'?
We saw initial growth in 2015 when we moved from Flash to HTML, and at that time I clearly saw a path to whatever 1 million, 5 million in revenue. It was great, we figured something out that could be quite a big business for ourselves at that time, and we were quite excited about that. But then in 2018, 2019, we started to see a path to 200 million. So that was quite clear that we figured out how to do enterprise sales and how to get into companies with our value proposition. And then when the pandemic started, it was quite clear that this would scale quite broadly. When the pandemic started, we had 5 million users globally, and then within 18 months it grew to 50 million.
Oh, that is crazy. So did it go down after the pandemic?
It flattened. So now we just passed 100 million users globally. It took us another 36 months to add another 50 million users on top of the 50 million we had in 2022. It's still growing quite fast, and we developed the platform quite a bit since then. We expanded from just being visual collaboration to what we call now an AI innovation workspace, where teams not just brainstorm or ideate, but where they move from original idea to the solution and to delivery end to end. So they can progress from one step to another, and now with AI it's a quite significant change because it's not just you do the steps manually, but actually you collaborate with AI and AI helps you progress from one step to another, so you can get to that outcome super fast. That's what we're passionate about now, because canvas is that modality for AI that can be quite powerful.
You grew to 100 million people because of the product. But then you figure out marketing strategy.
Originally, everything we were optimizing was product experience. So it's user experience, and we were optimizing for virality as much as possible. So people come to Miro and they start doing something, and then we were incentivizing them to invite other people to collaborate. It was a delightful experience because those folks who were interacting with Miro were like, 'Oh wow, this thing exists, and oh wow, what I can create with this.' So it activated word of mouth, people were sharing this with other folks. Then we added search optimization as a channel. Those three were our major growth channels. After several years, once we nailed those channels, we started to layer on top more intentional marketing and more intentional sales, but we had to build this organic flywheel originally.
Would you say you do the same in 2025 when AI makes building products so easy and every product looks kind of nice?
Yeah, I would definitely focus on product-market fit because it's now super cheap and fast to build product. The quality is questionable still; you need to invest quite heavily into making it a high quality product, but it's fast. The fundamentals stay the same: if your product is not solving the real problem, it will not grow fast. Brand matters more than ever. Trust, love for the brand, excitement about the brand — I think that's so important now.
Do you have any tips for finding product-market fit? Do you talk to your customers? Do you track particular metrics?
Now we are reinventing the next horizon for our company and for our business. I spend a lot of time with customers. There are different ways how you can explore product-market fit. It starts with what's the problem you are solving: is it a real problem, is it a big problem? Then you look at how big is the market on which you are solving that problem. So that should come together nicely. Once you've figured out this is the problem and this is the market, you go and have open-ended conversations with the customers: 'Hey, I heard this might be a problem for you. Can you elaborate why and what?' So you prove or disprove some of the hypotheses you have. You also build prototypes, because sometimes, especially in AI-first products and products which require a very special user experience, especially in productivity tools, people will not tell you 'I need this thing.' You have to come up with a solution. So you'd better build a prototype and put it in front of the customers and say, 'Hey, how does it feel? Does it solve your problem? Can you play with this?' How many customers? It depends, but in general, if you do deep quality interviews, it might be 7, 10, up to 20 customers, because then more or less you understand the signal.
I've heard you talk about the failure rate in your company, that you have a specific number in mind because if every experiment is successful, that means you're not experimenting enough. What's the failure rate and how did you end up with that?
In general, I would say you want to have a success rate of about 50 to 70%, and then you leave at least 30% for failure rate. And it's across not just product experiments or growth experiments, but also about acquisitions. If you are acquiring companies and 100% of your acquisitions are good, it means that maybe you're not pushing the boundaries enough. If a lot of them just fail, it's also not right. So it's super important to have a portfolio and to have some bets that are safer and make sense in terms of 'hey, this is how we can quite predict what the outcome would be,' but some bets should be moonshots, and those bets should fail.
How far do you push? Because sometimes you come up with an idea and it's not working, but after some tweaks and market research, you end up finding the right fit. Is there a strategy around that?
It's a great question. You always try to figure out: is it the problem that we're solving that is wrong, or is it the solution that we're playing with that is not perfect? If you believe the problem is there and if you believe this problem is solvable or solvable better than it's solved today, you have to go and iterate. It's totally fine. When I look at startups, a lot of them can do certain tweaks and get way stronger product-market fit. But some of them just stop, and their growth is very dependent on the strength of the product-market fit. So it's all about iteration on the solution. We have been building this for 14 years now. We redesigned our onboarding maybe hundreds of times. There's no right or wrong solution, but the time changes, the preferences of users change, even the user experience that you can create can change. We introduced a completely new interface, a separate mode called AI canvas, and this is a new set of capabilities. It's an experiment; we don't know if eventually it will merge into one experience or we will keep it as two separate experiences. But we zoomed out and thought, 'What's our day one thinking? If we create the product today, how should it look like?' That's how we came up with the solution. Now we'll see where the users gravitate towards, and then we will learn what works and what doesn't. We may kill that AI canvas, we may merge it with the core canvas. The fundamentals that we launch will remain, but it's more about how you position those capabilities as a separate mode or the same mode. What's prioritized? The previous experience versus new experience? You never know. That's what you have to learn.
Andrea mentioned something really important: you have to build tools that people trust and genuinely want to use. And honestly, trust is one of the most valuable currencies in tech right now. This really gets me thinking because I use so many AI tools every single day. My photos, my screenshots, my documents, even my voice goes into an AI app. The market is exploding with new services, and I'm always asking myself, 'Okay, I'm going to send this to this new app. I don't know who's producing it. What could happen? What might happen to my data?' Usercentrics did some research and it turns out many people feel the same way. According to their data, 62% of consumers feel like they've become the product. 77% don't understand how their data is handled, and 92% of Americans are concerned about their privacy when using the internet. This isn't just statistics; this is real consumer behavior and that affects the future of your product. If you run a website or online business, you might not even realize you're unknowingly breaking privacy laws. Most of us aren't lawyers, and the cost of getting it wrong can be devastating. That's where Cookiebot CMP comes in. It's an automated consent management solution that scans your website, identifies every cookie and tracker, and manages user consent at scale. It's Google certified and automatically updates when regulations change, covering GDPR, CCPA, ePrivacy, and more. Cookiebot integrates easily with WordPress, Shopify, Wix, and other popular platforms. Plans start at just $8 a month, and it's even free for small websites. If you're building something you want people to trust and use, check out Cookiebot by Usercentrics. I've got a special link in the description where you can get 15% off for six months. Thanks to Usercentrics for sponsoring this video. Now, let's get back to our conversation about building products people actually trust. So, you're changing the product with AI. Are you changing your marketing in the AI era?
We thought about that back in the day. I didn't expect there would be AI and commoditization of software building, but in general I believe that you have to stand out. That's how we came up with the Miro name, because originally the company was called RealtimeBoard, and that was quite literal. 'Board' — Miro is a great name.
Yes, ideal. For me, there are three types of company names. One is just a name, whatever, like RealtimeBoard, a descriptive name. Another would be a brand, how you can create a brand that people know and recognize. And then there is a lovemark. When I was thinking about this rebranding, my objective was to go from name to lovemark.
How do you define lovemark?
It's something that when you hear it, you have that kind of feeling. For us, this lovemark came from the inspiration of the artist Joan Miró. The idea was we want to be a canvas, the canvas that inspires people, where the creative mind is activated. Not yet another software, not yet another tool, but being that inspirational part of the day-to-day work.
I actually like how you plan in three-year missions. Is that right?
Yeah, we write a painted picture. We picked this practice from Atlassian back in the day. The idea is you sit down as a team and imagine what the future of the company and the product and our offering should look like, and you synthesize that in a couple of pages and share with the whole org, and then you go towards that. Last time we wrote this painted picture was 2022. It's now coming to the end, and actually we executed quite well against that painted picture. The one thing that we didn't predict is AI in that painted picture. So we're now rewriting the whole thing in terms of how AI fits into that vision. AI lends quite well into the vision, but also it starts to challenge some of the fundamentals of that vision.
2025 is when you're going to sit down and predict the next three years.
I'm not sure that we're going to do it this time. The reason is you can't predict from my perspective more than 12 months now.
What do you think is going to happen in 12 months?
