I Hit $1M ARR in 117 Days. Bootstrapped to $10M. Here's My Playbook | Chatbase, Yasser Elsaid

Intro

00:00We started now close to 3 years ago. I launched I think at like 1:00 p.m. I got my first customer 30 minutes after that. I think 10 minutes after I got the second person, maybe an hour after I got the third person. At this moment, I knew that I need to stop everything else I'm doing in life.

00:10We did 1 million ARR in exactly 117 days. We've been growing ever since uh bootstrapped without any external funding. It's just me and the team. That's it. My opinion is that you'll see an explosion of more and more bootstrapped companies doing insane numbers of revenue that were not possible before.

00:35Hi, my name is Yaser. I'm the founder of Chadbase. Chadbase is a platform for doing customerf facing AI agents specifically for customer support and sales. So, we're building an AI agent that is a brand ambassador and gives that conversational experience to every customer.

00:48We just bootstrapped Chatbase to $10 million in error. And now I'm going to share the exact playbook that got us here.

Why Bootstrap?

01:03Why did you bootstrap? I think there is many advantages and disadvantages of bootstrapping. The biggest advantage is that you have control. So you only listen to your customers and the team and that's it. Uh and I think this gives you a lot of freedom in building exactly what you want to build.

01:17And I think the other reason is the definition of success changes as soon as you raise. When you're bootstrapping like a successful outcome is like 50 millionaire or 100 millionaire or even 200 which I think is very possible for us. But I think raising changes that equation a lot.

01:35What I would consider a success might be even harder now to achieve as soon as you sign a term sheet. Getting a successful outcome for me, for the team and for the customer I think is a lot more likely in this path. And I think a lot of companies and founders are seeing that.

01:57Now you see a lot of bootstrap founders especially now with all the AI tools that you have now which you know help you with coding, help you with customer support, help you with marketing, you don't need as big of a team as you used to. So I think this era is the era of the smaller companies maybe like 10 to 50 people with a lot higher revenue per employee and when you have that uh equation bootstrapping becomes a lot more feasible.

02:22I think it's also more fun to to have control. So this is why I decided to to keep the company bootstrapped. What's the common mistake bootstrap founders make? When you operate as a bootstrap company,

The Common Mistake Bootstrap Founders Make

02:34you're more focusing on profitability. you're more focusing on everything you do is ROI positive. You're making sure you have like money in the bank. You're making sure to, you know, be as efficient as possible. The most common mistake bootstrap founders make is having the mindset of a bootstrap founder.

02:49That means being extremely, you know, like cost efficient with everything. Always trying to make sure you're ROI positive, being risk averse. But if you're like me and you want to build something huge and you want to have as much impact, hire people, see revenue go up, see number of customers go up, then the biggest mistake is not being aggressive enough.

03:13It makes sense to have that mindset early on. But I think you should very quickly stop having that you like as soon as you have some revenue you can rely on. You have to take risks. You have to like do experiments that might not be ROI positive.

03:25You have to you have to hire people that you think are expensive because they're good. You shouldn't delay that. This is honestly the biggest thing is is having a mindset of a boost company. And this is what we stopped doing and we've seen a lot of revenue growth since we started acting that way.

EO Partner Highlight

03:47What if your CRM kept up with the conversations that matter for you? Meet ATO, the AI CRM for modern businesses. Connect your email, calls, and product data, and Atoio instantly builds your CRM with complete context. Then ask anything, spot a deal at risk, draft outreach, uncover new prospects, spend less time in your CRM, and more time building relationships.

04:08Whether you're running product growth or enterprise sales, AIO adapts to how your team sells and grows. And if you're building with AI agents, plug any MCP clients straight into your CRM so your agents always have the context they need to act.

04:229,000 companies build with Atoio, including Granola, Modal, and Whisper Flow. Try Atoio free at aio.com/eo. Why did you become a builder? I started university in Canada studying

Why Become a Builder?

04:45computer science. You know, everyone is affected by their environment. The environment around me, everyone basically had the same goal actually like in most of universities in Canada, especially in computer science and engineering, and that's getting, you know, a fang internship uh in California.

04:56It's called Cali or Bust. I just started optimizing my work or my the the effort I'm putting towards getting that fang internship. I got it. I did an internship at Tesla. I did another one at Facebook. Everything is very new, so it's very exciting.

05:14But I think very quickly I did realize that maybe this is not the environment for me. The reason was like it just felt too, you know, like too safe, too structured. It's it's mostly optimizing for like the highest chance or like the highest likelihood of making like above $200,000 a year, which which is a fine goal.

05:38I think for me like I just was very inspired by people who are building companies and I think it's such a privilege to have the freedom you know like to wake up and decide hey like this is what I want to build I have this vision and I want to bring it to life.

05:49So while I was doing these internships, I very soon realized that maybe maybe this is not it. And I just saw if I continue in this path, what will my life look like in 5 years, 10 years? And it's honestly it looks exactly the same. Like you know the E3D engineer versus the E6 engineer, you know, that person has, you know, like a higher salary.

06:13They have more status within the company. I think this is this is of course like an amazing life and a lot of people would would want that. But I just wanted this ability to create things. When I thought of that, I just looked at, you know, what people are doing out there and I saw this community of indie hackers, especially on on Twitter.

06:31So people like Peter Levels or or Mark Lou, they have an idea and they sit down and build it and then they share it with the world. This gives them a lot of purpose and a lot of meaning. Like you imagine like you have an idea, you just like bring it into existence and then you also have people validate that by giving you money because this thing that you built is providing them value.

06:56When I did those things, it gave me a lot more skills that is actually needed when it came time to build chatbase. I don't think I used any of the things that I learned at those bigger companies, but I used a lot of the things that I learned when building those those smaller projects.

07:16And I think you'll see that even the most impressive founders, you'll see they always have like projects that they were building, some of them like very small oneperson projects, some of them bigger, but you always see like some hint that they started with an idea and they wanted to bring it to the world.

07:32And then by doing that over and over again, one they acquire the skill, two they acquire the confidence and then three they come up with better ideas for better products. And that's exactly what happened with me. You learn how to build a company by building a company.

07:45I think any skill that you learn at a big company is mostly irrelevant because it's just like a completely different dynamic, completely different goals, completely different kinds of people you're working with. If you know the goal is to like build this specific company, like the only way to get to that goal faster, is to just start.

08:02In building a company, especially early on, especially when you're solo, the outcome is like 100% correlated by the input you put in. You control all the inputs and all the output is a result of all of those inputs that you had control over.

08:22I think it's very apparent in when you're doing sports especially individual sports because you're responsible for you know your training your equipment your coaching your nutrition everything and your outcome is solely based on your input and your training you win because you won and you lose because you lost.

08:42Taking that mindset to building companies I think is is very beneficial. Knowing that you control your destiny gives you comfort. I knew I was smart enough to do this. I knew I had a lot of energy. I I had the conviction. So, it didn't make sense to me to not tie that skill and that energy to the outcome.

09:02I think for for anyone that has a similar mindset, starting a company is is a no-brainer. How did Chatbace start?

How Chatbase Started

09:12When I was building these other projects, some of them were using AI because this was like very early on in the start of the AI wave before Chad GPT was launched. So I was building things and when you're building things you see other people building things and you see some limitations in what you're building and you see some new ideas and new opportunities.

09:32So while building other things it became obvious to me that oh there is this one idea that it doesn't seem to me that anyone is doing but it was obvious to me that this is extremely valuable. This felt to me that this is if I don't do this now, someone else will because to me it was such an obvious idea.

09:50The idea basically was you have these powerful models, language models that are trained on all this general data, but they don't have specific data to a specific person or company or customer or like my my first example was a book. So the very first version of chatbase was uploading a textbook or uploading a book and then and then just chatting with it.

10:15I found that idea because I was doing other projects in the space and when you do that you see like the gaps you see the opportunities and then you come up with with new ideas. So I I I stopped going to class. This was in 2022 so before even Chad GPT was launched not mainstream like it is now and maybe in some circles they're talking about AI but of course not anywhere close to the same extent as it is now.

10:41It was just like hey we have this model called GPT3 and people are doing some interesting things with it. There's a few maybe companies that are using it. So back then I think a lot of people had like saw the models saw what they are now and thought of ideas of companies or products they can build on top of those AIs but based on the current capability.

11:05Now everyone knows you know like if you're building an AI you're building for the next model and the next model so that every time the model improves your product improves but I think that mindset was not what everyone was thinking back then when I saw this model like it has of course it was not not very powerful a lot of use cases were not very good but you can like hack in just enough to make it usable at least to do a demo at least to make it like useful for some customers and some use cases.

11:30To me, it was obvious that these models will improve. I didn't know how, I didn't know when, but if this model improves and I built the harness around it that makes it able to do like these these cool new things, then I'm winning and then my customers are also winning.

11:49I launched 1 p.m. I got my first customer 30 minutes after that in just, you know, like a pricing page on the website. I put out a tweet. I put a link and then I saw my first Stripe notification. I think yeah, I think that's such a special moment.

12:05You know, you built something from scratch and someone was willing to bet that this is valuable to them that they take out their card and spend money on it. I think 10 minutes after I got the second person, maybe an hour after I got the third person, at this moment I knew that I need to stop everything else I'm doing in life because I it was very obvious to me that this is this is this is a special moment.

12:29and this is a special opportunity. A lot of people don't have this opportunity and I just need to make sure that I take advantage of being in such a good position. So I stopped, you know, going to class. I stopped hanging out with friends. My main focus was was building.

12:46I think that was the right move.

How to Hit $1M ARR

12:52How did you hit a million in ARR? I didn't spend any money on marketing in the first 3 months because I didn't have money to spend on marketing. I like all the money uh that I has to saved up like personally even from like working and on and on the internships was mainly used to serve the models and then once I had enough revenue to cover the cost of the models then it started to become profitable and it became profitable ever since but to start like there was no marketing it was just building in public launching in subreddits launching on Twitter launching on LinkedIn talking

13:26about the product every day 100% organic it's good that I was forced to do it early on and not rely on, you know, like paid marketing or, you know, paid ads because this like paid ads and and paid marketing in general is very useful. But when you have the skill of organic marketing, one, you can build a brand around your company and about you as a person and then two, the efficiency of paid marketing becomes much much stronger.

13:54Having that constraint of not having money to spend on marketing made it actually so that when we had money that money went a lot further uh when we when we actually spend it on marketing. First month we were at 3,000 MR. I think I posted about being ramen profitable.

14:06I think second month it was close to 40 and then it was 60 and then we got to 1 million ARR 117 days in from that first tweet. It also helped me like make an intentional decision to not have chatbased as a lifestyle business. It is addicting to see the graphs go up.

14:29It is addicting to talk to customers and see how valuable the product is. It is addicting to hire people that you like and build stuff with them. I think getting there so quickly was so so so helpful to just like discard the idea of I'll just keep this as like you know a side lifestyle business and made me focus on making this a very big success early on.

14:490ero to a million versus a million to 10 million. What changes?

0→$1M vs $1M→$10M: What Changes?

14:55Like the problems you're facing when you're going from 0ero to 1 million are very different than the problems you're facing going from 1 to 10. And I imagine 10 to 100 is also going to be very different. 0 to one is more about finding what people want and like building that and like I don't think that's very hard.

15:07I think it's just about putting in the time and effort into talking to customers and like building and talking to them again. And it's just more about like you can brute force this. You can brute force this by putting in the work. I think going from 1 to 10 a lot of it is also there has to be like some you know like leadership skills you have to be you have to have some ability to to sell you have to have like ability to like explain your ideas explain your product clearly for some people maybe that's not very natural

15:37also a big part of it is building the culture building the team building the you know the as I said like the energy the incentives the processes that's just like a completely different game from going from 0 to one with like like maybe a threeperson team.

15:50That's not easy to do. Like a lot of people don't have the skills to do that. From what I've seen, going from 10 to 100 is also a completely different beast. But I think like a lot of the things you learn going from 1 to 10 is extremely helpful because like you have to have a lot more skills to be able to to have the performance to get there.

16:07How did you reduce churn early on? When

Reducing Churn Early

16:13we started our churn was high and the reason was especially at the start of the AI wave a lot of people were experimenting with these new tools they didn't know anything about like how capable these tools are what use cases they're good for and then the the company was also very new so a lot of features that we have now we didn't have back then as a result of that you you will have high churn the reason we don't have high churn now and the reason we we have customers stick with us for a long time is because they see the product improve every day.

16:43I think even if it's not something that that customer specifically needs or use, I think seeing like that there is this company cares enough to push out stuff every day to to customers and like this is this is how fast they're shipping. I think this sends a signal that like we're building this for you.

16:56We are putting effort into making this a better product. So that was one was just like making sure that we're shipping and then all of these improvements are visible to customers. I think that that helps a lot with reducing churn. The other aspect of it was just improving the product.

17:13Like when you start a product, especially when you get a lot of traction early on, that's not the best version of the product. The best version of the product is after you get feedback from the customers, after you talk to them, after you collect insights from all of these hundreds or thousands of customers, then you realize, oh, hey, like this is actually what I want to build.

17:29And then when new customers come in, they see like the problem that 10 other customers have faced is already solved. just listening to them and getting on calls with them, seeing how they use the product and making sure they're set up correctly, making sure they're live with the product in production and like seeing all the hurdle hurdles they faced along the way and making sure we remove all of them.

17:54I I don't think there is like, you know, the a specific like flow, cancellation flow or like hiding the cancel button or anything like that which will help with your churn. Honestly, I think it just comes comes down to how good your product is and how good is it at communicating its value to the customer.

18:10Early on, what we did was we built features, we like released them and then it's just very hard for people to see or use. Sometimes people would churn for a reason that we have already solved like there's a feature they want already exists, but back then our product didn't do a good job at communicating that feature or making it easy for them to use.

18:30So what we did is we just like overhauled our own onboarding flow to make sure that like there's different paths depending on like the persona of the customer. For every path we make sure that we surface all of these capabilities and like how to set it up uh as you're on boarding and then also we started to give them the option to talk to one of the team so that they feel like there's a person you know behind this website that cares enough to get on a call with them and help them get set up correctly.

19:01But even if for some reason you still couldn't, you at least know that there is a person that cares enough to get on a call with you and help you get get set up. It's just like consistent product improvement over a long time. You'll just see the the churn graph go down.

19:13Anything else that claimed to have like a big impact on churn that was not about improving the product was probably more of a waste of time. What are your thoughts on PLG? PLG is extremely powerful, but I think

Thoughts on PLG?

19:31it's just harder to do. It's actually much easier to do sales than to have a very powerful PLG engine that is always working and is always providing customers and making sure that those customers are successful. Like the the easy thing is to get on on sales calls, set up the product for the customer, make sure like the all the their questions are answered and that's it.

19:48That's the easy thing. The hard thing is to build a product that is extremely intuitive. people can just like go into it and sign up and without much effort be successful with it, especially if it's powerful. I think it's not smart to leave a lot of interest on the table without taking advantage of it for some reason like wanting to be PLG only.

20:09It is good to work with bigger companies. It's good to like help them on board. It's good to get on calls with them. It's good it's fine to have like the long sales processes if it means you're able to make sure that they're successful. So I think the reason a lot of companies start with salesled or even start with PLG but end up doing salesled at some point is because sales is just easier to do than like an extremely powerful product that's only growing from PLG.

20:35But I think there's a huge advantage of starting as a PLG product with self-s serve as the main interface and then from there going into enterprise and the reason is you're forced to make a good product because there is no one to like handhold your customers and like make sure they're successful because they're just like signing up on their own and like using the product on their own.

21:02So you're just kind of forced to have like a good intuitive simple but also powerful product to use and that's valuable whether or not you're doing sales. So having that foundation of like a strong self-s served platform and then on top of that adding like a salesled approach gives you extremely powerful products and that's why I think Stripe is ana like a very strong example of that.

21:27What's your SEO and AEO strategy? AEO. Um, a big part of it is just having

SEO & AEO Strategy

21:34strong SEO. Most of these models, they get their information by just doing a web search. And if you're able to rank for humans, well, you you'll be able to rank also for the AIS because the way it works is very similar. Like the page rank algorithm is very similar.

21:47You know, the basic SEO strategies still work. just do the SEO stuff, you know, like writing the blog posts, internal linking, external linking, making sure it's actually a good quality that people want to reference when they're talking about something in the same space.

21:59And I think also like organic content helps a lot with SEO because when you have a strong brand, people want to reference it in like when they're doing their own content. And that's such a big part of SEO. A big part of also AEO is is just doing the basic CRM but also like review websites or things like you know Reddit or user generated content or even like YouTube and Tik Tok those are also very important because everything is getting indexed by the when the model is trained and then also when the model is searching in runtime you just need to make sure that you exist everywhere and

22:38you're delivering the same message everywhere. You just need to have a message that is very clear to you and your team what's the value you're delivering, who the target customer is, and like what's your unique selling points. This is what we want people to think about us and then just like spray that everywhere as much as possible in all the uh surface area you can but also like when you find something is working like one maybe specific channel is working doing more of that makes a lot of sense.

23:02Now it's just one person on our team is most like just focused on SEO and AU. How does warm outbound work? Warm outbound is this idea that because

How Warm Outbound Works

23:16we started as a PLG first company because we're doing all of this content because we have all of this traffic on our website. We didn't start a sales letter. Like we didn't grow by, you know, cold calling customers and like convincing them to use us.

23:29We already have this massive amount of traffic and massive amount of following on organic channels. To boost it like a lot more, you add warm outbound. And basically what that means is you take a look at all the visitors that never sign up, visitors that do sign up, people who sign up, try the product but don't end up subscribing, people who subscribe but maybe like they stop using the product after a while.

23:54All of these are like the highest intent people like in the world that you currently have. You're not making sure that the the product is delivering the most value to them and you're not monetizing them the way that you can. Warm outbound is just making sure that all the effort you put into like getting all of this traffic like doing the SEO, doing the content is actually like pushing people in like that that last mile, making sure that they're successful.

24:18Like after they sign up, like you send them an email and say, "Hey, like we're here for you. who said you send them like a LinkedIn connection request say hey like let me know if you have any problem with the setup most of the time they're happy with that call because they see your content and like they feel like they know the brand and they have trust in the brand so that's what makes it warm and that's what makes it a lot more successful because cold is more about like numbers you know like you're spraying and praying that like hopefully like out of the 100,000 emails

24:51like I'll get two replies but this is more intentional like targeted outbound to a specific type of customer that already is invested in in in your company. For us, like it's 80% warm. So, a lot of it is like very personalized. A lot of it is like WhatsApp groups.

25:04A lot of it is like making friends with the customers and it is very hightouch and it's very like fulfilling also to our team because you get to like build relationships with those customers and like you get to actually solve their problems by just talking to them.

25:21How do you experiment with pricing? So when we started chatbas was more of a

Experimenting With Pricing

25:27B2C tool like you will upload textbooks or you upload books and then you chat with them. The pricing then was $10 and and I think $30 was the other plan. When we transitioned into B2B of course that changed. We still have like very inexpensive plans for maybe like very smaller companies or or single developers to try out the product.

25:44We started at like $19 uh as the lowest plan. We're now at $40 as the lowest plan. And then we also experimented with moving the highest self serve plan from 300 to 500. As we increased the pricing and we increased the pricing because the product just became more powerful and you're using more tokens for more reasoning and like you're solving much more complex issues when people saw the price increases.

26:09we we basically didn't have like any change in churn because people were seeing a lot more value from using the tools and like that's the the price increase. I think you want to make sure that like you treat your customers well and like you you're not just like trying to increase the price as much as you can all the time, but at the same time it is a good idea to experiment with the product.

26:29And also like experimenting with the pricing gives you a better idea on what type of customer is is getting the most value out of your business. So like you can talk to them more and understand their use case more. And as for like what pricing changed made the biggest impact, the biggest impact came from moving up market.

26:47When we did that, it was very obvious that this is the way we're growing from from this point onwards. Maybe the biggest mistake was not experimenting more. I've never seen a company like regret experimenting with their pricing, but I've seen many companies not experimenting enough for their pricing.

27:07The only balance that you want to make sure you hit is that you make sure that you still support your older customers that have used your product and have gotten value from it. Make sure that you find a balance where your customers are getting a lot more value than what you're charging them.

27:23The best pricing would be like you're just giving them so much value that they don't care about like the price they're paying you. Why revenue over margins? If you're choosing between like high

Revenue Over Margins?

27:34margins, less revenue or more revenue, less margins, better long-term play is higher revenue, more customers, even if it comes with with less margin. I think a year ago, we decided to buy billboards in San Francisco because we wanted to have more brand awareness here.

27:47It was not obvious how much of an impact that will make and like how much ROI that will give us. But some things, you know, are not measured by return on ad spend. Some things are just more like this is increasing awareness and this is going to help you in the long run.

28:05This is increasing um your brand capital in general. On paper, it will reduce your your margins like very drastically. Uh especially if you're like a a smaller company. These are the things that you need to be doing if the goal is to build something as big as possible.

28:18And some of the time it's not going to work. If you're smart enough, hopefully most of the time it will work. The job is to take calculated risks. Co-founder or solo. There's two different decisions people need to make when they're starting a company, whether you want to find a

Co-Founder or Solo?

28:37co-founder to work with or not. Uh and then whether you want to raise or not. Uh for both decisions, it depends. I think if you want to build a company and you don't have an idea of who you want as your co-founder, it's like you you want to still search.

28:49I think it's going to be very very hard to find like a co-founder. I think the best co-founders from what I've seen are people who have known each other for a while and they have, you know, like a lot of trust and confidence in each other's abilities.

29:01Of course, like there's some exceptions to every rule, but I think in general it's such a big risk. say, "Hey, I'm going to try to find a co-founder." And start working with this person. And a year in or like maybe even a few months in, you realize that this was not a good decision.

29:18Like having an amazing co-founder is better than being solo. Having a mediocre or maybe like slightly worse co-founder is probably a lot worse than than being solo. In my case, I didn't even have the time to think about whether or not I should have a co-founder.

29:35You just need to have extreme confidence in their values as a person but also their skills that complement you. When should you raise?

When to Raise

29:48It depends on what you define as a successful outcome. If your idea of a successful outcome is let's say like you exit and you get like 10 or or like even 50 million. I would argue that you have a higher chance of getting there if you bootstrap.

30:00If that's nothing to you and like your definition of success is a 500 million to like billions of dollars when you exit, then you need to raise because it's very hard to get to that outcome if you bootstrap. But I think it doesn't make sense if what you would define as success as like maybe 10, 30, 50 million outcome, but then you go and raise.

30:23I think it just makes it harder to achieve that because when you raise there's like a lot of variance because you you're you're trying to like 100x this investment and you need to like hire as fast as you can. You need to spend this money as as fast as you can if it makes sense.

30:36In most cases, you either like have an amazing outcome or like you're stuck in a place where it doesn't really make sense to sell the company for 100 million because after like the preference tag, after the all the terms, the dilution, you're not going to get that outcome.

30:53And I would say getting to a hund00 million outcome bootstrapped might be even easier than getting to billions of dollars after you raise.

Decision-Making Framework

31:05What's your decision-m framework? A framework for how you make decisions in life. Things change, right? Like especially in if you're building a company in a fastmoving industry, a lot of competitors, a lot of like new models released, a lot of um market dynamics change.

31:18These inputs are all factors into how you make a decision. A lot of people have a lot of, you know, pride and ego behind the decisions they're making, especially if they make that decision publicly because you don't want to look bad, you know, in front of your employees or in front of your team when you go back on a decision you made.

31:34But I think it's actually like it is a sign of low self-confidence. If you make a decision, you re like some time passes, the inputs change, you realize that this is not the best decision moving forward, but your ego is not allowing you to go back on that decision.

31:57And I think that's a recipe for a company to fail. And I think this is the case for founders, but it's also the case for everyone on the team. Everyone should know that dynamics change very quickly and we need to make sure that we're not sticking to a decision we made like two weeks ago because we don't want to hurt someone's feelings.

32:14Every day we wake up and we see the inputs. We know like why you made the decision two weeks ago. But you also know what changed and based on whatever changed does it make sense to do something else? And if the answer is yes, in 100% of the times you need to do that something else.

32:34And I think it's the job also of the founder to instill a culture where like going back on a bad decision is encouraged when you're building in such a fastmoving industry. Maybe it's useful to ignore the noise. Like if you look at all the people that fell into the trap of listening to what everyone else is saying in the space about like you know GPT rappers and like how there's no future so they stopped building but that was not the right idea.

33:04Now GPT rappers have rebranded to model harnesses and now like they're all the hype you know and now like you see companies that are have raised they're now like are doing more than 100 million in ARR in 3 years. you you could have like if they listen to the people that were saying, "Oh, this is a GPT rapper, then maybe they're not going to be in this position."

33:21Now, takeaway here is to just like people are going to speculate, maybe they're right, maybe they're wrong, but you don't lose anything if you just put your head down and just like work on building something valuable. I'm happy that I did that 3 years ago.