$100 Million ARR Pivot: From Platform Product to Vertical Apps With Treasure Data CEO Kazuki Ohta

Thank you everyone for joining my session titled 100 million ARR pivot from platform product to vertical apps I'm kazuki ota, I'm the founder and CEO of the company called treasure data and I was formerly cto but I became a CEO from june 1 this year. Foreign. I am going to talk about the pivot we had around year 4 to 5. So I started this company around 2011. So it has been almost 10 years and we grew from 0 to 100 million ARR plus in last 10 years. As you can see for the first 4 to 5 years we've been growing nicely around 0 to 20 million data ARR. But since then we grew from 25 to 100 million dollar plus in 3 years. This involved with the one huge pivot from becoming the technology oriented platform product, which is the big data platform product to the marketing application product. And today I'm going to talk about a lot of struggle and tough decisions we made. And we had like 15 minutes but really the presentation is about 30 to 35 minutes and I would like to have a lot of Q and A. So please ask any questions via Zoom and I would appreciate and I want this session to be more interactive. So let's talk about how we came from 0 million ARR to $10 million ARR for the first three years. First of all let me talk about my background. So my background is computer scientist. I was specialized in HPC high performance computing. I was a visiting researcher at Oregon National Laboratory in Chicago and they have a supercomputer called Bluegen which combines almost a half million compute nodes into one and create one of the world's largest and fastest supercomputer at that time. And they run a lot of nuclear simulation, aerodynamic simulation, various scientific simulation programs. And these programs generate a lot of data. I was fortunate enough to become a part of the team member to store and process large amount of data with that supercomputer. Around that time I realized the power of data and I also thought that okay, this type of environment can be only purchased from Department of Energy or the government, but this environment really really needs to be democratized to a lot of other companies so that they can have the power of the data. And I was also interested in another open source software called Hadoop and Hadoop was open sourced by Yahoo. But the software itself is really really modeled after how Google is processing large amount of the data. So I started committing some patches to Hadoop project and more importantly I started organizing the community in Japan where I was born around Hadoop and it was just a three people nerd community, but it expanded into 3,000 people community in two years. And I was fortunate enough to become a founder of that community. And within that 3,000 people community, I was talking with a lot of people who is trying to analyze the data in various industries and learn. There were three key common problems we need to solve. One, it takes time to analyze the data. It takes like almost 18 months to start analyzing the data. And, and the success ratio was only 30%. And two, you have to hire a lot of computer scientists and PhDs to do it. But at the time, Facebook, Twitter, those companies can do it. But more traditional companies cannot maintain or even hire these type of people and team. And three, you have to have a lot of upfront investment. You have to have a lot of huge data appliances, their warehouse appliances, network gears and hardware and disk drives and those cost like a lot. And you need to have a lot of upfront investment just to start the project. So those two, three key problems, I was just thinking, okay, why don't we host this software called Hadoop in the cloud so that people can one, instantly analyze the data. Two, you don't have to hire a lot of computer scientists and three, you don't need any upfront investment. You can just subscribe and having the monthly recurring payment and then start analyzing the data. So that was really simple. But key idea I had around 2011 and I was 25, nothing to lose. I came to Silicon Valley and then started raising the additional the angel round together with my co founder Hiro and Sada. But the investors were like who is throwing away the data to the cloud? And I guess everyone does right now. But that was sort of the biggest bet we had. And cloud is now the mainstream. But around that time it was not popular. And so that was the original idea. And those two slides were actually coming from our series A investor pitch. But long story short, we grew from 0 to 5 million ARR in two years. But in reality we closed 3.6 million and 1.2 million ARR deal led by founders. So if you look at this, 96% comes from these like two gigantic deals led by founders. And if you look at on the left side, the green is the recognized revenue, so it's going nicely. The blue is the booking. So if you can see, we closed $1.21, $3.2 million ARR deal and the ARR in terms of revenue is growing nicely. Right. And at this point, okay, we decided to raise additional Series A round and we decided to do more, you know, pipeline generation, hire more sales, work with more channels. Right. And although we released the product around 2011 Q4, within 6 quarters we we become 5 million committed ARR company. And more bookings are coming from these large deals led by founders. Within this six quarter period of time the product gets a little bit matured. Our product consists of three product lines. The data collection, data warehouse, data analysis. The FluentD is invented. This is the open source data collector. And we were collecting more than 80 billion records per month into the product around this time. We also ingest a lot of data in the bulk version. And in terms of data warehouse we built a really scalable columnar storage with the Hadoop MapReduce data processing job, really storing 360 billion records, executing almost 2 million jobs in total. Then we made a lot of connection with the BI tools and dashboard product. What you can really do is collecting a lot of data and processing the data and then visualizing it. That was really the product. Around this time we really didn't know how to call ourselves but we positioned ourselves as a big data platform as a service. At this point, if you look at on the right side the competitive landscape in the data analytics stacks we had a lot of layers. The data storage layer, data collection layer, Hadoop distribution and processing platform. So our products is designed for, okay, we become the cloud one stop shop so that our customers don't have to piece up a lot of technology into one place. And rather you just subscribe, treasure data and you don't have to worry about these complexities while you can bring in the visualization and BI tool you have. That was really the pitch and product around Series A. We grew from zero to $5 million in two years and then we became $10 million in three years. So if you look at on the right side. So those are the slides from our series B deck. Around 2014 Q4 we are about to cross $2.5 million revenue quarter, pretty little professional service, predominantly the subscription. What was in reality though is we were still the founder led sales company even at the $10 million. So at this point we were around 50 employees, more geographically distributed between Mountain View California and Tokyo and 120 plus customers. But I would say for the top five customers consist of more than 95 to 97% of the revenue. So that was a huge concern from series A and series B investors of course. Right. And we got a really good gross margin, 72% at this time and little bit of recognition from the analyst. Gartner recognized Ourselves as a Gartner cool vendor in 2014 and we've been building really good IPs such as Full ND and Message Pack and we open sourced it and then we got good traction from that open source angle as well. But again, those three years were having fun. But the problem is we were feeling like we're going to the graveyard at $10 million ARR range. So why we felt that is we really, really didn't feel like we had a sense of product market fit. So after we raised some more money, we decided to hire the VP of sales. But they couldn't consistently meet the quota, right? And if you look at the bottom last 10 years we've been going through six corporate logos changes in last 10 years. And that's how we changed how many times we changed the head of design and video marketing and we decided to hire VP of marketing from a couple of data companies who were really really successful but they couldn't consistently generate the pipeline. And then if this happens, sales and marketing is not going well, right? And more importantly all of the account executives we hired, only one of them could have consistently hit the target. But of course with the help from the founders. So we decided to, okay, we were now closing a lot of deals led by founders and now we need to scale the sales team. But we really really couldn't. So that was the real struggle we had. And the biggest threat we had at that time also was Amazon, Google, Microsoft are trying to commoditize everything from the infrastructure layer. So at year 2 to 3 AWS released new product called Amazon Redshift which is the cloud data warehouse product at aws Re invent, that's their annual big cloud conferences. And Google released a new product called BigQuery which is the cloud based data processing engine. And after these product gets released we really really have to compete against Amazon and then Google. And then as a result we observed all of the important SaaS. KPIs were suddenly about to go wrong. It was not necessarily bad in the current quarter, but we clearly foresee within three to four quarters it's just going to be really really bad. So the ASP was shrinking because the conversation with the prospects and customers always thought like okay, Amazon and Google offered 00 sign up and trial. Why Treasure Data is not right and the logo churn was going up. A lot of small sized companies tend to always look for better, cheaper alternatives. And Amazon, Google, they have better brands of course than Treasure Data in terms of the cloud offering. NRR was not going up. We closed A couple large customers and they run Treasure Data for the entire business line such as game analytics, telematics analytics analytics and then a couple others. And then they were adding more computation and storage every year. But at the time they also started evaluating Amazon and Google products. Our deal velocity went down. So the deal cycle gets longer, which is really, really bad. And our close rate went down because now we have to fight against these competitive products almost every single deal. So what really really happened on the right side is our deal complexity remained really high. The reason why our deal complexity was high is we were selling the platform product. So what we need to do in the sales cycle is we need to figure out what customer actually want. So doing a little bit of what is called a solution selling. So figuring out the customer pain points, putting some of the solution into together including the PS and dashboard and of course our platform product and couple of our products bundling some implementation service and really solving the customer's problem. So that was very complex sales cycle. But now what happened was we sold like seven figure large deals but the price is getting lower and lower so that's actually a graveyard as a startup. So we clearly needed the pivot about how we sell the product and then the product itself. But we don't know how to change ourselves. So let's talk a little bit about how we pivoted. So when the company is not going well, what you need to have is a really good investor base and an angel investor base. So we talked with people like Bill Tai, Dan Shehman, Tim Andy and Jerry Yan who is the co founder of Yahoo. And as we spoke with those great advisor for us, we had a 30 minutes meeting with Jerry Yan and he really laid out two ways to go in terms of the company direction. One is build more tech stacks on top of our platform and then going into the application layer solving the end to end problem. So that's the one way to go. The second is raising a lot of money and then fighting against Amazon and Google with the performance per price. There was a similar company called Snowflake Computing. So they took a route of two by raising $1 billion plus. But treasure data founders were not hyper profile founders like them in Silicon Valley. So we have zero idea how to raise more than like a billion dollar plus from the investor base. So really the option one was the only way to go. So I was really really or we were really really fortunate enough to have a really good advisors at Treasure Data. And I was also thinking okay, now we decided to go to end to end application Route. What's the pros and cons? If you think about the platform, we were just solving the one layer of the entire stack. But you can address wide range of the problems, which investor really really likes the story. Right? Because you can address a lot of problems and go big while end to end, you solve very specific but complete business problem. Right. So that's very different. And when it comes to sales and marketing end to end application, it's narrower but you can consistently pitch the value. Right. While on the platform side you can say okay, you can address problem broader range of use cases, but it's very different each time. So when it comes to consistent sales and marketing end to end it's much easier to ramp up the sales team, while platform it's harder. Right. And what's most important for us was also the reliance on the partners. Because platform solves one layer of the entire stack, we needed to rely on adjacent tech and implementation partners and it always comes together within our sales cycle. So it was really hard to control the entire sales process. While on the application side we need to solve the complete problems by the product so that sales and marketing can control the entire sales process just by ourselves. Treasure Data engineering mindset needs to be changed end to end apps you are deploying tens of times a day while on the platform side what you really need, stability, scalability, compatibility. A lot of customers actually didn't want us to change while they just want more performance per price. On the pricing side we had a really good margin for the first years like 70% plus. But we are getting a lot of pressure from our customers, especially compared against Google and Amazon. So the infrastructure ish SaaS company, they intend to have around 20 to 30% margin. But if you go to end to end application they tend to have more than 70% which is great. Platform side, it's more difficult to be replaced because you're just becoming the foundation for the customer's business. While end to end up, you can pick up the competitor and replace quickly. Right. So those are the pros and cons between the app and platform. And while investors really like the story of the platform. But you know, as an early stage company, end to end apps is much better way to go to market. Especially because you have more consistent and predictable sales and marketing motion. And also you can control the entire sales process. That was the take by researching a lot of SaaS companies at that time. Okay, now we understood the pros and cons of apps and platform. What we're going to do, listen to the customer Based on our observation and then interview with the customer turned out on top of our data platform, 90% of our customers were analyzing customer data, including the data coming from web and mobile and social and CRM. And even though our counterpart was a technology engineering team, it turned out 90% of the requests about the data were coming from marketing team. And it's all about user acquisition, retention, upsell and cross sell, right? And after that, through a lot of interviews with the customer, we decided to build a really simple marketing application on top of our data platform which does three things. Customer identification across multiple data sources, customer segmentation and an activation. And we built this little app on top of our data platform. We didn't know how to call this app, but we decided to call this as a CDP customer data platform. Because around 2017 I found this little category called CDP in the martech landscape or chaos map. And there were 20 to 30 vendors in it. But I really liked it because there's no Amazon and Google here. After fighting against Amazon, Google, Microsoft, everything feels like it's easier to fight against. We decided to jump in into this category. Of course I saw this category will be expanding a lot. The transition was not easy of course. Around year four to five, 18 months to almost two years, it was a wartime, we didn't add a lot of employee, but we were around 100 plus people company. So first of all this pivot, only a few people were convinced. I was fortunate enough. My co founders, CEO, CFO at the time believed in me as a head of product and engineering. I decided this choice right. And there was a lot of human dramas too. One VP were not convinced the destruction, so he came to the board and trying to fire me. So in this chaotic time, those human drama can easily occur too. What I was doing was instead of trying to convince every single one of people in the company, I formed 5 people small team like myself, another co founder, front end, back end, machine learning engineer, really pizza sites team. And then finding the two to three design partners and get them subscribed with the production mrr. Once I show more customers logos with the real amount of dollars, more people became more supportive. And while I'm doing this, which took six months to nine months to even just the prototype and get the first customer. Sales team of course didn't want to wait for it. So the team wanted to try out multiple products. So as a head of product, I decided to build two products in Parliament. Why cdp? I mentioned the marketing application and also we had a really good traction on the open source side. Which is full nd. So I built the enterprise open source full ND product and we also spent a lot of sales and marketing data to this product. And this line of business went from 0 to 1 million data in ARR in two years. But we decided to shutting down after three years because CDP was growing faster. So if you look at this sometimes, if you foresee that your company is going into the graveyard, you read very lonely and tough decisions. And it turned out for us this was $100 million plus ARR decision which I will talk about later. But we had a lot of blood and struggles. This means we decided to slow down the innovations for the existing $10 million ARR customer base. So we had a lot of customer churns. In fact we have 20% of our ARR is still coming from big data product but we're not of course adding new customer for that segment. We also decided to categorize ourselves as a martech SaaS company rather than the big data SaaS company. So a lot of early employees believed in this big data story and joined us by having a lot of risk joining the small startup. But now they were leaving because we're now working on different things. We rewrote every single piece of marketing collateral. We designed our messaging and positioning, case study and everything for the tech audience. Now we decided not to use any tech jargon because we're now selling into line of business. The background of AE is very different. Now we were selling to the engineering and, and tech audience. Now we're selling to the marketing department which is more a business line. Right. So those early customers, early employees, those are the ones who believed in our early vision and we decided not to, you know, I would say make them happy or make or driving the company into the direction they believed in. So it was really, really hard. But you know, instead of slowly dying as a company, we're determined to grow. So this was a much needed change for us. So after that everything worked out really well. Again we grew from 25 to $100 million in three years. Our growth rate got accelerated from 20 to 30% to 70% plus our gross margin got hugely improved to late 70s. ASP went up, which was a little bit surprising. The platform deal we did, we had a couple of seven figure deals but now we're closing 350k to 400k a year as an average. The pipeline velocity goes up because now we can have more consistent pipeline gets generated, sales cycle gets slower because again we can control the entire sales cycle by ourselves, not relying on the partners, most importantly AE ramp up time and repeatability went up because we're now selling the repeatable use cases. After this we finally felt like, okay, we had a product market fit, now it's time to conquer the category. So now we have the best product, a lot of recognition, the best CDP product. There are now 150 plus CDP vendors in the category because the category got expanded. But again, after fighting with Amazon, Google, Microsoft, we feel like, okay, it's a little bit easier to fight against those small to mid tier vendors. And ironically the underlying platform made us easier to handle the complex enterprise demands. And you know, which became the best differentiation for us, which I will describe now. So what I found out was if you become a truly, you know, $1 billion plus platform company, you really need a killer application by yourselves. So think about iPhone. When you buy iPhone you expect you have an access to the email, web and the messaging and without them it's almost unusable. If you have, okay, you have iOS, you can program and you can develop your application by yourself, you won't buy it, right? The same with Windows. What you're really looking for is you want to buy Excel, PowerPoint and Word as a killer application on top of Windows. And obviously they have a lot of application built on top, but those are a killer apps owned by Microsoft themselves, right? Similar pattern with the hundred billion plus SaaS companies. Salesforce has Force.com or Hyperforce as a platform layer and they had the CRM or sales cloud as an initial kill application and then later on they expanded into multiple application. ServiceNow is same, they have now platform and then it. Service management was of course their killer app and they expanded into customer and also employee solution later on. So really the learning is you actually need to have both a platform and application if you want to become 100 million data plus ARR especially targeting for the enterprise. And if you have 100 million ARR, someone can build an application which generates a million dollar ARR and at that point you actually become the true plot one. So what treasure data now is landing by solving specific problems by an app by using the underlying platform to expand the use cases within our customer base. What we are doing right now is transitioning from one applications to multiple applications. And this is the diagram I'm looking at almost every single day. And thinking about it is this is how Salesforce has expanded the ecosystem. So they have what which is the product lines, where is the region they are selling to and who is the target industries and customer segments. We were selling CDP which is the one product and we were selling into America and Japan and then we specifically sold to the enterprise. And in this segment, CDP category itself is expanding, but it's expanding 30% annually only. But we as a company we want to grow more like 50 to 60% or beyond because that's how SaaS company gets the higher valuation and then growth. Right. So what we ended up doing is okay, we are selling to more region which is obvious. So we raised a lot of money and then expanding the sales and marketing team to multiple regions. But we're also now adding multiple applications on top of the platform. So we were selling to marketing. But the power of customer data is not just limited to the marketing. So I'm now working on the initiative called Beyond Marketing so that we can unlock the power of customer data into sales division and then contact center division. So by doing this you can actually expand your time like greatly. But this requires a lot of complexity. Now the sales needs to handle multiple products with the overlay, how the commission works, the region, we have global deals, how to split the commission across multiple regions. And we really need to train our sales rep with multiple product lines. So this journey continues and this is the journey we are on into the $1 billion plus ARR. So the key takeaway is although we started out as a platform company, a lot of customers or businesses always fund the project by specific problems. And you always land the customers by solving specific business problems. And you can say I can solve hundred of your problems but that doesn't resonate. And then of course that doesn't how it gets budgeted or the project gets budgeted. So you need an application to solve the complete business problems. And once you land the underlying platform allows your customer base to expand into more use cases. The investor really likes the platform story to be told. And you can keep talking about the platform for the right audience, but that's not how customers will purchase or fund the project. And once you have the platform and then forced application, you can always seek for additional applications or ecosystem to expand your total addressable market time so that you can grow more and acquire more new customer segments. And finally the big learning is, you know, we felt like around year 5 to 6 this company is going to be dead or graveyard and it was just intuition. But when things are going to wrong and you feel that wind, believe your intuition. And having the great advisors are really, you know, the key. Just talk with them, having more advice from them, getting more insights. And once you decide to pivot and change, start proving small and Turn it around. And sometimes you need to make a lonely decision to grow the company faster. So thank you very much. Now I would like to answer the questions. Okay, So let me reading through. Okay. Why did you decide to go to one build more stocks and going into application instead of battling Amazon and Google? How did you see that playing out during that period of time? Okay, let's go back to this slide. Okay, so I guess you're the SaaS founder. At certain scale, Amazon, Google, Microsoft or GAFA becomes a competitor at some point. Right? And for us it was more direct. They had a direct competitive product announced and then we just need to fight against them. Right. And the reason why I decided to go to one is for two. They have hundreds of engineers, custom hardware, software, network and infrastructure so that you can make the system more economical and performant per dollar. So again the company like Snowflake raised the money and they're doing really good. But again we were not the high profile founders in Silicon Valley and we, we're not confident we can raise like this like huge amount of money with the higher valuation and take the risk. Rather we are confident that we are really close to the customer and we learn how, what, what type of application we have to build. Right. So like this. So that's why we decided to go to 1. Hope that answer your question. So the second question, did this happen to you when you pivoted some of the original customer that helped you succeed during your 3.5 million range stage no longer had need of your product? What is the best way to handle this? Right. So again we had a lot of customer churns and we tried to make it as more like the soft landing, meaning they shrink rather than stop using. So what we had done is we built some connectors such as redshift connector or bigquery connector so that you can still keep using treasure data, but you can summarize the data and then put that into the redshift or bigquery, the Amazon and Google products so that you can use both. So we allow our customers to use both products at the same time, which slowed the customer churns in certain sense. But you know, at the end of the day a lot of customer have churned, but we tried some some of the way to do the soft landing. So I hope that answered your question. Thank you. Previously you mentioned that during the PMF phase you struggled to think sales efficiency via AES and VPs. And AES. VPs and AES being tailored for business line versus engineering audience. How did sales performance increase so dramatically after you pivoted to vertical, how much percentage would you attribute to a superior product and market versus other factors like talent? Yeah, so this is a really good question. Thank you very much. The reason why I had a great sales efficiency after the app is because we can target the same audience. I think the key to PMF is solving the specific problem for the specific Persona. Right. So in this case we're solving marketing division's problem which is customer data gets siloed and they can't have the customer 360 view for their marketing campaign execution. But if you go to on the platform side, which is big data platform, the use cases were so diverse. Okay, day one you're talking with the healthcare company, they are trying to analyze certain type of text document within the corporate. And on day two you're talking with the automotive company trying to analyze the telematics data. Three, you're talking with Internet and social gaming company trying to analyze the customer behavior. So what was really difficult is for the salespeople to each time you talk with the prospects, the industry then use cases and then pain points were so much different and you can only solve a part of the problem. So we had one sales rep who consistently hit the quota, but he was previously the engineer and Those type of AES were really, really hard instead of having AE who is selling the product to the marketing organization. Because there were a lot of reps working for large companies like Adobe, Salesforce and Oracle selling marketing tech products to the large enterprise. Right. So that was the answer. So I hope that answered your question. You said that when you change the buyer Persona from engineering to marketing, there was the difference of fit in the employee with multiple product lines. That's interesting. Do you see the same issue arising now that you have multiple Persona rather than pivoting and solving one? I don't think so. I would say the reason why we had a lot of employee churn is we are completely shifting from big data story to martech story. Right. And then it's just the truth that we didn't need a lot of talent, especially in the marketing and sales side for the, you know, the platform side. But when you are adding multiple product lines, it's a pure addition to your team, right? Yes, I'm hiring more people who have more sales and service background rather than marketing. But they are all excited to coming on board because it's just a pure addition to the market we're addressing. And then the marketing side of the talent is happy because now we're addressing more Use cases so that you can bring more products to the customer. Right. So that was the answer. Thanks for asking. So how do you sequence going directly after an app first versus building the platform initially to decrease cost of building the apps. So I think this really depends on the founders and how you start the company. For me my interest and passion was data analytics which was really, really broad. So it was really natural to start from the platform and then transitioning to the app. Right. But if you are the type of founder who deeply understand specific segments, the pain points, I think you should start from the app. But later on, if you want to become a truly enterprise billion dollar unicorn, you have to think about, okay, what's the adjacent areas you can expand, what's the common layer you can have within the product. So going from app to platform, I do think the sequence really depends on how you start the company and the initial product. So next question. How do you get the meeting with Jerry Yan? How to get meetings with great advisors and investors? I would say that was a lot of luck. So we had a really good angel investor base, especially Bill Tai who is the legendary angel investor in Silicon Valley. To be honest, I was coming from foreign country, which is Japan where I was born. I was 25. Coming here I couldn't speak any English like 10 years ago. And we had a lot of investors who didn't want to invest us. The team was, they were not convinced to the category and product and team. But once we had Bill Tai, he made a lot of introduction to a lot of people including Jerry Yan. And over the course of the time he gave us a lot of advice. So I think having the really good angel investor initially or really good VC firm in the early on is really the key because they're almost like a doctor and you're the patient and every time you have a tough situation, you can just call them and ask the help. And they were really, really happy to help us. So we were really fortunate. So, next question. Hello Kazuki, could you share the team skillset difference between software system integration company versus CDP product company, if any, the skill set difference, I would say the product side was not very different. The platform side we had a lot of engineer working on stability, scalability. We also added the application engineering team who focused more on the higher velocity application development. So that's how I organized. But on the sales and marketing side it's very, very different. So we hired the sales and marketing team who have experienced more market products and you know, also we were lucky because because we're selling to the marketing department. Most of the marketing people were using it and then experienced that problem. Right. So the marketing side they were the Persona, they're the user. So that's natural. On the sales side we are trying to hire from, you know, Salesforce, Adobe of the world who are selling the products into the marketing division. And also we had a 150 plus CDP companies and a lot of them are small so they wanted to join more mature companies. So we're sourcing the talent from them as well. So I hope that answered your question. Have you considered open sourcing big data platform during your chaotic moment to offset your small team? Why do you think it's better that it didn't? Also that is a really good interesting question. So we do actually open source a part of our stack which was full md. So fluentd is the data collection part. Let me find the additional another diagram. So if you look at on the right side, the full ND is the open source log collector. So we decided to open source everything on the client side because it sits within our customers infrastructure. But on the cloud side we made it more proprietary because we didn't want our competitors to just host our software into the cloud and then offering the same. Right. So we decided to use open source as a way to get more adoption initially because once you install this log collector using treasure data was instant. Right. So using open source as a way to generate the lead initially but we keep our differentiation as a proprietary. Maybe we had a path to open source it, but I couldn't feel any business value at that time and also now as well. So I hope that answered question why was the complete pivot needed in addition of the verticalized marketing and then sales functions? Yeah. So I would say yes, practically what we have changed is one on the product side we are adding the application side and the sales and marketing side we completely change it. But I would call it as a pivot but because we are selling to completely different audience and solving very different set of problems. Right. So on one end the reality is okay, we had 50 to 60% of employee were in the sales and marketing division. And what you're saying is that is going away, right? So that was really happened and that was really really tough. Okay, this is the last question. Did you have any real really productive partners? If so, why could you not replicate those to make the partner route to selling the platform to the success? Yeah, so again you could have a good partner but really the problem we had was with the partner you really can't control the entire sales process and that was that made it hard. The sales cycle becomes longer, the use case is very different. So that's why we decided to go from the platform to end to end. And I think as an early stage product company, you really should address the solving the problem by yourself. So that was it. Again, thanks for joining my session. I really appreciate it. I hope you got some value by sharing a lot of our treasure. Data struggle.