He raised $1.3M, hit $100K ARR—but failed. Here's the lesson every founder needs to hear. | Tolga Ermis, Founder of PromiseQ

Tolga stumbled upon a problem in the security monitoring space. Motion cameras generated way too many false alerts. So he decided to solve it using AI. He raised over a million dollars and got several customers.

But he always felt he was pushing a boulder up a hill. At one point, one of his large customers churned and went with a competitor. Tolga pivoted, but it was too late.

Now, thinking back, he realizes that one of the main issues is he may never have been solving an important-enough problem. And in my experience that's the reason that 95%+ of startups fail.

You learn a lot from the massive successes— but you can only really know what drives success if you pay attention to the failures as well. So here it is, back by popular demand.

Why you should listen:

• Why learning from failures is as important as celebrating successes.
• Why you need to deeply validate the problem with customers before building. 
• How to tell if your product isn't a top priority based on sales cycles.
• Why you probably need to pivot sooner than you think.

Keywords
startup failure, founder stories, learning from mistakes, customer validation, market dynamics, business pivot, entrepreneurship, PromiseQ, AI technology, security solutions

Timestamps:
(00:00:00) Intro
(00:02:50) What is PromiseQ
(00:11:06) Working on a Non-Priority Pain Point
(00:16:10) Realizing the Different Costumer Profiles too Late
(00:18:23) The Business Model
(00:20:52) Why you need to pivot fast
(00:28:32) If Things Went Differently

Chapters

• 2:50 What is PromiseQ
• 11:06 Working on a Non-Priority Pain Point
• 16:10 Realizing the Different Costumer Profiles too Late
• 18:23 The Business Model
• 20:52 Why you need to pivot fast
• 28:32 If Things Went Differently

Transcript

Pablo Srugo (00:00):

I love talking to some of the most successful founders in the world, but I've also realized, but a lot of the things that they associate with their success, a lot of other founders have done as well, and yet have had materially different not as good outcomes. And so what I've realized, and I think is so important going into this new year, something that we're gonna do more and more of is talking to founders who had some success to raise some money, who got some revenue, but ultimately failed. And trying to dive into what it was that they didn't get right. It is actually equally as helpful as trying to understand how somebody grew a business that got to a hundred or a billion dollars in revenue. And I think as an early stage founder, listening to stories from both sides of the spectrum, looking at the hyper successful ones and trying to think through what they got right, but also looking at the ones that had some traction and ultimately failed and trying to think about what they got wrong.

Pablo Srugo (00:47):

And between those two things, patterns are going to emerge and reasons are gonna emerge that are going to help you figure out how to handle things in your day-to-day at your startup and get to product market fit and beyond that to true success. So today we have the founder of PromiseQ. He raised over a million dollars. He got to a hundred thousand in ARR. He had customers, he had investors, he had a solid thesis, and yet it didn't pan out, and he ultimately had to wind down the business. So today we go deep into why that happened. Think back to the last few months, the last few years as you've been running the startup, how many different founders have helped you out? The reality is, founders help each other out. That's just who founders are. They pay it forward. So help a founder out, take literally five seconds, take your phone outta your pocket and hit five stars. Tolga, welcome to the show, man.

Tolga Ermis (01:34):

Thank you. Thank you for the invitation, Pablo.

Pablo Srugo (01:36):

You had a startup that, you know, ultimately failed, didn't work out just like I did. Just like many other founders that have been in the game for a while have, and but in your case, I mean, you raised like 1.3 million euros, which I guess is like, I don't know, 2 million. Let's call it one and a half to 2 million USD, but you hit a hundred K ARR. So like, you actually had, I mean, not just like some investment, but you actually had some traction and then things ultimately didn't pan out. And I think those stories are - you know, we've had just great responses to these kinds of stories. Obviously, somebody who never raises anything or never gets any traction, you know, people just assume, okay, maybe the idea just just wasn't that good, or maybe they weren't all in on it, these sort of things.

Pablo Srugo (02:16):

But like the difference between a company to get some amount of traction like yours and the ones that really hit is more nuanced than you might think from the outside. And, just doing these stories and having lived it myself, like you realize how much work goes into that and how many things actually the founders tend to get right.  And it's just a handful of things that maybe either timing or some other subtleties that didn't really pan out, and that's the difference between like, success and failure. We'll dive into all that. Maybe just like, if you could tell us what, what was PromiseQ and like, kind of how did the idea come about in the first place?

Tolga Ermis (02:50):

Of course. Sure. Yeah, you're absolutely right. We don't talk about these failures that much. And I think it's very important. So thank you for really giving the stage to founders that went through that. I think we have so many lessons to share with other founders that are up and coming that are maybe overlooked when just, you know, highlighting the ones that got super successful. May it be by luck or luck and skills and all, you know, right time, right connections, right place. And yeah. So I'm Tolga. I am, I am or was the co-founder and CEO of Promise Q. it's a Berlin based AI startup in the physical security field. So my background and the background of my co-founder Elias is actually an automotive. And before PromiseQ we were working in R and D for autonomous driving cars.

Tolga Ermis (04:06):

And we were developing and testing object detection algorithms basically that would come into a car and would basically be the perception of the cars, you know, the camera based perception. So we were doing that and by coincidence we figured out- well, it was actually the father-in-law of Elias, my co-founder who was working at the school as an administrator. And yeah, they had installed security cameras that were connected to his phone. There was no third party commercial security company in between, and he was getting so many alerts in the evenings, on weekends, et cetera. His phone was just buzzing non-stop. And then he came up to us and was like, Hey guys, you’re engineers, why don't you build a robot that would, you know, go there and check if the alarms are actual and relevant alarms.

Tolga Ermis (05:10):

And as you can imagine, most of those alarms, over 95% that we figured out later on are fault alarms, or irrelevant alarms. Because these cameras are super old that are in the field. they trigger on motion, they trigger in combination with other simple sensors when there's contact or something. And you can imagine like most of the data and most of the triggers are just waste of time and completely irrelevant. You have animals walking by, you have the cameras shaking a little bit in the wind and creating alarms basically. And yeah, then Elias and I were like, okay, we, we do have the technology, the technology is there. Why didn't anybody automate that yet? And then we set out and built PromiseQ of course we did like all the things you're supposed to do. We talked to actual commercial security companies. security operating centers to verify there is an actual problem. There is an actual need.

Pablo Srugo (06:22):

Walk me through that. 'cause I think, I think that's very important. I mean, I think at the surface level, it's an obvious thing when you describe it, it makes sense in theory. You say, okay, there's cameras and I have a home security camera. A lot of people do. any animal, anything moving around. Like that'll just set it off. So obviously a lot of false alarms. What, what motion do you go, like, when you talk about research and talking to customers, like how many people in the space, whether it's customers or potential partners or other providers, other vendors do you speak with? And what do you start hearing through, through those conversations?

Tolga Ermis (06:55):

Hmm. We started, I mean, there are templates online. You can see things like B2B, lean B2B, lean startup stuff. You can, you can find those templates online and you have a questionnaire. You have to really define your problem. I mean, we already knew the kind of technology we wanted to push, which is what you're not supposed to do. But then we tried to validate this, this problem that we yeah, coincidentally figured it out. Then we started out talking to these companies to, to leads in these security operating centers, probably. We identified them, we reached out to them through either people from our network or cold outreach on LinkedIn. And then we had our first few dozen interviews. And yeah, the problem was not only with Elias's father-in-law, it was actually a problem in the security industry and the whole security process chain, basically the cameras.

Tolga Ermis (08:08):

And then we have like a few companies in between that do the monitoring, that do a bunch of work, that send cards, et cetera. And our initial hypothesis was basically verified like that. We were still working full-time in our other jobs, in our previous job. So we did all that on weekends and in the evenings which was quite hard, but very exciting also. And then we were basically at a point where, okay, we had the problem verified. We did start building an MVP and then it was like, okay, we can half ass our day jobs, we can't also half ass the startup idea. So we had to make a decision, okay, now we are kind

Pablo Srugo (09:04):

Of going all in.

Tolga Ermis (09:06):

Yeah, we have

Pablo Srugo (09:06):

To go all in ,committed

Tolga Ermis (09:08):

We exactly,

Pablo Srugo (09:09):

I remember when- so gym track, which was my startup, was a five year journey. And, the first one was product one. And then at one point we pivoted to product two, and I remember the event, CEO and I, 'cause we'd hired a professional CEO as we started thinking about product two, we went down to this conference we went down to this conference in, I think it was in Palm Springs, and we met with like 20 different gym operators. We were selling it to gyms, and we described the new thing that we were doing, and we asked them whether it was truly a problem for them. And we had literally all 20 of them tell us that this was definitely a pain point, that it was definitely a problem. We even, at one point, it was two days on the second day, we were listening, we were hearing people tell other people about what we were doing.

Pablo Srugo (10:04):

So you, as you can imagine, we left there elated and convinced that we were actually really solving a pain point. And what happened, Tolga after. We started obviously trying to sell to the most interested ones out of those, the ones that were like, yeah, that not only do I have a pain point that pricing makes sense, definitely want to be a buyer. And out of those 20 conversations, we made exactly zero sales. We didn't convert a single one. It took me a long time to realize that we solved a problem. It just wasn't a top priority problem. It was not a number one problem or a number two problem for them. And we probably could have figured that out if we'd done the verification a totally different way. You said something that struck, that struck this chord with me because you said we kind of pushed it in our direction. So I guess I'm curious, when you reflect on what you did, do you feel that you were solving a top priority problem? Or was it more like what I ended up doing at gym trusts

Tolga Ermis (11:06):

Exactly like yours. When I look back it's two extremely good points. I think for any new founder that has, you know, this really idea for a product and then they try to push it so hard on a not top priority pain point, I think that's really the issue. And you have to really be fast in figuring out that, okay, that's not it. And for us, it took way too long until we re realized, okay, it's a problem, but it's not their top priority. That's why they were interested, but they were not buying that quickly. The sales cycles were super long. Then we also had the issue, so what we built was this cloud-based false alarm filtering with the computer vision module that would automatically analyze the incoming alarms from the cameras and then get rid of the irrelevant alarms and only forward the real alarms to the security operations center.

Tolga Ermis (12:19):

So the operators only focus on real alarms. They save time and money, they can really focus on the actual alarm. They don't miss real alarms anymore. Sounds everything super great, but because it was cloud-based, we had two integrations to do. So we were a module in security in a tool chain, basically. So we had to get the data in from the cameras. And the data was all different formats. So we had to build many, many integrations and all integration. Every customer was basically a custom integration project, which made the sales cycle super long. And then at the end, the outbound integration into whatever system they're using, they would use alarm management systems or video management system. So you have to bring the verified data into their system again. And so you had like two integrations and not a top priority thing because we always thought security operating centers, it's their main business to verify alarms, but they, they do so much more. You have to look at their total revenue and alarm verification is only a fraction of it.

Pablo Srugo (13:47):

What was- so yeah, maybe walk me through, again, back to my example, like what I realized is these gym operators, they care about two things. They care about opening up new locations, they care about getting new members to a smaller extent, maybe they care about retaining existing members. Everything else is just not, is just not gonna get you know, all that prioritized. And, you know, I'll give you actually a different example. I was speaking with this founder. Founder of a company called FlashFood. Really interesting product. What it does is it's this app that lets big like that, let's say grocery stores - you know, how grocery stores, when their food is about to go bad, they tend to discount it, but it's not as simple as it might seem. So this app is this kind of mechanism by which they can put out these grocery stores can, can actually surface up anything that's about to go bad.

Pablo Srugo (14:39):

And then people that are on the app go and they actually drive to the store to buy that stuff. That's like 50% off. And they started off by obviously selling into smaller grocery chains, like four locations, 10 locations. And it was pretty crazy because he told me, obviously he fought tooth and nail as we all do to get those first few pilots. He finally got two different pilots, two different sets of grocery chains, and both of them turned, both of them were like, you know what? Like, this is just not important enough. And then he got another one, which was like a hundred times the size. Like instead of having four or five, they had like 400 or a thousand, like 500 to a thousand locations. And that one not only worked, they rolled it out across the entire division. Like it turned all the stores.

Pablo Srugo (15:24):

And so I asked the founder, I said like, Josh, why would you be a fit for the massive enterprise and not for the little ones? And he said, I think a really great point. He's like, the little ones could grow by expansion. So the focus was how do I go from four locations to eight locations, eight locations to 16? And in that context, what we're doing and what they were doing, which was like effectively on the margin, kind of adding a new revenue stream was never going to be top of mind for the executives. Now on the bigger one, 500, a thousand locations in their geographies, they really had nowhere else to go. So everything was about optimization. And so play like this could be top of mind, could solve a top priority. I guess I'm curious, like in your case, as you think about that, what do you, what do you think was top of mind really  for your buyers?

Tolga Ermis (16:11):

There were different customer profiles, which we realized also when it's too late, there are for example, operating centers that have a different business model than other operating centers. They would, for example, make money with false alarms as well. So if the end customer, you know, you have a business car dealership, or even if you, if you're a, a private customer, you have cameras installed. And if their business model model is that they also make you pay for false alarms, then they don't care to optimize that. That's that you know, how they make their money basically. Right?

Pablo Srugo (16:51):

So they were charging per alarm. Yeah,

Tolga Ermis (16:54):

Exactly.

Pablo Srugo (16:54):

Like per alarm resolved sort of thing. Right.

Tolga Ermis (16:57):

Exactly Or they, they would have like a flat rate with that, that many is included in, in the monthly pay and it exceeds then they make them pay for anything, even if it's a false alarm.

Pablo Srugo (17:08):

Well, it's kind of like, you know, a lot of legal tech, like lawyer, lawyer technology is like about, you know, efficiency but it's billable hours also. So the more hours they can bill, the more they can charge. And it's not, so it seems like the incentives are not fully aligned. 

Tolga Ermis (17:21):

Yeah, exactly. And then you really, but there are other types of security companies that are bigger and that install those cameras. For example, those ones, their interest was actually to optimize and to sell a holistic solution to their customers that is really streamlined. So that was a good customer profile. But again, still looking, looking back we focused on, on small and middle sized companies. So we were putting a lot of effort and putting in a lot of time for not a lot of customer lifetime value or average-,

Pablo Srugo (18:13):

so, you know, 'cause you're talking about integrations, custom software, all this big stuff. Like what was a CV for you? Like what were these customers paying? What was the model?

Tolga Ermis (18:23):

Really different. We had one that was really big and with them we made like 80 or 90% of our revenue which in hindsight was bad because when they turned it was basically the beginning of our end. And then all the other ones, we had a bunch of small ones. They are, contract value was just between a few hundred euros and a thousand euro,

Pablo Srugo (18:53):

Like a year or a month?

Tolga Ermis (18:55) 

A month.

Pablo Srugo (18:55)

And then, so you mentioned the big one, churned, which I guess they were paying probably like 50,000, 80,000 Or so a year?

Tolga Ermis (19:00):

they were like 80 around about

Pablo Srugo (19:03):

 Why did they churn?

Tolga Ermis (19:04):

They churned because we did several things wrong there. First of all, we didn't know our product costs that well. So our cloud-based false alarm filtering system, we didn't know exactly how much we're paying for one alarm. And they, they bought like per month I think half a half a million of alarms that were processed. And we had a deal with them from the beginning. They were one of our first customers. And I knew back then already, okay, this is, we're losing money with them, but it's a good pilot customer. It's a good reference customer. They have a few thousand cameras out there all on our platform. But then we wanted to adjust, right? We didn't want to lose money. We wanted to either ring the cost down or increase the price. So we went there, tried to negotiate a higher price, and that made them look at the competition and then eventually change. And I was really convinced that they were not gonna go to the competition because we had put, again, a lot of time for customization for them. And I thought, you know, the competition is big and they're not gonna invest that much time to do the same customization. But in the end, they used the APIs and they were able enough to build their own workaround. And yeah, unfortunately. 

Pablo Srugo (20:45):

And what, like what was the competition like? Did you enter a market where there were other people solving the same problem, or were there any, like, big differences between you and the competition?

Tolga Ermis (20:52):

It was in the very beginning, three years ago, 2021, when we started, it was only this one competitor. They had done a pivot into this space. They were already big. They did an exit two years ago as well. There were not that many. And there's still not that many. I think it's getting more and more crowded, but this cloud-based false alarm filtering, yeah, there is competition and the price sensitivity of the market is very, very high. That's why they changed, that's why we lost that client, which yeah, was very unfortunate. Yeah. And then we did realize all of that, beginning of that beginning of this year that's why we did a pivot as well. So moving from cloud-based video analytics solutions to an edge based solution. Because we thought, you know, to work around all of these integration issues and shorten the sales cycle we need to be on the edge, we need to be right at the source of the video camera, video source, basically.

Tolga Ermis (22:09):

And we built, I mean, Nvidia had just announced their new Jetson series and we are super big fans of that real time AI. And yeah, we said we will use a Jetson board, bring it to the cameras, automatically have a plug and play solution that connects to the cameras and has sophisticated AI functions running and to where we create the alarms ourself. So we are not dependent on whatever the camera system does. And we don't have to standardize, we don't have to do the integration. We just create the alarms ourselves on the Nvidia Jetson board and then connect it to our cloud other system. And yeah, that was the idea. And vision was amazing with this Nvidia Jetson, you can do really crazy things.

Tolga Ermis (23:08):

You can really combine different AI models, not only computer vision, but you can have an input, output control where you control an actor. You can use speech models, LLMs, and really have a virtual security officer basically that can automate a bunch of processes that is currently done by either guards that are on the field or the operating center. And that was our goal or product vision, new product vision. And yeah, I was out on the field pitching that. I had a really great team. We did a funding round at the end of last year. So I had money, I had a really great team, and we had this new product idea or this pivoting strategy that was extremely promising. And then we went out there, we had a bunch of new big pilot clients that started actually testing. We sent them the hardware and yeah, ultimately what happened, they tested for a few months and what I, and my team, basically what we promised them. We just did not deliver it in time. So they just-

Pablo Srugo (24:31):

It’s just harder? it was just harder to do it on the edge than you’d foreseen?

Tolga Ermis (24:34):

Exactly. And it just took way longer than I had hoped. So traction and the revenue did not come, did not come as expected. Then that other client from the other previous business model dropped out. And that was basically our downfall. And then we had taken a public grant from the government here, here in Berlin, in Germany which is basically for a research project, you have to hire 12 people. They want to create jobs here. You get a million euros, 600 in loan and 400 in equity free grant. Sounds amazing. That, you know, was a super good addition to the VC money that we got. But yeah. But then when the problem started I did have an investor at the end that said, I still believe in you guys. I want to finance it until the end of 2025. This is gonna be a good thing, but you need to -

Pablo Srugo (25:45):

How much money would that have been? Another, like million, half a million?

Tolga Ermis (25:48):

I think another half a million.

Pablo Srugo (25:51):

Hmm. So material, good amount of money.

Tolga Ermis (25:53):

Yeah. And with a minimal team, super lean, you know, downsized really efficiently, which did make sense. But the people, the bank from the government grant they were like, no, either you continue with the original project with 12 people with this super high burn rate, or you discontinue the project and you have to pay back what you got. And then we were like, okay, we're screwed. Yeah. And then that made it also impossible to sell the company. I did also have a few potential buyers lined up, but when they heard that there’s three hundred thousand euros of debt they were like, no, we're not gonna, we're not gonna go in there. 

Pablo Srugo (26:45):

So maybe, maybe just now, like reflecting, looking back, I mean, I'm, I'm curious obviously, like you can look at it as, you know, you ran out of money and, and technically that's true. There's the issue with the grant, and that's not the first time that I hear that those things can come back and, and bite you, though you often do them 'cause you have to. So it's not like there's that much choice into it. But what do you think were either some of the big mistakes or reasons why the startup didn't work? You know, fundamental reasons. Mm-Hmm

Tolga Ermis (27:17):

<Affirmative>. I think there are several things we had to be. I think this customer profile, we did change it in the end. But the ideal customer profile you have to have, it has to be scalable. And if as a founder you don't want to see it or you want to really push hard for that specific customer group then it's not gonna work. And then this is, I think what, what we did for too long until we realized, okay, it's not gonna scale with this with that customer group. We have to rethink the thing. We did that in the end, but, but we did it too late. And this whole integration stuff, I think we should have done the pivot way earlier than we would've had more time to to actually deliver.

Pablo Srugo (28:11):

Let me ask this, let's say this edge, in your best guess, if you had started off with AI on the edge and let's just say that the tech was there from day one do you think, do you think things would be different? Like, do you think there was enough market pull and demand for that value prop or that idea? 

Tolga Ermis (28:32):

Mm-Hmm.  It's very hard to say. I would say yes. But it is a very new field. It's a new market that you're kind of creating. So the security, physical security industry is very conservative and very fragmented in Germany, at least in the EU as well. So you need to-, the timing needs to be right, and you might need to educate the market and the customers a little bit. But as I said, once we had that new pivot, the new idea and the new product vision they were extremely keen to test it. And we had, I think 5, or 6 really big ones that were testing it. But where we didn't deliver the new promised features in time, and then they, obviously it didn't convert.

Pablo Srugo (29:30):

Makes sense. Well, Tolga, we'll stop it there, but thanks so much for sharing your story. I think like, there's so much to be learned in stories like, like this one where everything didn't, didn't pan out, everything didn't go according to plan. And I mean, that's most startup stories, that's the, that's the 99%. So thank you for, for openly sharing yours with us.

Tolga Ermis (28:52)

Thank you, Pablo. I hope it helps.

Pablo Srugo (28:54)

you remember the first person who told you about Bitcoin, the first person who told you about Uber. You want to be that person because being first is cool. So be a cool person and tell your founder friends, set it to them on WhatsApp. Put it in a WhatsApp group, put it on a Slack channel. Let people know about the show. Let people know about this episode. Don't let somebody else beat you to the punch and share it with your founder friends. First, remember what Ricky Bobby said. If you ain't first, you're last.