Let's get real. We all screw up. What's the dumbest thing you did in your first startup? Share your epic fails so others can learn from your mistakes. Let's help each other avoid disaster!
Because they repeatedly demonstrated a very poor grip on reality: they seemed to think that inventing a bland two-word phrase (a) gave them all rights to all interpretations of that phrase and (b) made my part as prospective tech/CTO easy to conjure that up as spec and reality.
I have a similar past failing. Someone that was a super friendly people person that never met a stranger. Would talk a good talk about plans, but would just not execute. Staying focused on one thing was impossible.
We tried to compete on price in a market that really wasn't as price sensitive as it claimed to be. Yes some people wanted to "pay less" but in reality the big players were happy to pay the "industry standard" rate. Many of those big players we would later find were happy to pay far more even. For every player who wanted to "pay less" there were easily 10 that would pay more to get the opportunities their competition wasn't willing to pay for.
We did pivot to add a more premium, high touch offering and it sold very well. Thats how we came to the second unhappy realization: Despite having what we considered a very nice TAM the industry didn't really want to be scaled. This meant that our clients who paid for our premium offering were very difficult to service and create value for.
Our business was in a sense connecting buyers with sellers for a fee, and there just weren't enough sellers to satisfy the industry of willing buyers eager to pay for value. A fine boutique business and a solid way to have a $12M/year company focusing on BIG DEALS. Not much of a play to scale with tech and profit by reducing costs and friction. Nobody cared about costs and friction was oddly desirable for our customers as it meant "we're in the game - things are happening!".
Had we not "raised money" and encumbered ourselves with investors we'd have continued operating. As it turned out, we more or less took on a bunch of debt to operate the same business we could have made with no tech at all.
That's a tough pivot. So, you went from low-cost to high-touch, only to find a market that didn't scale? That's gotta sting. Did you consider a hybrid model or focusing on only one end???
We had a team of data scientists who were developing models (in the abstract sense) that we would then train on specific and then deploy as concrete models.
This was just before the BERT paper came out, we were working with a few different models, sometimes the classical models from scikit-learn, sometimes these were
for text trained with a GPU. Ideally I wanted to make it easy for the data sci's to put their models in front of customers so it was not about getting one model up and running but developing a process to grease the skids. Back then, for instance, a particular version of Tensorflow required that you had a particular version of the CUDA libraries installed. We'd start with models we downloaded off the net so it was important that we could support whatever version of Tensorflow the model needed and have to click-click-click on various permission forms the way NVIDIA insisted on. (Turned out the CUDA libs are just userspace libraries and you can pack them up in a wheel)
There was that and the fact that pip's resolving algorithm was and is incorrect (e.g. works OK all the time for very simple project, screws up occasionally for medium sized projects but it is easy to work around and maker excuses for, a big project might fail to build with complete reliability) that you can sabotage all your virtualenvs with this one weird trick
pip install --user <some_package>
that data sci's often think they are too smart to have to have any discipline, that Docker is part of the problem and not part of the solution when it means the data sci's now can easily find a Python where the default charset is Hungarian, etc.
After seven months or so I'd figured out close to 100% of the reason why our Python builds (really anybody's Python builds) were not reliable but the wetware problems in our organization and outside of it were severe. (If there was one root cause for "python builds being unreliable" it is that the Python community was willing to live with unreliable pip for so long)
Management though was sick and tired of Python and we were also focused on other aspects of our value prop so we quit working on the Python trainer and they had me working on another part of the system in Scala + Typescript. (There it was frustrating that we had a data analysis pipeline that didn't give the right answers consistently because even though we'd thought a lot about how to initialize it nobody had thought about the problem of how to tear it down)
I thought it was quite "incredible" (a favorite word of the CEO) when it was announced they'd been bought by one of the world's biggest footwear brands but it was true -- I really did believe in our vision and told everyone that I thought our product could make so much value for one of our customers that they'd buy us and that happened.
Forming a partnership with a salesman who couldn't sell and who worse kept me from selling work I could have sold myself. Boy did we get a lot of calls and we learned a lot about the industry we selling into, competitors, etc.
(1) He was more ambitious in terms of the scale of the work. I could (and did) sell and execute small consulting jobs that would have never made me rich but could have kept me cashflow positive. For him anything had to be bigger so he could get a commission. I did sell a few projects after he left and also had another collaborator for a consumer product startup who I didn't feel was listening to our subject matter experts so I gave up on him and got a job for somebody else's startup.
(2) As a team we were highly effective at getting calls and getting people to talk, often about things they shouldn't have told us. We found out a lot about the internal tools at startups, failing projects at three-letter agencies, cloud migration plans at the world's most secretive hedge fund, etc.
(3) The worst problem with this guy was that he was (at least sometimes) dishonest. One person who we were talking to warned me that he'd lied to him and I didn't take it too seriously. Then he lied to me in the heat of the moment and I accepted his apology. There was a third time that I went down to NYC to meet somebody I had met through all the marketing activity we had done and he told me how my partner had told him totally different things with him individually as opposed to the group and that was the initiator of the breakup.
Even if you're willing to put up with small dishonesty from someone or you don't take it very seriously you can be sure that it can have a deadly impact on your business.
> (1) He was more ambitious in terms of the scale of the work. <snip> bigger so he could get a commission.
Finding a sales person is definitely a two-edged sword. Have them eat what they kill means they have to constantly sell small jobs, or wait for that perfect big kill. Pay them a base rate they are comfortable with, and they have no motivation.
Other sales related issues we've all probably seen where sales make promises that are damn near impossible to deliver or fail to recognize an valid sales because the truly do not understand the product. I've been hit by both of these from the same person.
I do the same thing! How do you find people who want your product before you make it? It's hard to get people to know about it without showing them something first.
What if I don't have any distribution or social followings?
If you can't find people before you make it, then you won't find them after you make it either.
Which makes this an excellent question. Indeed perhaps the most important question to ask before you build anything.
Too many programmers subscribe to the school of "if you build it through will come'. (Hint: that's not how it works.)
Firstly you don't have to show anything to potential customers, (and even when you have it, it's better not to show it.) You are selling benefits, not features.
Secondly, how to find them is exactly the hard problem you need to figure out (it'll be different for different products.) For me personally, I the late 90s / early aughts, I traveled to user groups (all over the world.) Turns out that "showing up" spread the word very quickly.
(That worked for me because it's a niche area and nobody else bothered to travel, so I stood out.)
The key lesson here- if you can't define your market, and if you can't reach them, you can't sell to them.
Every business has hard problems. This may be yours. Solve this problem first, not last.
That’s a mistake I see too many people make. They try to solve a problem and build a product from afar, using their imagination or trying to copy an existing product without having personal relationships with people having that problem (or without having it themselves). I guess it’s possible to create a product first and then find people who need it and iterate from there, but I’ve found that having personal relationships and then making something in response to a common problem they have is the only way to start and grow. That’s what I’ve found “talk to users” actually means.
I made this mistake as a solo founder. It still ended up opening new doors for me, but still could've done a lot better if I had internalized this from the get go.
I can't give details over why I learned these lessons without dox'ing myself, even speaking in broadstrokes is hard and the actors are litigious.
- Don't make your product hard to buy.
- Don't take investments from a potential customer.
Long story short, the key innovation was fiscal (looking to draw revenue from the real meat of the industry and not where it was traditionally placed in the supply chain), the tech was compelling enough to get investment from a large player, we built the product, then tried to sell the product to the investor, they dragged their feet, and eventually offered to buy the company instead of license the technology. But only after we ran out of runway and everyone lost their jobs.
The founder said "no" and the company is gone.
---
When I say "don't make it hard to buy," in B2B sales think about how much money you want to make off a single customer, given that number, what level of the organization is there someone that has that purchasing power, how hard is it to get them in a room, and can you walk out with a sales contract finalized or do they need to kick it back to their team for a final approval. And if that timeline exceeds your runway and you need the sale to close you're fucked.
In less crass terms, as an early stage startup you want your enterprise sales to scale horizontally through an organization. If your ideal user is at that company, you can only make so much money by selling it to them, because they have a limited budget. So you can instead sell to their manager, by getting the user to convince them to get more money to buy your product. But if you get greedy and then try to sell to that manager's manager, all of the sudden your advocate is two levels removed from the person making the decision and that's much more difficult to close. What you want is lots of deals closed fast, and that manager to tell another manager at the same level their team is using your product, or people to talk about you at the water cooler and get their managers to buy.
And once you scale up, then you can offer discounts for expanding to all teams under an org instead of each team buying individually, and now you have an enterprise contract that's a signal to get another enterprise contract with a competitor.
I worked at a startup where the founder learned this the hard way. We were trying to sell to government agencies. He insisted on a pricing model that meant we had to go through complex and expensive RFP processes.
We had potential customers come to us and tell us directly that they wanted to buy our product. They explained that with some adjustments to our pricing model, we could avoid the need for an RFP. The founder would not budge and we really struggled.
This is from work over 30 years ago on our garden simulator by my wife and me. Being too ambitious (and perfectionist) at the start with product plans for software. Building a Minimum Viable Product (MVP) instead which prioritized essential needs would have engaged customers years sooner in our case and provided valuable feedback for deciding on future directions for new features or changes. A couple of people early on warned us to simplify and prioritize, but we ignored that. Learned lots of other lessons from that project and others, but that was the biggest and most costly lesson overall.
Sorry, off topic but a garden simulator like animal crossing or garden simulator like planning a garden?
Because if it's the second one, please try again. There are ZERO user friendly garden planning software options that account for climate, soil type, partner plants, and rotation schedule.
Thanks for the reply. It is not quite either, but between the two it is more of the second (planning). It actually focused more on learning the science of gardening than planning an actual garden in a way you might use for your own backyard. https://www.gardenwithinsight.com/ "The Garden with Insight garden simulator is an educational simulation that uses weather, soil, and plant growth models to simulate a simple garden in an open-ended microworld setting. You can plant vegetables and grow them to learn more about plants, the soil, the weather, gardening, and science."
It was in Delphi Pascal for Windows, and I would still like to port it to the Web with JavaScript/TypeScript at some point.
it was supposed to be a folksy, communal, whole-earth inspired bookstore that would interact synergistically with actual book stores and the rest of the world.
instead, we built a fucking monster that is awesome for consumers and a nightmare for almost everybody else.
> it was supposed to be a folksy, communal, whole-earth inspired bookstore that would interact synergistically with actual book stores and the rest of the world.
A few things have changed since you left, it has to be said! There is no need to apologize - you weren't involved in the later stuff.
Amazon has been wildly successful in most financial measures, so surely you can claim some bragging rights for that.
As a monster ... ... yes it is. Awesome for consumers? Perhaps. Nightmare for almost everyone else? Again - perhaps.
I'll drop this now, its a glass half full/empty type discussion and way too involved for now. I will suggest you might take pride in being in there at the start and having a hand in creating a unicorn that went full on pegasus.
I have shared my startup story here, and it was a full-on cargo cult. I am now convinced that Cargo Cult thinking is a far bigger killer of startups than we presume.
Hiring from the network sounds like a fast way to get qualified employees, but it backfired when we ended up with groups of employees who were too tight (some even shared apartments)
The work place was transformed into a reality show where you had groups of people trying to maneuver "up" in the system, by convincing the management that their way was better. No matter the discussion, they would back each other and support their own guys. As a group they would always win a discussion.
The company stagnated, but they were able to get new jobs and left before the long term impact was (finally) realized by the management
It is applying random chance to hindsight, absolutely. They were bigger than us and growing fast, and we’d have had no impact likely, but it was not the wrong decision when we made it I think. It just worked out poorly to the tune of millions so I count it as a fail. Sometimes you make the right call with the information you have and it turns out badly in hindsight.
But, we still did alright. Worse things have happened to me. After a certain personal tragedy happened to me a few years back, I’m now in a place where I am only minimally bothered by anything that doesn’t involve someone dying.
Worked with a bunch of ex Google employees who expected to write code and money pour in. This did not happen. A sales guy we hired warned us that at big companies you usually start developing sales materials and approaches years before launch.
I used a nosql database (Cosmosdb) when I should have used a relational database. The Azure cost plus the lost time in transitioning to SQL plus the cost involved in the implementation ran into embarrasingly debilitating six figures.
1. Shortly after I started working on Tarsnap, I was introduced to a "serial entrepreneur" who expressed an interest in getting patents on Tarsnap's technology and licensing it out to other companies. I wasn't interested in doing this myself -- I wanted to provide the world's best secure backups, not sell patent licenses -- but I agreed that if he covered all the costs and paid me an up front fee plus a share of royalties then he could resell the "spinoff" IP.
Distracted me for about a year before he backed out. And then he asked me to pay for the money he had spent on lawyers (I laughed).
2. About five years after Tarsnap launched, I got a phone call from a YC company saying they were interested in "working together with me". The conversation went around in circles for half an hour before I figured out that was a euphemism for them being interested in acquiring Tarsnap.
I was in 2nd Year of my engineering. as many engineers before going to the college i used to make computer games on C++. Games caught my interest really early, because thats what is really fun and challenging in programming. One day while making a ping pong game, i thought what would happen if both the sides computers will play? who will win? from that i developed a programming game, in which you need to write algorithm for your battleships to destroy other battleships. You could try really good algorithms to test them for eg. min-max, monte-carlo tree search, RL, Deep learning etc.
Since i was in India, people here do programming for the sake of coursework or to get jobs. Really passionate people are difficult to find.
I tried to launch the product in my college, but sadly no one would want to play it. The game was really challenging to grasp at the beginning, I also pitched it to my professors to include it in the course curicullum of AI, they liked the idea, but refused it by saying it will be an overhead for the students to learn first about the algorithm and then about the Game API.
for an year i dejectedly saw that not everyone is as passionate as you are. I found no market for my programming game. If it would have launched somewhere in US, it could have been better, since MIT has such a kind of competition in which students needs to make the bots. it is not that programming games have no market, there are games like : Battlensakes, coderOne etc. but their market share is very less.
I learnt the lesson the easy way i guess, because i had a safety net since i was still in college, and had a job from the following year.
But then i really understood about product market fit, which i used to ignore while they taught in entrepreneurship classes. If anyone wants to see how the game looked : https:\\aiplaygrounds.in . I have revamped by business idea and working on something else.
59 comments
[ 4.4 ms ] story [ 134 ms ] threadMe still bitter and twisted? Never!
~The co-founder who in addition to delivering ZERO value of any kind in addition to demanding a large salary, CEO title and Chairman status.
We all run into this guy, I think everyone has a story like this.
Looking back I feel totally stupid about it.
Our business was in a sense connecting buyers with sellers for a fee, and there just weren't enough sellers to satisfy the industry of willing buyers eager to pay for value. A fine boutique business and a solid way to have a $12M/year company focusing on BIG DEALS. Not much of a play to scale with tech and profit by reducing costs and friction. Nobody cared about costs and friction was oddly desirable for our customers as it meant "we're in the game - things are happening!".
Had we not "raised money" and encumbered ourselves with investors we'd have continued operating. As it turned out, we more or less took on a bunch of debt to operate the same business we could have made with no tech at all.
This was just before the BERT paper came out, we were working with a few different models, sometimes the classical models from scikit-learn, sometimes these were
https://en.wikipedia.org/wiki/Convolutional_neural_network
for text trained with a GPU. Ideally I wanted to make it easy for the data sci's to put their models in front of customers so it was not about getting one model up and running but developing a process to grease the skids. Back then, for instance, a particular version of Tensorflow required that you had a particular version of the CUDA libraries installed. We'd start with models we downloaded off the net so it was important that we could support whatever version of Tensorflow the model needed and have to click-click-click on various permission forms the way NVIDIA insisted on. (Turned out the CUDA libs are just userspace libraries and you can pack them up in a wheel)
There was that and the fact that pip's resolving algorithm was and is incorrect (e.g. works OK all the time for very simple project, screws up occasionally for medium sized projects but it is easy to work around and maker excuses for, a big project might fail to build with complete reliability) that you can sabotage all your virtualenvs with this one weird trick
that data sci's often think they are too smart to have to have any discipline, that Docker is part of the problem and not part of the solution when it means the data sci's now can easily find a Python where the default charset is Hungarian, etc.After seven months or so I'd figured out close to 100% of the reason why our Python builds (really anybody's Python builds) were not reliable but the wetware problems in our organization and outside of it were severe. (If there was one root cause for "python builds being unreliable" it is that the Python community was willing to live with unreliable pip for so long)
Management though was sick and tired of Python and we were also focused on other aspects of our value prop so we quit working on the Python trainer and they had me working on another part of the system in Scala + Typescript. (There it was frustrating that we had a data analysis pipeline that didn't give the right answers consistently because even though we'd thought a lot about how to initialize it nobody had thought about the problem of how to tear it down)
I thought it was quite "incredible" (a favorite word of the CEO) when it was announced they'd been bought by one of the world's biggest footwear brands but it was true -- I really did believe in our vision and told everyone that I thought our product could make so much value for one of our customers that they'd buy us and that happened.
What did you do different when you took over? (Just curious)
(2) As a team we were highly effective at getting calls and getting people to talk, often about things they shouldn't have told us. We found out a lot about the internal tools at startups, failing projects at three-letter agencies, cloud migration plans at the world's most secretive hedge fund, etc.
(3) The worst problem with this guy was that he was (at least sometimes) dishonest. One person who we were talking to warned me that he'd lied to him and I didn't take it too seriously. Then he lied to me in the heat of the moment and I accepted his apology. There was a third time that I went down to NYC to meet somebody I had met through all the marketing activity we had done and he told me how my partner had told him totally different things with him individually as opposed to the group and that was the initiator of the breakup.
Even if you're willing to put up with small dishonesty from someone or you don't take it very seriously you can be sure that it can have a deadly impact on your business.
Finding a sales person is definitely a two-edged sword. Have them eat what they kill means they have to constantly sell small jobs, or wait for that perfect big kill. Pay them a base rate they are comfortable with, and they have no motivation.
Other sales related issues we've all probably seen where sales make promises that are damn near impossible to deliver or fail to recognize an valid sales because the truly do not understand the product. I've been hit by both of these from the same person.
rookie mistake! lesson learnt:
don't make before you sell
What if I don't have any distribution or social followings?
P.S. I'm in the same boat.
Which makes this an excellent question. Indeed perhaps the most important question to ask before you build anything.
Too many programmers subscribe to the school of "if you build it through will come'. (Hint: that's not how it works.)
Firstly you don't have to show anything to potential customers, (and even when you have it, it's better not to show it.) You are selling benefits, not features.
Secondly, how to find them is exactly the hard problem you need to figure out (it'll be different for different products.) For me personally, I the late 90s / early aughts, I traveled to user groups (all over the world.) Turns out that "showing up" spread the word very quickly.
(That worked for me because it's a niche area and nobody else bothered to travel, so I stood out.)
The key lesson here- if you can't define your market, and if you can't reach them, you can't sell to them.
Every business has hard problems. This may be yours. Solve this problem first, not last.
This quote really stod out tho: If you can't find people before you make it, Then you won't find them after you make it either.
literly every time i just wasted money in business was because i didn't work by this quote!
2. Recognize a $space shortcoming that users would pay to resolve
3. Start building
Bonus points if it's an area you're passionate about! But don't expect everything you're passionate about to solve a marketable problem.
If you represent that you have a product, sell it, collect payment, and actually don't have a product at all, that could be fraud.
- Don't make your product hard to buy.
- Don't take investments from a potential customer.
Long story short, the key innovation was fiscal (looking to draw revenue from the real meat of the industry and not where it was traditionally placed in the supply chain), the tech was compelling enough to get investment from a large player, we built the product, then tried to sell the product to the investor, they dragged their feet, and eventually offered to buy the company instead of license the technology. But only after we ran out of runway and everyone lost their jobs.
The founder said "no" and the company is gone.
---
When I say "don't make it hard to buy," in B2B sales think about how much money you want to make off a single customer, given that number, what level of the organization is there someone that has that purchasing power, how hard is it to get them in a room, and can you walk out with a sales contract finalized or do they need to kick it back to their team for a final approval. And if that timeline exceeds your runway and you need the sale to close you're fucked.
In less crass terms, as an early stage startup you want your enterprise sales to scale horizontally through an organization. If your ideal user is at that company, you can only make so much money by selling it to them, because they have a limited budget. So you can instead sell to their manager, by getting the user to convince them to get more money to buy your product. But if you get greedy and then try to sell to that manager's manager, all of the sudden your advocate is two levels removed from the person making the decision and that's much more difficult to close. What you want is lots of deals closed fast, and that manager to tell another manager at the same level their team is using your product, or people to talk about you at the water cooler and get their managers to buy.
And once you scale up, then you can offer discounts for expanding to all teams under an org instead of each team buying individually, and now you have an enterprise contract that's a signal to get another enterprise contract with a competitor.
I worked at a startup where the founder learned this the hard way. We were trying to sell to government agencies. He insisted on a pricing model that meant we had to go through complex and expensive RFP processes.
We had potential customers come to us and tell us directly that they wanted to buy our product. They explained that with some adjustments to our pricing model, we could avoid the need for an RFP. The founder would not budge and we really struggled.
Because if it's the second one, please try again. There are ZERO user friendly garden planning software options that account for climate, soil type, partner plants, and rotation schedule.
It was in Delphi Pascal for Windows, and I would still like to port it to the Web with JavaScript/TypeScript at some point.
it was supposed to be a folksy, communal, whole-earth inspired bookstore that would interact synergistically with actual book stores and the rest of the world.
instead, we built a fucking monster that is awesome for consumers and a nightmare for almost everybody else.
sorry.
Like Discogs, but for books?
EDIT: looks like Amazon bought that too: https://news.ycombinator.com/item?id=33933791
Amazon has been wildly successful in most financial measures, so surely you can claim some bragging rights for that.
As a monster ... ... yes it is. Awesome for consumers? Perhaps. Nightmare for almost everyone else? Again - perhaps.
I'll drop this now, its a glass half full/empty type discussion and way too involved for now. I will suggest you might take pride in being in there at the start and having a hand in creating a unicorn that went full on pegasus.
https://healthio.notion.site/How-Cargo-Cult-Thinking-Nearly-...
The work place was transformed into a reality show where you had groups of people trying to maneuver "up" in the system, by convincing the management that their way was better. No matter the discussion, they would back each other and support their own guys. As a group they would always win a discussion.
The company stagnated, but they were able to get new jobs and left before the long term impact was (finally) realized by the management
But, we still did alright. Worse things have happened to me. After a certain personal tragedy happened to me a few years back, I’m now in a place where I am only minimally bothered by anything that doesn’t involve someone dying.
1. Shortly after I started working on Tarsnap, I was introduced to a "serial entrepreneur" who expressed an interest in getting patents on Tarsnap's technology and licensing it out to other companies. I wasn't interested in doing this myself -- I wanted to provide the world's best secure backups, not sell patent licenses -- but I agreed that if he covered all the costs and paid me an up front fee plus a share of royalties then he could resell the "spinoff" IP. Distracted me for about a year before he backed out. And then he asked me to pay for the money he had spent on lawyers (I laughed).
2. About five years after Tarsnap launched, I got a phone call from a YC company saying they were interested in "working together with me". The conversation went around in circles for half an hour before I figured out that was a euphemism for them being interested in acquiring Tarsnap.
Built for two years, failed to go to market, failed to raise funding, and ran out of time. What followed was even worse.
And after 3-4 years, we paused development to explore the world of blockchain.
We are all in and back fully on Automatio. Gonna catch up with lost time.
Since i was in India, people here do programming for the sake of coursework or to get jobs. Really passionate people are difficult to find.
I tried to launch the product in my college, but sadly no one would want to play it. The game was really challenging to grasp at the beginning, I also pitched it to my professors to include it in the course curicullum of AI, they liked the idea, but refused it by saying it will be an overhead for the students to learn first about the algorithm and then about the Game API.
for an year i dejectedly saw that not everyone is as passionate as you are. I found no market for my programming game. If it would have launched somewhere in US, it could have been better, since MIT has such a kind of competition in which students needs to make the bots. it is not that programming games have no market, there are games like : Battlensakes, coderOne etc. but their market share is very less.
I learnt the lesson the easy way i guess, because i had a safety net since i was still in college, and had a job from the following year.
But then i really understood about product market fit, which i used to ignore while they taught in entrepreneurship classes. If anyone wants to see how the game looked : https:\\aiplaygrounds.in . I have revamped by business idea and working on something else.