Pursuing a Business I'll Love
mtlynch.io
mtlynch.io
> Another recurring theme in interviews was that nobody wanted a managed service. This surprised me because so much of the trend in software in the last 20 years has been toward SaaS, managed services, and monthly payments. Sheet metal shops wanted none of it. [...] While SaaS businesses often highlight the advantages of low up-front cost and minimal IT maintenance, sheet metal shops didn't value these things. Independence mattered far more than convenience or cost. Multiple shops told me that they didn't want their businesses grinding to a halt because their software vendor went out of business or experienced a network outage. I asked if they minded maintaining their own servers, and I was surprised to hear consistent “no"s. [...] Given that some of these businesses have been operating for 70+ years, I see why they plan to outlast their software vendors. Still, this sentiment surprised me because it seemed to contradict most of the conventional wisdom about SaaS sales.
"It is difficult to get a man to understand something when his salary depends upon his not understanding it." - Upton Sinclair
Look at where this "conventional wisdom" comes from and it'll all fall into place.
If the customer is a startup themselves and therefore has a very short expected lifespan, this does not necessarily matter because the lower upfront cost enables cheaper experimentation and pivoting. If you are in a business with product/market fit that has been around for some time already, the lowest possible upfront cost is much less of an issue compared to the total cost of ownership and the risk of a critical dependency disappearing. SaaS startups do poorly on both of those metrics.
Basically every company that's been around or plans to be around for more than a few years is primarily self hosted.
Chat and email, e.g. stuff that's tangential to the work, might be outsourced, and they might use (the equivalent of) EC2 and S3 but ultimately it's your software running on your VM's.
Scrappy startups that don't actually make any money but are trying to thrash something together quickly to convince a greater fool tend to use way more.
I honestly don't know, so that is an honest question. I would love to hear from Autodesk customers on that topic. The former is possible. But the latter is too. Businesses love subscription revenue for all kinds of reasons, including just that it makes the accounting work out well. They might be willing to even make their customers take a bit of a hit relative to what they want to force something into a subscription model for their own needs.
It also carries enormous risk of anything happens to the device with the data on it. Employee that just got fired drops laptop with project plans in an oil drum on the way out? Yeah that happens.
All of the big players in construction project management are cloud and mobile and anyone that tried to pitch anything different would be at a substantial disadvantage for all but the smallest contractors.
>> I thought that rewriting the Zestful website as a static site would improve its SEO. Sadly, Google Search Console suggests that it had a negligible impact, the enterprise customer approached me only a few days after I rewrote the site, so I can pretend to myself that the rewrite was responsible for ~$3.9k in additional revenue.
I had a chance to connect with Michael (about keto stuff) a couple years ago. He had reached out in a very humble, welcoming way.
On the call, I recall I pompously gave him advice, not realizing how far ahead he was on his entrepreneur journey. He graciously welcomed it all, and let the conversation flow.
Looking forward to learning from you and reading future posts!
I wasn't being gracious at all. You had a lot to teach me! You were great at building a community and an engaged following in a way that I still haven't been able to replicate for my keto site. When I write ad copy now, I remember the good tip you shared with me about writing it as if to a single person.
It's a surprise to see this on here because I mainly write the retrospectives as an exercise for myself and don't expect others to find them very interesting.
I'm glad there were lessons I learned in this one that people found useful or fun. I'm running around today, but I'll be back on later tonight to respond to comments.
The metal shop example was a surprising and good one - so some businesses actually don't really want you to solve their problem in the best way, they want you to solve their problem adequately but it's more important to give them flexibility and options for the future.
The biggest insight is what makes me a good technologist, does not make me a good anything else (entrepreneur, SaaS proprietor). I gotta figure out how to get there.
The line about Word having too many features is that everyone uses 10% of the features... the problem is that everyone uses a slightly different 10% of the features.
One thing you should try hard not to do is look at the competition and write it off as crap. It's highly unlikely that your antecedents set out to make a shit product. It's likely that when they were starting out, they wanted to create something efficient and streamlined, but then years of paying customer feedback took things to where they are today. The folks at 37signals always used to remind us that it's much easier to add a feature than it is to remove one.
I wonder how much of the shopkeepers reactions were fighting change vs just not trusting someone who didn't come from where they came from. There's real class anxiety around the blue collars vs the 20-something software millionaire set (which is likely how they saw you, as the other, even if you're actually their neighbour). I wonder what would happened differently if it was the owner of the first shop leading the pitch. I often advise founders that they shouldn't get into any business where one of the founders doesn't have deep industry experience and connections. If nobody on your team has ever managed a hotel, it's highly likely that you will fail to disrupt hotel management. Anyhow, it's unclear from your writing but seems like your first customer guy was just that - a customer. In other words, he found someone to spend a month doing free fan service for an idea that if he was truly that excited about it, he would have hired you or at least proposed a partnership and been much more aggressively involved in doing the initial customer development conversations with you.
I'm intrigued by the notion that they have no problem installing their own servers. This pretty much lines up with a certain segment of blue collar and [what I'd describe as Meyers-Briggs 'S'-type professions]... heck, I've seen those computers. They are always covered in a slick of oil and dust particulate. I wonder if they would have been comfortable with an on-site solution that is also connected to a central service, like Github Enterprise. In this way, they have working infrastructure even if you decide to ditch software for comedy, and you can both push new updates and offer remote backup (which is good for risk mitigation) as a paid extra service.
Do sheet metal shops have a trade magazine? Is there an annual sheet metal trade show in Vegas? You might still find it interesting to attend.
Note that of course you're not passionate about sheet metal. The entire premise is ridiculous because even the sheet metal people see it as a job. The whole point of doing something unsexy is that in some ways it's hard because not everyone is willing to do it. I would change and do some things differently, like getting that first customer to put a lot more time and money on the line to explore the concept in a partnership capacity, but if you want to follow the FreshBooks trajectory you have to start with a fractional subset of functionality targeting a micro-demographic and THEN grow into the Intuit competitor over the next decade or so. There was never going to be an overnight success and you'll probably need someone to put in some adventure capital so that you don't burn out after a single month of customer development.
While it's true that you failed to come to the table with a solution to a hair-on-fire problem the other sheet metal shops were having, try to remember that they have good reason for conservative skepticism about something that could turn their business upside down. A lot of this stuff is just an expression of the lowest levels of Maslow's Hierarchy. Everyone wants to be the hero of their tribe, but nobody wants to be humiliated in front of their tribe. The only way to sell into those shops would be for your first customer to make the leap, and then have someone else that already agrees with that customer make the leap and show everyone else that it's safe.
https://www.youtube.com/watch?v=V74AxCqOTvg
It's not the first customer that is the most important. It's the second one! Also: flying doesn't kill you, landing does.
In the end, his comments about how someday they might need the extra features is probably most telling. Deep down, people are scared of the future and they make a lot of irrational decisions based on things they've already done but also on things that they imagine they will do in a perfect future. People will double down on purchases that they've already made so they don't have to confront the quality of their first decision. And people will buy books on Amazon that they don't have time to read because in an imaginary aspirational future, they have an opportunity to real all of the books, and that person will be so smart and awesome compared to themselves today. A huge amount of purchasing in North America is aspirational. Kitchen appliances much?
Onto the comedy app! Bandsintown/Songkick for Comedy is a really great concept. I agree that scraping is a fools errand. However, it's also really important that you replace scraping not with your time but with someone you hire from the internet. Not only is your time way too valuable, it's not a realistic test of what will be necessary to scale it even a little bit. I definitely think that you should find some comics willing to pay even a token amount to use your service to host a reasonably nice homepage for them. You might find it useful to watch this presentation by the founder of Bandcamp at #XOXO. https://www.youtube.com/watch?v=MaUkS-lr-ZM
I do have a small beef: you say that you were making an MVP. What could possibly justify using Vue or any SPA to build something this simple? When did it become the default to start an MVP using tools that are designed to solve Facebook scale problems? I know that it's not just you and I'm probably farting in the wind, but this topic gets under my skin because the day absolutely will come where people look back on the SPAs-for-everything fad and profess that "not everyone was doing it, just the people who didn't realize they were conforming to the pack and reinventing simpler tech that already worked better than it ever had in the past". Your filtering needs were pre-solved by Isotope, for example. https://isotope.metafizzy.co/
Maybe they just did a sensible assessment of the options.
- Is there a risk in being dependent on an SaaS business that may rise prices or go out of business? Yes, definitely
- Are sheet metal shops in a position for upfront investments in an asset that will be useful in the long term? Yes, definitely, and more so than an enterprise customer who'll just expend a 200$/mo charge on the company credit card
- Are sheet metal shops used to taking care of business-critical assets? Yes, definitely
Much of today's SaaS software is based on the premise that people can buy it without permission from the higher-ups because
- the monthly charge is low enough to escape scrutiny
- the function of the software is not business-critical enough that lock-in is a problem or it looks cool enough to the CEO and the alternative is something clearly inferior (nothing, or Excel Macros)
Once you move out of the enterprise/startup space into structures which function differently (and it's not that they are "fighting change", it's just that they're built on different foundations), you need to check these assumptions.
Limiting yourself to passion projects really limits the markets you can move into because there's a lot of correlations in what we all have passion for. Comedy is at least not in the A-list for programmers (it's not a text editor/IDE, video game, or web framework, etc.), but it's still not sheet metal, logistics, accounting, or something else that can have clear value to customers, who will give you real money.
I have a clear through-line on how a sheet metal software shop will make money, in quantities sufficient to live on, and indeed, to start hiring people from. I've got a much less clear through-line on how a "comedy directory" is going to make that money. I mean, I see one: ads, of course. But it's a much more amorphous and fussy through-line. Heck, the reasonable best case for the comedy site economically, "low-level life-style site", is perilously close to the worst case for industrial software, "$0 in revenue". That's not a lot of daylight there in business terms, partially due to the volatility inherent in an ad-based business.
Not saying you should hate what you work on, but opening things up to things that you're doing to make money broadens your possibilities a lot. You may well find some unexpected passions along the way.
Yes, this also applies to porn stars.
>One thing you should try hard not to do is look at the competition and write it off as crap. It's highly unlikely that your antecedents set out to make a shit product. It's likely that when they were starting out, they wanted to create something efficient and streamlined, but then years of paying customer feedback took things to where they are today.
Yeah, you're right. I definitely had a bias to look for the weaknesses in their product, and I'm probably overlooking the positives. And I might be overweighting the fact that the UI is ugly and seems to have poor design. But it also suggested to me that it's a company that's been comfortably entrenched for a long time and has never had a competitor to force them to continue improving it.
>I wonder how much of the shopkeepers reactions were fighting change vs just not trusting someone who didn't come from where they came from. There's real class anxiety around the blue collars vs the 20-something software millionaire set... In other words, he found someone to spend a month doing free fan service for an idea that if he was truly that excited about it, he would have hired you or at least proposed a partnership and been much more aggressively involved in doing the initial customer development conversations with you.
I agree. There was definitely a disconnect. I'm not sure if it was blue collar vs. spoiled software kid as much as sheet metal vs. not.
The first shop owner actually did offer to partner with me, but I decided against it. I know it would have increased the odds of success, but the independence of this lifestyle is one of the most important parts to me.
>The only way to sell into those shops would be for your first customer to make the leap, and then have someone else that already agrees with that customer make the leap and show everyone else that it's safe.
Yeah, I think there's kind of a catch-22 there. Because I'm not super passionate about sheet metal, I wouldn't want to commit the next X years of my life to sheet metal unless I had assurance that I could make at least a few thousand dollars per month. But my first customer doesn't want to take a leap of faith with someone who's going to disappear if 4 more customers fail to materialize, so we're sort of stuck.
If I found 3-5 customers who showed strong interest in purchasing, then I'd be willing to invest 3-6 months building an MVP for them, but with just one customer, I'm afraid to put months of my life into something and have just one customer for years.
>I agree that scraping is a fools errand. However, it's also really important that you replace scraping not with your time but with someone you hire from the internet. Not only is your time way too valuable, it's not a realistic test of what will be necessary to scale it even a little bit.
Oh, I actually disagree with this. I think scraping is too expensive for the MVP, but I plan to eventually move to scraping. The website I mentioned, KetoHub, works on scraping. It scrapes 16 different blogs, and I've only had to fix the scrapers 2-3 times in 2 years. I can usually write a scraper for a previously unseen cooking blog in about 2 hours.
I expect comedy clubs to be pretty easy to scrape because they don't change their layouts very often, and there are probably O(1000) comedy club websites in the US.
>You might find it useful to watch this presentation by the founder of Bandcamp at #XOXO. https://www.youtube.com/watch?v=MaUkS-lr-ZM
Interesting, I'll check this out. Thanks!
>I do have a small beef: you say that you were making an MVP. What could possibly justify using Vue or any SPA to build something this simple? When did it become the default to start an MVP using tools that are designed to solve Facebook scale problems?
I'm a bit confused by this point. What is the simpler alternative to Vue? Just write bare HTML and copy/paste it for every route? PHP?
I developed KetoHub using Angular, and I think your feedback applies to that project. Angular was such a hassle to work with because there are a million configuration things that are a pain, and then there are breaking changes every 6 weeks. It's fine for enterprise, terrible for a one-person dev shop.
The project I did after that was just Flask, Jinja templates, and two lines of JS. But that was kind of a pain to work with, too. There wasn't much complexity, but there was a lot of repetition and it was hard to organize the code. I ended up with lots of small template files but without any of the compiler support I'd get from an Angular/React/Vue.
For WanderJest, I just reused the architecture I had from a previous project where the backend is Go, frontend is Vue. I haven't yet run into an issue where the tech stack has been a bottleneck. I feel like it's actually been the opposite, and I'm moving faster on this project than I've ever moved on a previous project. I'm comfortable with the stack, and I feel like it's fairly simple.
>Your filtering needs were pre-solved by Isotope, for example. https://isotope.metafizzy.co/
That requires me to research options, understand Isotope's commercial license, buy it, then learn how to install and use it. And then repeat that process for every feature I need.
I already know how to get around in Vue, so that's the most expedient path for me.
What I'm about to say is pretty fuzzy, but I have noticed that services based on scraped datasets trend towards being predictably shittier than services lovingly harvested by a human. Obviously it's domain dependent but my bias is that it seems like scapers pick up a lot of crap; in music, it'd be the singer-songwriter showcases that literally even the performers families don't want to go to more than once a year.
I would filter out a service sending me 100+ mediocre, repetitive listings every week. I would consider paying for 5-6 human curated listings that strike a blend between "don't miss Judd Apatow" and "you might like Little Esther" up-and-comers. It's the human giving a shit that is providing value. I am someone that if I go to something I don't love, it is far worse than if I didn't leave the house.
So, yeah. It's all going to come down to what kind of customer you want. O(1000) comedy consumption sounds like someone who doesn't care how they are entertained so long as they don't have to contemplate how small they are in the universe. Like, if someone in my circle said "let's go see some comedy!" I'd say "oh, who?" and if they looked at me perplexed and said, "I dunno... comedy!" I would consider this grounds to reevaluate this relationship.
"Let's go see some music!" "What band?" "Oh, just some music!" gun
Anyhow. :)
And hey, if you're moving fast and love Vue, keep rocking it. I just meant... use Rails. Use PHP. Use anything that emits HTML that doesn't need to be rendered in JS on the client. This is just my opinion. If you've never tried Rails, perhaps consider trying Rails. Most of the other server frameworks (Laravel, Express, Django) are just efforts to copy Rails in other languages, but they never quite nail it.
Oh, I think scraping and curation are orthogonal. I agree that if I just scrape anything from the web that looks like a comedy show and don't do anything to help the user sort the good from the bad, it's going to be a bad time.
I want my site to be useful for casual comedy fans who just want to see big names like Judd Apatow or Jerry Seinfeld. And I also want it to be useful for comedy nerds who do want to see open mics and alternative comedy shows. Both experiences should be possible with automated scraping + curation/ratings/filtering.
> And hey, if you're moving fast and love Vue, keep rocking it. I just meant... use Rails. Use PHP. Use anything that emits HTML that doesn't need to be rendered in JS on the client. This is just my opinion. If you've never tried Rails, perhaps consider trying Rails. Most of the other server frameworks (Laravel, Express, Django) are just efforts to copy Rails in other languages, but they never quite nail it.
Ah, okay. I've never tried Rails, but I have been interested. Moreso in the last few years as I become a bigger fan of Basecamp. Maybe I'll find an excuse to learn it this year for a hobby project. PHP, I used for a while ~15 years ago, and I don't really miss it, but maybe the experience has gotten better since then.
Here's a promise: if you ever want to jump into Rails and have any questions, I'm willing to put in as much time and energy helping you get started as you are to soak it up.
I had to switch to reader mode in Firefox to get any text?
The same behaviour occurs on all the website...
What gives?
Here's a capture for anyone interested: https://demo.intellique.org/nextcloud/index.php/s/xM6QCQXReF...
I've been a corporate coder all my career, but sometimes wonder what it must be like to run my own show. Your writings are highly interesting.
Facebook, on the other hand, often takes nearly a minute for the UI to be fully rendered.
OpenComedy has a funny bug where it seems to think I live in Northampton, UK because I like near Northampton, MA, USA: https://imgur.com/BVEZWbA
One of the recurring ideas that comes up a lot in my discussions with bookers and comedians is that bookers have a hard time finding good comedians to fill their shows and comedians have a hard time finding out which shows need comedians and what the details are. I've thought about some sort of matchmaking functionality, but as you said, it's hard to make money from something like that. I'd still like to tackle that, but I feel like I should get the money part figured out first.
Would like to know more if you have a source.
~200 companies a batch and supposedly a 2-3% acceptance rate puts it at 6,600-10,000.
They stopped updating their statistics: https://twitter.com/sama/status/618247698378985472 and https://blog.ycombinator.com/yc-portfolio-stats/
If it kept growing they would have proudly put that everywhere.
At the same time they introduced startup school which got 15,000 applicants, many of whom would have applied to Y Combinator but chose the startup school instead.
My bet is it peaked around ~8,500 and decreased when they introduced the startup school program.
Also, from the figures presented at the conclusion of the program last year, there were ~23k applications to the 2019 summer & winter batch in total.
fewer hackers