How I run multiple $10K MRR companies on a $20/month tech stack
52 points by olegkovalov
52 points by olegkovalov
I am much more interested in how the author is identifying multiple problems to solve that earns $10K/month. In fact, without information on the problems being solved, the infrastructure cost lacks necessary context.
Agreed. A discussion about how to identify niches that are worth $10K MRR would be far more interesting.
The whole tech discussion is ... not exactly useless, but pretty close. If $5 vs $200 per month is relevant, you're likely doomed anyway.
The answer at the kind of production levels being talked about has always been "Whatever stack you know well, USE IT! Knock the features out, forget about the tech, and go spend your time on actual business tasks like sales, marketing, support, etc."
Computers are so powerful that its really not relevant whether you are using Go, Django(Python), Laravel(PHP), whatever.
He also wrote this post on finding ideas: https://stevehanov.ca/blog/finding-great-ideas-for-your-startup
And this post on sales: https://stevehanov.ca/blog/bending-over-how-to-sell-your-software-to-large-companies
Yeah me too. I spend so much time trying to think of passive income ideas and it's probably no surprise to anyone that it's freaking hard!
That would result in this article to be moderated away as buisness stuff if off-topic here.
Obviously AI-generated image on the sequence diagrams page. After that:
Stop fighting with automated graph-routing algorithms. Our custom C++ engine effortlessly renders dozens of actors and complex conditional loops without overlapping lines or visual clutter.
But the post says:
I write my backends in Go.
... I've flagged as Spam. [Edit: Removed my flag based on @singpolyma vouching for the author below.]
I don't understand what exactly your complaint is? Yes, Hanov writes most of his new backends in Go. Yes, websequencediagrams is a much older project and has a big chunk still written in C++. This is not a contradiction. And certainly isn't SPAM.
I understand that it's certainly plausible to have that tech combination.
Looking at the sites linked, those seem to be very similar to typical vibe-coded sites, which reduces the credibility of the post.
There is no way to flag as "Seems AI-generated & potentially fake"; I think spam is the closest flagging option.
I can assure you that Hanov is a friend of mine and websequencediagrams predates the concept of vibe coding by many years.
Not that he hasn't used any AI assistance for updates to it recently, he admits as much in the article, but it's not just a piece of slop with no revenue. It is what pays his bills.
I see, I didn't make the connection that you knew the author, I thought you were coming up with a potential interpretation for it.
I've removed my flag.
Nah, I don't see a problem when most of your (web)backend is in Go except some intense computation ins C++/Rust/Zig tbh.
Can be treated as a spam but this jumped to the HN Top and I decided to share it here. (I'm not an author btw).
The marketing copy's "dozens of actors and complex conditional loops" sounds like a far cry from "intense computation"...
As someone who is considering launching a small project as a little commercial venture the things I’m running into:
Merchant of Record: If I take payments, I’m responsible for tax compliance for every jurisdiction someone signs up from. Places like Lemon Squeezy will be the MoR and handle taxes but they take $0.50+15% per transaction, which is a ton.
Authentication and Identity: The author here says they wrote their own auth library with SAML integration and everything, which is neat, but I feel like handling account lifecycle internally is asking for trouble. I’ve looked at Clerk for auth/identity and it seems okay and, more importantly, it’s one less thing to worry about.
Subscription and Billing: Related to the MoR thing above, I don’t wanna keep people’s credit cards on file nor do I want to have the risk of accidentally billing people incorrectly or handling refunds. Stripe and Lemon Squeezy do this but they aren’t cheap.
Monitoring: A single VPS is great but I guess I don’t know the state of the art for monitoring these days. What do people do to make sure their stuff is up and actually working?
What I've learned:
MoR: overkill without traction
Auth: local auth is fine (Devise/BetterAuth/etc), or use Supabase free tier if you want it to be hosted
Payment processing: there's no getting around needing one. I don't have enough experience to recommend a particular one yet.
Monitoring: also wondering about this
One problem with this space is you go to learn about something and it sprawls out further. Today I learned card testing is a thing that your site can be used for, so some rate limiting is helpful.
Would you mind explaining “overkill without traction” there? I guess I don’t wanna find out I owe a billion dollars in fines because I didn’t account for city sales tax in some French village.
If you didn't make a billion dollars you won't owe a billion dollars. Especially if you never visit that village...
From doing more research, I'll probably walk back my MoR comment here. They provide more than just help with global taxes. Evaluating polar.sh at the moment...
Merchant of Record: I've seen paddle on a few invoices from SaaS apps, which also handles taxes for $0.50+5%.
Monitoring: I just set up Honeybadger for my project. They provide a pre-baked solution for my stack, so got it done in ~10 minutes. Not as powerful as Grafana, but has enough for my immediate needs: Health checks, response times, and a list of slow db queries.
Subscription and Billing: Related to the MoR thing above, I don’t wanna keep people’s credit cards on file nor do I want to have the risk of accidentally billing people incorrectly or handling refunds. Stripe and Lemon Squeezy do this but they aren’t cheap.
Check also Authorizenet (Dinosaur interfaces, when I integrated with them -- but may be ok). Also check with your business bank may be they will recommended the online CC processor with a recurring billing option (that's how I got the Authorize.net recommendation). May be you will get a better deal.
Correct, you do not want to store or even process through any credit card info -- instead you need to display the Auhotirze.net (or whatever processor you choose) page to take users info, and then get back a secure token, and then use that.
you can also choose a 3rd party SaaS to implement recurring billing (search for recurring billing SaaS software -- they will let you typically configure the rates, suspensions (if you support that, and so on).
In the past I wanted to build one of these recurring Billing saas (with lots more features) -- as I am 'back belt' in that area (and taxation, etc) -- but I just did not figure out a model where I could do it on my own! ).
Stripe and Lemon Squeezy do this but they aren’t cheap.
Stripe is basically standard. What are you looking at that is cheaper?
Stripe is cheaper than Lemon Squeezy but doesn’t (to my knowledge) act as Merchant of Record, so I’m on the hook for tax compliance.
A direct merchant account and doing your own card processing is cheaper (Visa CyberSource for example) but it carries a much higher assumption of risk.
Looking at the linked websites I’m surprised the author brings in 10,000 USD (CAD?). Maybe I underestimate how much you can make from these products.
I’ve spent most of my life building software tools for human genetics. I would love to somehow make a steady 10K per month by solving problems for that community but I kinda fear the big companies won’t buy from a one man startup. I guess maybe the lesson of this blog is that you can bring in quite a bit of revenue from individual end users if you solve a real problem for them?
personal observations:
a) Individual earning's do not often correlate to the value they represent to a community
b) The authors tools seem to be aiming at specific end-user segments: professional or advanced amateurs (as in case of stock trading). This type of aiming is somewhat more profitable than targeting occasional end users.
I think the general sentiment (my translation, so to speak) that the author is expressing for professional programmers:
Learn how to express your business ideas into applications that can be done by a single-person, with a minimal investment into infrastructure.
This way, you do not need outside investment to start. And by the time you choose to get outside investment, you may haves significant business, so that the financial success you share with investors, is appropriately sized.
I think it is a practical advice (but also not easy to implement!). I am sure that technology suggestions in the article are not the 'core' ideas, though.
Overall, in my mind, it is unusual to see this type of post in this forum, i would expect it more for the orange site where there are abundant stories and subsequent discussions of financial success, geo-political/policy/economy (that have a very casual link to software dev technologies).
This way, you do not need outside investment to start. And by the time you choose to get outside investment, you may haves significant business, so that the financial success you share with investors, is appropriately sized.
But to put this in "reality" context: $10k/month is not enough to raise interest with any significant investors. From my own experience, you'd need to hit between one and two orders of magnitude more than that, with a seriously steep growth curve as well. And at that point, you really don't need investors 🤣
37Signals / Bascamp have some articles and or books on this - their strategy back in the day was to sell to teams not corporations. Managers have credit cards and the ability to get their teams the software that is needed - monthly pricing should be somewhere below a typical single transaction limit and also remember to let users put a billing email that is different to users email.
Sometimes you need the absolute cutting-edge reasoning of Claude 3.5 Sonnet or GPT-4o for user-facing, low-latency chat interactions.
On a month old blog post.
I press X to doubt on this one. Sustaining a business like that not only needs software, but also support / marketing / sales / acounting. Also the author didn‘t really touch on some other limitations of sqlite like schema migrations and failover. Those things are solveable (and for the app I am developing, I did), but not mentioning them seams like the author‘s advice isn‘t a 100% genuine.