Ask HN: How to Do a Startup in 2020?

It's the same as it's always been: build something people want and acquire customers.

> I think it would need the backing of something like ycombinator, techcrunch or techstars

That's a bad mindset, imo, and probably just a distraction. If it's such a great idea that you think it is, build it and acquire users.

I think it's worth glancing at https://www.ycombinator.com/resources/ - but I think more people use this as an alternative to building things and getting users.

Thanks yeah I started reviewing a bunch of these docs right after posting this. Appreciate the suggestion.

Back to the basics. Just build something people want.

Thousands of companies build things people want every year. Almost all of them go out of business without any customers. Building something people want is not nearly enough, not by far. You also have to sell it to them, keep them using it, and outcompete other companies in the same space. This takes wildly different talents, and if you don't have those talents, it takes a lot of money.

Saying "just build something people want" is reductionist well past the point of being useless, it's almost actively insulting. Customer acquisition is the death of almost every failed company.

You are correct. “Build something people want” should be followed by “don’t die” and “find an audience”. In that order.

Startups are still the same thing as they have been for 10, 20, even 100 years. You build something of value, and try to convert that value into assets. Acquiring companies won't look so much at your cashflow, but rather you as an asset that complements their existing processes. So the goal is basically to build as much asset as fast as possible, take it as far as you can, and sell that to someone who can take it further.

Bootstrapping is tough for tech, because the curve is usually exponential. If you were selling lasagna, you could easily make $1000 profit the first week, and bootstrapping is easy.

With tech, it's quite common to spend a year trying to get to $1000/week, and then make $30k/week the next year.

You also have to consider competition. There are many kinds of ideas. Making a Tinder for X might not have too much competition, but if you do something that Google/Amazon/Facebook is building, you'll get crushed if you move too slowly.

This is usually where venture backing comes in - investors are happy at getting good returns, and you're happy at skipping years of consulting, etc to pay the bills.

Don't spend too much time negotiating or optimizing; you're aiming for 3300% annual growth and a 20% better deal isn't worth your time. Paul Graham/Viaweb started at 10% for $10k. (http://www.paulgraham.com/ycstart.html) It seems unfair, but it was still a good deal for both sides.

well, you answered your own question - bootstrap it yourself. i know exactly where you're at, unfortunately.

my advice is - go with backend as api and separate frontend. define the api(protobuf+swagger plugin, not raw anonymous json because you will get lost in fluid schema later on and with 10+ routes you will go crazy without schema) and strict interfaces. so you know how data will flow. just pure basics, not all the functionality you have in your mind/vision. avoid DRY, copy code all over the place, go with kiss and most importantly YAGNI. use tech you know, not what is fancy right now.

then proceed to implement it in the fastest way possible. the end result should be a MVP with the worst code you can imagine. I myself have never been able to execute it, unfortunately. Each programmer strives for perfection which in this stage means time wasted not delivering the end product. I am certain that this way, no matter the size of the project, you should be able to produce something usable within 3 months.

You WILL be struggling with authN but mostly authZ(unsolved problem) so go for the simplest solution you can think of.

Next important thing is to do a monolith first. NO EXCEPTION. Forget microservices or SOA at this stage. I too fell into this trap and kinda keep repeating this mistake over and over. People like to use too much abstraction. Just brute-force it.

Then make sure you run one single VPS. No cloud, not kubernetes, no docker, nothing. Run it as close to metal as possible without any infra mumbojumbo fancy scheduler...tech..whatnot. And avoid cloud VPS. Look at OVH, Hetzner, DO or whatever. Just not cloud VPS because they are crazy expensive. Don't go for bare metal just yet(and avoid cloud altogether, it will bankrupt you in the long run, bare metal is where it's at).

Once you get the MVP up and running, look for first paid customer(s) and listen to their feedback. That feedback is free and no one else can give you better information than actual users. Also don't bother promoting the project on forums looking for people to test it for free. That is a dead end, trust me. Spend few bucks on some basic presentation page and some ads. Most notably it should be immediately clear on the website what your product does. I'd say 80% of companies these days fail at this horribly.

When you take payments, always upfront and annually will be better than monthly.

Once you get there, let us now :)

Good luck soldier.

Thanks man super solid across the board and mirroring what I've already kinda decided on almost all this. Good reminders too (aka use what you know, kiss, etc). Good stuff!

It really does depend on the idea. Is it a consumer product? B2B? In the last few years I’ve seen a growing number of companies start with little or no seed funding and grow to a respectable size. If you’re technical, I would start by building a prototype and trying to get a few paying customers to prove it something people are willing to pay for.

I’m curious to learn more. Reach out to me at sam@sellout.io if you want to chat.

It is a consumer facing product. Like the sites I mentioned (employment/dating, etc) it's driven by connecting people with a unique twist. I'm a little reluctant to discuss it more than this because I really do think this is the kind of idea when you hear it you think 'oh shit, yeah I'm surprised nobody thought of that already!' In fact I had this idea originally in 2005 but didn't have the confidence to try to go against the existing players (I also didn't know anything about incubators, or any of that world yet either).

So yeah I would reach out but I also don't really want to disclose any details until I've got something running and hopefully users using it. Of course there are NDA's but I've always wondered if those are worth anything.

Thanks for the response either way, appreciate the offer.

Ideas are a dime a dozen. Execution is what matters.

Is your idea building "Tinder for job search"?

I don't say that to discourage you. Quite the opposite. If you're passionate about it, do it. As others have said, at the end of the day, it's quite a simple formula: build X, get people to use X.

Even more simply, "solve problem Y for lots of people".

Totally get it, and agree (and I have started building it - I am always down for a new fun, exciting side project).

But this one feels different... and no it's not 'tinder for job search', I was just trying to use something to compare my idea market/space, etc without giving away anything (I'm paranoid that way).

I have bootstrapped a few apps... most notably a signup system to manage pick-up hockey games... I knew writing it it was a niche 'product' and didn't expect to make any money off of it and mostly just wanted it to cure a bunch of personal pain points of running pick-up/drop-in hockey games. Truthfully I thought I could get rinks to adopt it but after meeting with 3 rinks all owned by local muni's that idea fizzled out quickly. I did get one privately owned rink to use it (and they still do and love it), so it's fine, I make a little off it every few months (like 50-100 bucks) and it made/makes running my skates a breeze so all good.

Which is why I am trying to avoid that same fate... aka if I bootstrap this idea, there is no way I'm going to be able to get the level of users to compete with a Linked-In or a Tinder. To me that requires something like Ycombinator getting onboard and throwing their weight and resources behind it.

So I guess I'm saying I am pretty realistic about this idea... it's not a rink management or drop-in/pick-up management type thing that is very limited in the scope/scale. That idea never had the potential of truly blowing up.

This idea is more like a linked-in, tinder, facebook, instagram, giant, massive, holy fuck type of idea... and yeah ideas are a dime a dozen, but to me that's only true when that idea is coming from a friend, an uncle, a buddies wife... aka someone that doesn't really know how to qualify an idea like this.

I think I have my decision though, I feel that apply for Ycombinator is the right way to go; I'm convinced it's that level idea, that warrants that level support to get it to world domination levels.

Thanks for the reply.

I think aiming for YC long term is smart, but YC doesn't really take ideas anymore as far as I know. To increase your odds of getting in you still need to do all the things that you would do anyway: build a prototype, get some users (preferably paid). You have to show the ability to execute to get YC on board these days.

Thanks - I would never apply with just an idea. I'm a senior software engineer and I've already started building the prototype.

My bigger challenge is whether to try to find 2 people to partner with, and if so who/what roles that would be. I know generally speaking solo founder startups are less likely to be picked up. Debating bringing on a designer or maybe a mobile dev(s) (I do back-end/db/web so I have that covered).

Interesting, I think the opposite. Of course there’s the bare minimum needed to execute properly, but past that I think the idea and market are much more important in the long run.

For example, if the TAM isn’t there, you can execute as well as you want, but you can’t increase the physical limit on number of customers.

Your opinion is totally valid, and I agree.

I should've been a bit more clear... On the list of things that are important, having a "good idea" is near the bottom ... unless you're lumping in market validation, timing, total addressable market, etc. into "good idea". In this case, I wasn't. Those are all on the list of things that need to be considered though, and to me, fall directly in line with "execution", rather than "idea".

Again, this is just the way I see it, and it's mostly a semantic argument. But to me, "Idea" is like the hypothesis, and problem statement. "Execution" is proving that out by building a really good product and iterating on it, along with analyzing TAM and PMF. If you're better at the "execution" component, you're going to find more success. Especially if you're good at doing that portion quickly and reliably and can cycle through several "ideas". Therefore, ideas are a dime a dozen (in my opinion).