Ask HN: What open source business models have worked?
18 comments
> If you have a technical user who is perfectly capable of self-hosting, and I would argue even gets joy out of doing so, why would that user let go of that joy and pay up?
Many years ago, I installed and managed OpenWRT on my routers, and FreeNAS on a home-built NAS. It was fun and interesting for a while, but I grew tired of the maintenance and keeping up. It had moved from hobby and became a job. I was full-time network administrator for my own house.
People are infinitely variable. Some will get a lot of fulfillment out of doing it themselves. Some will not. Some will lose interest in self-hosting, but still want your service. There might even be someone who makes an OSS version of the services that you offer for money (see SorryCypress)
Avoid getting too wrapped up in the psychology of your users.
Many years ago, I installed and managed OpenWRT on my routers, and FreeNAS on a home-built NAS. It was fun and interesting for a while, but I grew tired of the maintenance and keeping up. It had moved from hobby and became a job. I was full-time network administrator for my own house.
People are infinitely variable. Some will get a lot of fulfillment out of doing it themselves. Some will not. Some will lose interest in self-hosting, but still want your service. There might even be someone who makes an OSS version of the services that you offer for money (see SorryCypress)
Avoid getting too wrapped up in the psychology of your users.
> Many years ago, I installed and managed OpenWRT on my routers, […] but I grew tired of the maintenance and keeping up. It had moved from hobby and became a job. I was full-time network administrator for my own house.
Wait, what??
I check OpenWRT once every several months for a new release, and if one exists I spend an hour or two updating my gateway and wireless bridges. There is no extensive maintenance needed even when running multiple servers behind it.
Honestly, I run two VM platforms, a dozen or so VMs, several workstations and a mess of personal machines and devices, and I think I’ve spent half a day on my network in the last year.
Wait, what??
I check OpenWRT once every several months for a new release, and if one exists I spend an hour or two updating my gateway and wireless bridges. There is no extensive maintenance needed even when running multiple servers behind it.
Honestly, I run two VM platforms, a dozen or so VMs, several workstations and a mess of personal machines and devices, and I think I’ve spent half a day on my network in the last year.
Well, all I can say is that this was about 15 years ago. Updates to OpenWRT were not fun, and I was never very confident that I had done things correctly. Whether it was my shortcoming in technical knowledge, or poor documentation, it was more than I had patience for.
Let's be honest --- giving your work away for free has never been a good, reasonable basis for a "business" model.
Attempting to monetize after the fact often involves some sort of ethically questionable, bait and switch type behavior.
Attempting to monetize after the fact often involves some sort of ethically questionable, bait and switch type behavior.
There are things you could charge money for even if it is FOSS, such as:
- Priority support (or even, any support at all, if you prefer to not offer any free support at all)
- Preinstalled hosted instances of the software (in case the user does not want to set it up themself on their own computer)
- Physical media, e.g. CDs and DVDs
- Printed documentation
- Customized software solutions (if the user does not wish to modify it by themself)
- Precompiled binaries (possibly, only when using commercial "app stores" e.g. in iPhone/iPad)
- An explicit license to use it in case your company requires an explicit license (this is how it is done with SQLite, which is public domain, but charges money for a "Warranty of Title" in case you need that)
- Computer hardware designed to work with the software
You can combine multiples of the above. (Which are appropriate would depend on what your business is; some are not appropriate for some kinds of products/services.) (In some cases, I might even be willing to pay for some of the above, e.g. for some software, I might like to purchase printed documentaion if it is available, even if I can also access it for free on the computer.)
There are other possibilities, but I think that many of them will be unethical and should be avoided.
You do not have to force everyone to pay; someone who does it by themself will do so, and might refuse to pay anyways even if you do change the license (might instead prefer to no longer use it, or write their own, etc).
- Priority support (or even, any support at all, if you prefer to not offer any free support at all)
- Preinstalled hosted instances of the software (in case the user does not want to set it up themself on their own computer)
- Physical media, e.g. CDs and DVDs
- Printed documentation
- Customized software solutions (if the user does not wish to modify it by themself)
- Precompiled binaries (possibly, only when using commercial "app stores" e.g. in iPhone/iPad)
- An explicit license to use it in case your company requires an explicit license (this is how it is done with SQLite, which is public domain, but charges money for a "Warranty of Title" in case you need that)
- Computer hardware designed to work with the software
You can combine multiples of the above. (Which are appropriate would depend on what your business is; some are not appropriate for some kinds of products/services.) (In some cases, I might even be willing to pay for some of the above, e.g. for some software, I might like to purchase printed documentaion if it is available, even if I can also access it for free on the computer.)
There are other possibilities, but I think that many of them will be unethical and should be avoided.
You do not have to force everyone to pay; someone who does it by themself will do so, and might refuse to pay anyways even if you do change the license (might instead prefer to no longer use it, or write their own, etc).
Cygnus was founded in 1989 and had success. Red Hat started in 1993 and was successful. MySQL was released in 1994-1995 and was successful. JBoss started in 1999 and was successful. A lot of these companies ultimately were swallowed up by IBM, except for MySQL which was ultimately swallowed up by Oracle. They were successes and lessons can be drawn from them.
Google releases the Android Open Source Project, the basis of things like LineageOS. Facebook releases React under the MIT license. These are not open source companies, but they release major software under FLOSS licenses.
Google releases the Android Open Source Project, the basis of things like LineageOS. Facebook releases React under the MIT license. These are not open source companies, but they release major software under FLOSS licenses.
There are a small number of OSS companies you can point to as being successful. Mozilla is successful, but basically their revenue comes from Google. So certainly one lesson there is to find a single giant benefactor.
Google, Facebook etc do have open source code, but not their "revenue earning" product. So perhaps the lesson there is to make lots of money somehow else and use that to fund your OSS projects.
MySql was kinda Open Source, but their license violated OSS principles. So the lesson there might be "call it Open Source, but lie".
Red Hat took a project made by someone else, avoided the development costs, and monetized packaging and support. (The then pushed some of that revenue back at the upstream projects.) So I guess the lesson there is to minimize development costs and maximise marketing.
None of which negates your point that some small number of companies have found success. But I'm not sure this helps random developer monetize their OSS project today.
Google, Facebook etc do have open source code, but not their "revenue earning" product. So perhaps the lesson there is to make lots of money somehow else and use that to fund your OSS projects.
MySql was kinda Open Source, but their license violated OSS principles. So the lesson there might be "call it Open Source, but lie".
Red Hat took a project made by someone else, avoided the development costs, and monetized packaging and support. (The then pushed some of that revenue back at the upstream projects.) So I guess the lesson there is to minimize development costs and maximise marketing.
None of which negates your point that some small number of companies have found success. But I'm not sure this helps random developer monetize their OSS project today.
>> Conventional wisdom says that you should build a large community of users, and then you'll figure out how you monetize a subset of those users
This is really just wishful wisdom. Conventional wisdom starts by deciding what value you can add to society and who will pay for it.
Doing the work now, and figuring out the business later isn't a strategy that often ends I success.
As a provider I create OSS out of the excess created selling commercial software to customers. The OSS stuff doesn't pay for my coffee, never mind anything else.
This is really just wishful wisdom. Conventional wisdom starts by deciding what value you can add to society and who will pay for it.
Doing the work now, and figuring out the business later isn't a strategy that often ends I success.
As a provider I create OSS out of the excess created selling commercial software to customers. The OSS stuff doesn't pay for my coffee, never mind anything else.
A "simpler" tool that is easily self-hostable is hard to monetise. For example, I self-host several tools especially frontend such as Ghost, Gatsby, etc.
However, I also have experience of seeing "open core" work as a good business model for several providers. In a complex setup, while the core can be self-hosted for smaller workloads, the need to run production-scale workloads means you need to buy the SaaS license from the provider. Gitlab comes to mind as an example.
However, I also have experience of seeing "open core" work as a good business model for several providers. In a complex setup, while the core can be self-hosted for smaller workloads, the need to run production-scale workloads means you need to buy the SaaS license from the provider. Gitlab comes to mind as an example.
These are not business models, but is a quick list of revenue models:
* retail
* advertisements
* subscription
* partnership/affiliate
* white label
* data broker
* consultant/freelancer
* resource timeshare
* licensing
* support fees
Normally business models focus on how to do business, this could include sales or operations, not build revenue. Examples of business models might include:
* security, threat analysis, and risk management framework
* talent identification/acquisition model
* regulatory compliance policy
* performance/research analysis model
* total cost of ownership
* procedures for operations and plans
* knowledge management and internal engagement
* systems architecture
* retail
* advertisements
* subscription
* partnership/affiliate
* white label
* data broker
* consultant/freelancer
* resource timeshare
* licensing
* support fees
Normally business models focus on how to do business, this could include sales or operations, not build revenue. Examples of business models might include:
* security, threat analysis, and risk management framework
* talent identification/acquisition model
* regulatory compliance policy
* performance/research analysis model
* total cost of ownership
* procedures for operations and plans
* knowledge management and internal engagement
* systems architecture
Open hardware, maybe.
It’s hard because it’s hardware, but generally speaking people are happy to pay for a finished product vs build themselves (which is also more expensive without economy of scale). Until you reach the scale of an Arduino, then knock offs become an issue.
So, I guess, up to a certain scale.
It’s hard because it’s hardware, but generally speaking people are happy to pay for a finished product vs build themselves (which is also more expensive without economy of scale). Until you reach the scale of an Arduino, then knock offs become an issue.
So, I guess, up to a certain scale.
Donations, Open Core, dual-license, paid binaries, to name a few. Important to note that models aren't a 'one size fits all', what works here may be a disaster there.
Have a look at the Commercial OSS (COSS) community at
https://www.coss.community/
NextJS on Vercel would suggest that you're wrong.
That said, perhaps anyone using NextJS is inclined toward convenience rather than deriving joy from self-hosting.
That said, perhaps anyone using NextJS is inclined toward convenience rather than deriving joy from self-hosting.
Building a service on top of the OSS, where some people prefer to self-host, but many don't want to self host.
Ex. Bitwarden, Infisical, Docker, etc.
Ex. Bitwarden, Infisical, Docker, etc.
You forgot to mention another model, which is to simply charge for support. No license change needed.
Well I didn't list all of the possible models, but yes.
I think the support model worked once, exactly once, and that was for Red Hat.
Not saying that you can't charge for support, but the days of building a company on support fees are long gone.
I think the support model worked once, exactly once, and that was for Red Hat.
Not saying that you can't charge for support, but the days of building a company on support fees are long gone.
gitawonk.com's premise is hosting your repo but charges cloners $2 to
get full commit history. We do for software what itunes did for music.
Yeah, we all know you can get the song on youtube for free, but pay us a
pittance and we'll make it more convenient.
Conventional wisdom says that you should build a large community of users, and then you'll figure out how you monetize a subset of those users. The standard argument that follows is that you'll offer a SaaS version of the open source for those users who don't want to deal with self-hosting.
I'm not sure that argument holds. If you have a technical user who is perfectly capable of self-hosting, and I would argue even gets joy out of doing so, why would that user let go of that joy and pay up?
The inevitable path then is a license change (Business Source, anyone?), or even going entirely closed-source (e.g. Panther did that)
Curious to hear how everyone thinks about this topic, both from a user and provider perspective.