[WalterBright]

A friend of mine is CEO of a medium sized company. His only product is software that helps back end servers run faster. The software is open source, he gives it away to any company that wants it.

His company is doing extremely well. How is that possible? After companies use it for a while, they come back and are happy to buy a support contract rather than manage it themselves.

His business would not exist without open source software that his company developed.

[dalke]

I have an open source package. It comes in two forms: a no-cost one you can just download, and full-cost one where you pay me and I provide support. Both are MIT-licensed. It's a Python library and set of command-line tools for chemical similarity search.

I've had a few sales. Those plus support renewals make me about $25K/year. Consulting and custom software development augment my income.

Recently I was at a workshop, and found that actually many companies use my software. I had no idea it was so widely used. One even presented the results of some 60 CPU days spent using my software. They used the no-cost version, and had never considered paying me for it.

(I asked. They said they had no internal mechanism to give gifts to open source developers. I said I could offer them a PO. They said they would need to justify it. I said I had a new version with better performance and more capabilities. It was a frustrating conversation.)

I am not convinced that the open source, no cost, business model, where you hope that people are "happy to buy a support contract", will work for this combination of product and field. It may be that there simply aren't enough users in my field to support an open source company in the way you describe.

So while I know it works for some - as you said - I believe for me and this product, a non-free/open source license would make me more money than a free license.

I have a few concrete problems I've come across in selling free sofware.

Given that I distribute a software product, how do I provide an evaluation version of the newer, better library? Do I simply trust that they will either pay me or not use it? If they do use it anyway, do I shrug my shoulders and walk away?

How do I do market segmentation? That is, if a non-profit research site, or academic group, wants to use my software, do I give them a discount? But these are also the groups where some graduate student might look at it, see that it's open source, and put it on their github account.

Which is fine. I support them doing that. But I need to price things appropriately. So it's odd that I may need to charge more for academic groups than I do pharmaceutical company (pharma being my main customer base), because academic groups are riskier sales than pharmas.

Now, I could simply give it away at no cost, and hope there is enough support contract interest for the future. But I get the strong sense that people will do a one-off purchase to get the software even if they aren't willing to get a support contract after they download and start using it.

[WalterBright]

My friend said that actually only a small percentage bought support. But he was happy about that, because the user base was so large that even a small percentage made for a lucrative business.

It's just a variation on give away the razor, sell the blades. Give away the music, sell the concert tickets. Open source enables reaching a much, much larger audience than would otherwise be possible, and at a much lower cost than doing advertising. (You also don't have to expend time and treasure dealing with piracy and the inevitable customer problems and anger with your DRM mechanism.) Another benefit is customers fix bugs and feed those back to you, so you've got a stronger product.

Another model for making money off of open source is using it as an advertisement delivery platform. We're all familiar with that stuff.

I also have a friend who expended a great deal of effort creating a very nice dev tool. He wanted to get paid for it, and made it closed source, pay only. An expensive marketing campaign is necessary to make that work (like what Coverity does), but he didn't have one. He made zero sales and was bitter about it.

I don't know your particular situation, and it might be worth a try for you to go pay-only. But be prepared to invest a lot in marketing.

As for the trust thing, my experience is that trusting customers works. Treat them as worthy of trust and respect, and they will respond likewise. Treat them with suspicion and DRM, and they'll live down to that expectation.

[dalke]

> "much larger audience"

Yes, that is the problem.

The biggest conference in my field has ~200 people. There might be 10,000 people around the world who might ever use my software in one way or another.

They tend to work in one of the one to two hundred or so companies doing drug discovery cheminformatics.

A "small percentage" of 200 is only a handful of companies. That's not enough to make a living. Rather, in that case it's more profitable to do custom software development in that case than off-the-shelf software. Because (oddly) people are willing to pay more for special one-off solutions than general solutions.

While if 10% of the field was willing to pay me once, and the handful of companies willing to pay support, then that's good money.

> "Another benefit is customers fix bugs and feed those back to you, so you've got a stronger product."

I've heard that a lot. I've also been doing free software in this field for 20 years. I've only ever received a handful of minor bug fixes.

My users are computational chemists and their IT support staff. They are not professional programmers. It's easier for them to pay me money to fix bugs than for them to do it.

Also, at some pharmas, code needs to be reviewed by the lawyers before being sent out.

I know one company that distributed source under the GPL, then found that their customers weren't willing to upgrade. Because they had made local changes, and found it too difficult to go through the lawyer review in order to send the changes upstream. It was easier for them to ask for new features instead.

> "advertising delivery platform"

Sure, and in-app sales. Except my software works on proprietary chemical compounds. I don't have access to the machines or the data, and no pharma in their right mind will allow me that sort of control of their in-house informatics platform.

> be prepared to invest a lot in marketing.

I wish to clarify that I am not looking to make my software proprietary. My argument is that I'm sure I could make more money with proprietary software than with free software.

Even if I make it proprietary, customers will get full source, no DRM, and the right to redistribute after (say), 10 years.

As for marketing, and as I found out, most people know of my software, and already use it. The problem is one of conversion less than marketing. Which is still a form of marketing, but less expensive.

> trusting customers works

So, one potential lead said they might buy my software. I gave them my list price. They said they didn't have the money to pay full price so could they get it for only $10K? This London-based organization has a full-time cheminformatics staff member, and over $100M in funding. I don't know if I should trust them, but I suspect it's a negotiating tactic.

Again, my point isn't to say that open source is a bad approach. I agree that there are business models where it does work. But it's not a panacea.

[WalterBright]

I would not argue that it works in all cases, my only point was that open source is very compatible with capitalism and lots of companies find it very profitable to go open and free source.

> I suspect it's a negotiating tactic.

Of course it is!

[dalke]

Certainly. So, what does "trust" mean?

Do I trust them to undercut the price I think is reasonable? Do I trust them to tell me what the software is really worth to them? Do I trust that they will value my time appropriately?

That's why I have a problem with the word "trust" in this context.

[WalterBright]

Trust means if you make a deal with them, they'll abide by the terms.

Trust does not mean they won't negotiate hard to get the best deal they can. Trust does not mean they are obliged to show you their cards. Trust does not mean they value your time as you would.

I suspect you'll be very disappointed if you consider the latter to be trust issues.

[open-source-ux]

Could your free version still be sold for a price but with limited or no support? For example, two licences

1. Commercial license, no support: pay $/£/€ amount

2. Commercial license with support: pay $/£/€ amount

By commerical use, I mean companies or academic insitutions. Option 1 is cheaper, but presumably more than affordable for companies who want to use the software.

I guess what I'm saying is that even when the software is open source, can you stipulate that payment is required if used by a commerical company or academic institution? Some companies may not pay, but I'm guessing many other companies will be happy to pay for option 1. Or am I being hopelessly naive?

[dalke]

While possible, that runs into another question: why would someone pay for support when they already have the software?

There are a few reasons: 1) they want to support new feature development, 2) they want to support me personally, or 3) they want support with understanding the API or doing integration.

I think most F/OS software has lousy documentation. I think that's because developers rarely like to write documentation, because free documentation writers are rare as hen's teeth, and because users aren't willing to pay for documentation.

Which is great if you want to sell your consulting services as well. That is, if the tool solves a problem, but is a hassle to use, then you get paid for configuration and customization. But that disincentivizes putting together a solid API with extensive documentation, where people don't need your help to understand how everything works.

> "even when the software is open source, can you stipulate that payment is required if used by a commerical company or academic institution?"

There are two issues. First, for the software to be open source, then there can be no legal mechanism to require later payment. What you propose is not "open source".

That said, it's might be possible to have a side agreement which does that. That agreement might not be legally enforceable, but a company will likely comply anyway, because they made the agreement in the first place.

But there's no way I'll obligate an academic group to pay me $20K should one of the grad students, or student with access to the shared disk, decide to redistribute it, after seeing that it's under an open source license. There's no way that will do me any long-term good.

[WalterBright]

If you stipulate that payment is required for commercial use, companies will pay. My experience is that commercial companies by and large are honest and quite willing to abide by the terms.

[thescribe]

This is a little off-topic but the software sounds interesting, do if you wouldn't mind I'd appreciate a link.

[WalterBright]

Send me an email and I'll forward it to him.