Ask HN: Should we offer selfhosted or cloud version first of B2B SaaS?

What does your first paying customer want? Start from there, is my advice.

Also, better figure out early on whose job will be to get the product out (dev) and the customers in (sales).

You should ask your customers/prospects which they'd prefer as it depends a lot on your specific product and how you are targeting it. But in general I'd recommend starting with cloud because the companies that will take a chance on a new service will tend to be smaller and prefer cloud.

With self-hosting, you also need to be sure your contract size is large enough to justify the time you'll have to put into support, which can be significant.

A good model for starting out imo is to offer cloud, but have a contact us option for self-hosting. Do self-hosting once you have a customer or two who wants it and is serious enough to pay for a contract that will cover support costs.

Dev tools are hard to grow/price. Most of them never make money. Those that do make money from a very small % of their customers.

What will your self hosted version look like? Both should be containerized and so apart from managing keys you shouldn't have a ton of overhead.

How about "hosted" in isolation using a server(s) which are guaranteed to be single occupants?

You retain control over the software and they have assurances the hardware is not shared

I started with SaaS and only recently after over 8 years moved to offering self-hosting as well. I tried code escrow for awhile, but that was hard and required a lot of hands-on support because at the time the app wasn't built with self-hosting in mind. I'd say go SaaS and offer self-hosting once the pain of not offering it outweighs the cost of offering it, because there is a cost to offering it -- from support, to onboarding, to setup flows, etc. If your customers want it, build it, but if they don't, you save a lot of time and money until they do.

You could do an AWS marketplace kind of thing where people could spin up your configured server in their own cloud.

Long term vision should support both but in short term you can focus on your hypothetical ICP or beta customers. If you are focused on customers who prefer self hosting, that could be your MVP version but in the long term cloud may give you scale. Some upfront Pros and Cons of Self hosted vs Cloud. Self Hosted ( As assuming stand alone or cloud self hosted) - Pros: Cost to serve customer is less for you as you pass on infra cost to the customer. - Cons : Maintenance and continuous support will cost you more as every deployment release and support will be specific to customer.

Cloud - Pros : Easier to get started and scale if it is a self serve product. - Cost : Infra cost is added to your cost to serve customer.

I tried both, and I'm currently focused on self-hosting.

Selling a hosted version has usually less friction (customer has less to do), but it also requires a lot more resources from your side (both time-wise to run/maintain the service, but it also costs you money to run it).

With my self-hosted analytics platform I decided to go self-hosted only, so the main focus now is making the installation and distribution as easy as possible. One way to do it, is to provide images to hosting providers and their marketplace, for example what I did with my DigitalOcean app: https://marketplace.digitalocean.com/apps/uxwizz

In this way, the customers can easily try the app, and install it.

The main drawback of self-hosting, currently, is that there are A LOT fewer people in your target audience. I think this will grow in the future, but at the moment, hosted services are still the easier sell.

How do you do licensing and entitlements enforcement in the self-hosting B2B SaaS space?

Where to see the tool?

I've been thinking about this lately my self. At this time I don't have much interest in running a hosted SaaS business.

I anticipate my target customer will be also be cloud customers. So I have decided I'll provide them terraform code so they can deploy the infrastructure needed to run the app. I had wanted to move my dev environment from ClickOps to IAC anyway, so that has been a good motivation to do that.

Once the dust settles from the Broadcom buyout of VMware I'll probably look at create the IAC for an on-prem version too.

This might not work for you, hell it might not even work for me, but I'm going to give it a try.