I did this last year. Just figure out how much it will cost you to package things up, write out explicitly what will be your responsibility (maybe bugfixes, updates, whatever) and what will be theirs (keeping the server running, backups, etc).

Expect to spend three or four times longer than you think you should dealing with their IT department.

Make sure you charge a healthy amount for all this because it will take more resources than you think on your side.

This is solid advice. I can't emphasize how important it will be to clearly define ownership.

One of the main initiatives of my current role is to carve out responsibilities between a company and its infrastructure support partner, it tends to not be a pretty process. I recommend you land grab whats required to properly maintain the app otherwise you'll be reactively accountable for adapting to a changing environment as they update OS, antivirus, etc. The more you can black box a particular VM from the rest of their IT organization, the less headache you'll encounter.

This is good advice if you are going to do it.

I'd argue though that the best option is to offer them a dedicated instance of your app and secure access via VPN (as someone else already suggested). Having the app installed within their datacenter will likely require you to have two different deployment models, one for your standard usage and one for their datacenter. It may even require you to make custom code modifications to work around limitations or security settings that do on their "standard" OS images. So once you go into their datacenter, you need to provide prepackaged deployments that have a configuration tool etc. Not saying any of this is insurmountable but you have to make sure you properly charge and expect it to take far longer than you anticipate as has been said.

This is pretty common. I field a half dozen requests like this a year for Twiddla. Very few of them pan out to actual deals.

A good plan to avoid wasting your time is to give them a ballpark figure up front for what it will cost to make this happen. That figure should not be less than six figures if it involves any significant effort on your part.

If you've already packaged everything up, and it really is as simple as delivering a VM image, you might only quote them 20X - 100X the retail price of your most expensive subscription plan. Naturally, this will be an annual contract, with the first year paid in advance and a fair amount of lead time.

Keep in mind that your email outlining this will typically be the last contact between you and the customer, since they're often simply middle manager types with an expectation that the price will be the same as your "Big Customer" subscription level, except that it will somehow magically run behind their firewall. Your mail will simultaneously give them a heart attack and correct their understanding of Enterprise licensing. If they do engage after that, you can start your Enterprise Sales cycle.

Good luck.

When my startup launched in the B2B finance space a few years ago, we were SaaS - but faced the same dilemma. Our first customer wanted on-site for security reasons, and we needed the money, so said yes.

A week or two after signing the contract, they said they couldn't get the hardware after all and could we go hosted instead. I look back at that moment, and thank my lucky stars. Now 2 1/2 years later we would never offer on-site as an option.

Be sure of what you're getting in to. Will they allow automated deployments with updates to your appliance (as that's essentially what it is). Will you need permission to access it? How will you support users accessing it? Often those same 'security' reasons apply. And then you'll find yourself stuck in the old model of yearly release cycles, supporting old versions and all kinds of pain.

Think long and hard and be 150% sure it is worth it.

You can say YES, and:

- we'll start with a 2-week discovery phase, at $225/hour, in which we work with you to see how all the details would work etc.

- once implemented, we charge an annual license of $XX,XXX, and we're available for consulting for $225/hour.

Hey,

Do they consider having it hosted on a private cloud environment? So you guys still control the systems while they can access it securely (i.e. VPN tunnel between their premises and your cloud servers).

The difference with your current offering is that this "instance" of your APP won't be shared with anyone else and you can land a nice maintenance contract, plus keep everything under your control.

I think the reason why the company wants it hosted in-house is primarily so the data never leaves their own hardware. Even if a completely new and private instance was brought up and hosted somewhere external of the company, it would still be hosting some of their data which could compromise business operations or infrastructure security should any of the data be ops critical.

> implications like updates, new features

They might be understanding if you told them there would not be updates or new features aside from critical security patches. When they're desperate for new features 10 years from now, dispatch a team of on-site consultants to help them upgrade.

I would probably consider it, but put a really premium price and make clear that they will get substantial discount if they choose SaaS version.

Can you build and a deliver a box in a U form factor that you can setup a service contract around?

Yes, go for it. Every other SaaS has an on-premise license - github, for example.

Once you have a battle-hardened product with your key features ironed out - absolutely. I'm pretty sure GitHub didn't launch with an MVP on-premise product, however. Obviously do-able but you can't help but increase friction in the early days of building out your product if you're following SaaS for your other customers.