"we focus on the "? You left something out.

A better approach to figuring out how to charge for this is:

"what workflow do I anticipate my likely customers will have, and how would this product integrate into that workflow?"

You should also look at what other products are in this space, if any, and see how they price themselves.

All that said:

I'm not sure selling this as a service is the right approach.

You could set yourself up as a specialized consultancy (give us your tired old schemas, we give you a shiny new data-access layer using your choice of the current best-of-breed technologies), and just keep your code generator as your secret sauce.

Then you market your consultancy to various outsourcing shops, offering to take the initial persistence layer coding off of their hands for their new projects. You can turn around better quality code for this stuff, faster, than any of the sweatshop operations, and you're cost-competitive with the sweatshops, too.

I know the dream is to set up a service, let it run, then sit back and collect the checks, but the consultancy route might be an easier route to monetizing this.

Otherwise, I would attempt to negotiate big-dollar site licenses with large outsourcing firms, who probably stand to benefit the most from this technology (b/c they are the ones who most often are given a legacy database and asked to build systems on top of it).

If you go that route, being available over the web isn't much of a selling point.

For a trial versions, making an unlimited use trial for N months (N = 1, 2, or 3) available after a phone call works:

* the phone call step slows things down, but makes it much more difficult to "farm" an infinite # of free trials (!)

* since the primary value-add for this service is at the start of a new project, and since the typical large outsourcing firm has new projects starting all the time, a short-ish trial doesn't run much risk of giving the customer everything they'd want from the full product (b/c it only helps them with projects that were getting started in those months)

(!) If you go this route, don't hard-code a time limit, but do set up some automated emails to remind you when a trial is coming close to wrapping up. It's helpful as a sales thing to be able to extend the trial a week or two (or, at least, it's terrible from a customer's standpoint if the trial cuts off hard right in the middle of something).