Comparing enterprise, self hosted pricing to SaaS pricing for a smaller client? Changing industry pricing models?

We're trying to figure out the pricing for a service we've been asked to provide (a couple times, in passing, actually in reference to something we've pitched that hasn't hit the mark "well that's great, but what we really are struggling with is xxxx").

One of our advisors is a director in a larger (much larger) company in the same industry and he's shared with us the amount they paid and are paying in initial license and maintenance for the defacto standard software (which is terrible) ($xxx,xxxx/year).

Is there a formula or some way to determine how to take that pricing, for a self hosted enterprise product and come to a reasonable SaaS pricing?

So far, we've sliced it two ways:

• take that number, divide it by the entities it covers (units in this case, not users/seats). Then use that formula for the SaaS. That gets us to about $40/unit/year. The smaller players in our market will have about 100 units. $4000/year -> $333/month = not enough to cover our costs.
• create a sliding scale that creates a reasonable, "buy it now without thought" pricing for the little guy and then if we get a very large client, work a special deal for them (they'll likely not want SaaS anyway, so that's a whole 'nother story). We've been throwing out $5/unit/month for $500/month for the smaller player that's a prospect. This is a more favorable number for us and begins to scale well financially for us.

Or should we break the industry model of pricing on the unit and go to a seat model (i.e. by the user, potentially more based on the user class as they would have different functionality)? The concern here is that a small (people wise) client with a large number of units could sign up and potentially overuse the system (comparatively).