b) If you're in email, and you're not working with clients at the very, very top of the sophistication chain, you should not expect them to host their own email servers because their deliverability will be terrible. I suppose you could theoretically let them use your product with a bring-your-own-MSA, but that should probably be in-scope for your product. (You'd also be crazy to do this yourself, but you can use an MSA on the backend, like substantially every company providing UI and logic on top of email does.)
c) The operational difficulties of doing release cycles and maintenance for client-hosted software, particularly hosted software which has to play well with client-provided infrastructure, strongly, strongly, strongly suggest you host things yourself.
1. Yes for email delivery they can always bring their own smtp like sendgrid or mandril, in that case may be they may not have to think about delivery? 2. Don’t you think customers can save money on self hosted if we sell for onetime payment (let say price at 6 or 8 months subscription cost) like whmcs. 3. What do you think about customer acquisition cost for both options will it be equal or any one will have any advantage over other?
As to your original question: Unless your software is special in some way that you don't want anyone to find out about, you can offer both SaaS and self-hosting at the same time. My advice would be to do that, but use a yearly-license model for the self-hosting. Calculate the costs a self-hosted version would cost in total to a client (license, hardware, maintenance etc.), and make the SaaS version slightly cheaper than that.
The recurring subscription model fits nicely with saas software also and your growth over time starts to compound.
DON'T.
SAAS is a great deal easier - it's easier on you support-wise and it's also much easier to sell to most clients. I'd expect your cost of customer acquisition to be lower by far with the SAAS version.
Patio11's point about deliverability is also very important given your sector.