Lots of the SaaS products we use mean we don't have to hire more employees. So I'm not sure if dollars per employee is a meaningful stat.
This is becoming a joke.
In my whole life I didnt buy a one single software product, that is based on subscription model (unless necessary as there are ongoing costs for its development like antiviruses (which I dont use, but this is another story)). Why? As it tells me, the product is at the end of its inovative phase and has nothing more to offer. I am fine with paying for update hunderds of dollars if, and only if, i need the added functionality. Or in different words Microsoft Office 97 was more than enough for 99% of its users (I dare you to install it and find something that you miss, it wasn't perfect but it did get the job done).
But there is more. I wont pay for cloud products either. Why? I believe that what is giving you the "edge" is your knowlidge. Using, for instance, gmail is taking you the chance to increase it. Yeah sure, time to market, but if you stick to DIY logic, this is not going to be a problem (I can set up a mail server with left hand while coding with right hand... and offer it as a SAAS to those that ignore DIY and earn extra bucks ;) /s ). If you ignore the lack of knowlidge, you will end up always paying to someone else or stuck in a problem that wouldn't be a problem if you wouldn't be ignoring it for so long. Not to even go into lack of inovation and reinventing / copying the solutions that were solved long time ago, but this is just another story.
As for the rest of your argument, most people are not like you, as they will not want to spend time setting up their own mail server. For companies this is doubly true, as they will likely spend more money setting it up, maintaining it, and upgrading it, than the subscription actually costs.
I wonder if those same companies intend to socialize the wealth generated by their automation.
I don't like it at all.
Running the mail server is simple enough, but then you need to set up storage. Emails are sensitive and often can't afford to lose. Then you need to set up backs, monitor and regularly test DR scenarios to see if your failovers, recovery plans are operational. Of course you need to maintain your data centers as well if you don't want subscriptions.
Then think of login. Employees don't want to log in to 15 different software. So they are going to ask for SSO. You are going to have to deploy and maintain something like active directory. Sometimes they need to be maintained across networks between different offices that are far from each other. After all this, you will still have to buy support packages from principle vendors for times when things go wrong (and they go wrong all the time when you host your own stuff). Shit escalates mate.
Login is surely one of the best arguments for non-SaaS solutions. SSO is a total mess in the SaaS world. Everyone ends up with a billion passwords that they manage individually, change at different times, have to use a password manager to keep track of (but they all use different ones), etc. Active Directory is actually a solution to that which got lost in the move to SaaS-in-the-cloud.
If you have 10,000 employees that's $100M per year in expenses. If you're just renting payroll, purchasing, etc software for that price it'd be far cheaper to roll your own, or pay a few people to work on OSS full time.
This is the kind of hubris I come to this site for.
https://en.wikipedia.org/wiki/Chrysler_Comprehensive_Compens...
Payroll is harder than it looks due to all the edge cases. For example, it is literally possible for an employee to have so many payroll deductions including court-ordered wage garnishment that net pay would be negative. How do you handle that in all legal jurisdictions where the company operates?
By issuing a error and demanding that a human deal with it. Obviously.
If you mean it's impractical and error-prone to do payroll manually, well that's the point of automating >99.999% of it. For the remaining <0.001%, someone is going to have decide how to deal with it, and I'm sceptical that a programmer seven years ago will know better than a lawyer today what the legal requirements today are.
You could argue that payroll shouldn't be so complex but that's the reality of it today.
Second month was a disaster in the end they had to physically cut the tape on the last remaining partially destroyed copy - about 20% got paid on time.