Ok so you have an idea to launch a new product to help you reduce your service desk call. Maybe it is a Employee Password Reset System, a Knowledge Base, or even a tech support chat bot.
What are items due you use for calculating your ROI and proving to the bean counter your idea is worth the investment?
The basic math is how many man hours per month will this system/feature save the company. Multiple that times the average salary of the people who's time will be saved. That's the money saved per month, move that out per year and per three years.
Now figure out the cost to the company to create the system/feature. This is done by figuring out the number of hours it'll take to create the system/feature up front, then how many hours it'll take to maintain and upgrade the system every month or year. Take the number of hours times the average salary for the person doing the work.
Subtract the money spent creating the system/feature from the amount saved using the feature and you have your ROI. Be sure that the amounts that you use are based over the same time period.
Using an employee password reset system an as example. We'll say the helpdesk gets 20 calls a week to reset a password. It takes the help desk employee 10 minutes to verify the user and reset the password, plus the 10 minutes the user takes to find the help desks number, call and wait on hold. We'll say the help desk guy makes $15 per hour, and the user makes $10 per hour.
So by giving the user a way to go to another users computer and reset there password (working under this assumption we're talking about Windows passwords here) is costing the company $66 per week in user salary and $50 per week in help desk salary. So that's $6032 per year in cost.
Now, assume it'll take 30 hours to build and test the system, and basically nothing to maintain (since AD isn't changed all that often). The Developer makes $35 per hour, so that's $1050 to make the system plus we'll throw in $100 in salary to maintain annually. So over 1 year that $1150.
Since three year ROI is standard, over three years our costs are $1350 and our losses by not having the system in place is $18096. So over three years the company will save $16746 by building the system.
We use a very simple algorithm, with the following inputs:
Number of times operation is performed per day
(Old Time - New Time) * Frequency = Time Saved
Then it's just Time Saved * Hourly Rate = Money Saved
That gives the daily amount of $ saved. It could be a very very small number, don't distress if it is, because a single day is a very small time measurement in the business world.
Then once you know how much money it's saving your per day, it's not hard to extrapolate that to a yearly savings, and then subtract from that the implementation and support costs.
Edit: Actually, just refer to Mr Denny's response. He has vocalised this method in a much clearer way.