You get what you pay for.
But not often. It’s what mathematicians call an upper bound, and writers call a cliché.
Our topic this week: CIOs running IT as a business. That’s the cliché part, and I suppose we should support the attempt to de-cliché it with a new buzz-phrase: “IT as a Service” (ITaaS).
Now for the mathematics part — how IT should price its services and charge its internal customers.
Chargebacks.
For ITaaS to work your service catalog will have to be a catalog: It will need prices. Which means you have to understand something you should understand anyway: What IT’s products and services cost, and why.
Drop IT’s costs into three broad buckets: (1) overhead costs (the cost of turning on the lights); (2) per-capita costs (those driven by the number of employees who need support); and (3) metered costs (costs driven by utilization of one or more resources).
Everything else is detail — irritating and tedious, so borrow a financial analyst from Accounting to help you. That’s what they’re for.
Now you’re ready for chargebacks.
Overhead
With exceptions (departments whose employees make little or no use of information technology) allocate IT’s overhead costs to cost centers in proportion to each cost center’s budget. This isn’t perfectly fair, but it’s supportable and simple. The cost of the physical data center goes here. Enterprise systems too: Even if a department doesn’t make direct use of the company’s ERP system, for example, it makes indirect use, and piece-mealing per-module costs to “owners” throughout the company is a pointless headache. Worse, it encourages each department to substitute a cheaper alternative, increasing the cost of integration … and who’s going to own that?
A popular alternative is to take overhead costs as a percent of the total IT budget and add that percentage to per capita and metered costs. This approach has only one gotcha, but it’s a big one: Forecast volumes wrong and you’ll over or undershoot your spending, possibly by a lot.
Better to keep fixed costs fixed.
Per capita costs
These are relatively easy. Fully provisioned PCs, for example, have a known cost of purchase and require a set amount of labor for acquisition, configuration, quality assurance, and dropping them onto an employee’s desk.
Metered costs — non-labor
Metering the costs of physical resources is easy, assuming you have the proper tools for data collection. Storage, for example, costs what it costs (if you capitalize it, use the depreciation amount rather than the purchase price). Tally up who uses how much and you’re done.
Don’t get too granular, though. In principle, for example, the cost of the electricity needed to run the equipment in the data center should be metered — the more computing a department does, the more electricity it uses. In practice the difference between the metered number and the number you’ll arrive at by lumping it into the data center’s overall cost allocation isn’t worth the aggravation.
Metered costs – labor
Where chargebacks get ugly is when the resources are human.
What to charge isn’t hard. Whether it’s the number of physical or virtual servers a sysadmin can support, how many incidents of a given degree of difficulty a service desk analyst can handle in a day, or how much business analyst and programmer time it takes to develop a module of a given degree of difficulty, all you need are accurate timesheets. (Yeah, I know …)
IT can tally time and report it, but demonstrating to anyone’s satisfaction that this is how much time and effort the job should have taken?
That’s where ugly happens.
Some CIOs defend their numbers with industry benchmarks, which are fine except that they’re worthless. Worthless?
Imagine, for example, your programmers take longer than the benchmark says they should. That’s a problem, unless your programmers produce cleaner code than average, at which point it isn’t. Or, maybe your business users change their minds more than those in “average” companies. Think anyone benchmarks that? Think again.
This is an unsolved and probably unsolvable problem. My advice: Use last year’s numbers as your benchmark and make improvement your goal.
Preparing the invoices
The critical balancing act is dividing costs into fine enough categories that IT’s invoices are meaningful, without so much precision that the invoices are too complicated to understand.
Because if a manager can’t understand the invoice, the result will be an argument and a damaged relationship, and won’t be a business manager who understands how to manage IT costs better.
That’s a cure that’s worse than the disease.
Bob,
Really useful column for those IT managers who haven’t had to delve too deeply into business before – thank you. Just one question – the second to last line seems to be missing a word (or perhaps several.) Could you please update the text.
Thanks for the columns and your books, all have been helpful to me.
Nope – they’re complete as is, although possibly a bit cryptic. My point is that if a business manager can’t understand IT’s invoice, the result will be an argument over whether the invoice is correct and fair. That will damage the relationship between IT and that business manager. A second consequence will be that one of the most commonly stated reasons for charging back costs – that it makes business managers accountable for managing their IT spend, can’t happen because if the invoice is incomprehensible, it won’t help the business manager understand how his/her decisions affect IT costs.
Hope that helps.
The section on preparing the invoices needs to be forwarded to all practitioners of medical billing. They are the epitome of too granular and complicated to understand.