CFO: You want to spend HOW MUCH?!

CFO: You want to spend HOW MUCH?!

If you’ve ever tried to build a business case for an IT solution, one that you have to present and justify to the top brass including the CFO, then let me put a comforting arm around you and say, “There there. Would you like a nice cup of tea?  It isn’t easy for anyone…”.

Gartner popularized the Total Cost of Ownership (TCO) model for IT back in the 80′s but the concept is about 100 years old.  Maybe a TCO model is difficult because it is comprehensive – it must include all costs, meaning capital and operational expenditure – for the life of one solution.  But an IT organization runs many IT solutions all at a different stage of their lifecycle, so there’s another complication.  And don’t forget that these different solutions are managed by the same, centralized groups of resources… so how on earth does someone work out the OpEx for an individual solution (e.g. virtualization)?

To complicate matters further, there are other models that can go into an IT solution’s business case, namely

The difficult thing about all five models is how to represent Operational Expenditure (OpEx) for one solution in today’s monolithic, complex and beastly enterprise IT organization.  If your solution uses a Linux platform then how do you calculate how much OpEx will represent the efforts of the centralized, shared Linux team on your individual project? Unless you can work this out, then your TCO is missing a number.

If you are thinking Work Packages then I think you are facing in the right direction, but without proper Work Packages with level of effort and costs in them (FTE hours x labour rates) then this is still a long way off from giving you your missing OpEx number for the Linux team as part of your solution.

Why is OpEx important? Well, depending on who’s numbers you believe, anything between 50%-80% of an IT budget can be spent on OpEx (as opposed to CapEx).  Horribly, up to 80% of OpEx can be spent on Unplanned Work.  On a $100m IT budget, that’s anything between $40m-$64mthat is not delivering value! (see end of post for some references)

The crazy thing is when you try to do a TCO model in IT you always struggle to get a handle on the OpEx numbers, especially as they apply to one solution picked out of a whole, centralized, group-wide IT budget.  These numbers are often hidden, obscured, obfuscated, made up or just plain not available.  How do you know the OpEx cost of the “development lab”, and how that might be improved by the introduction of VMware LabManager automation.   You know that OpEx will be better, but how do you prove it to the CFO?

It makes things difficult when you’re trying to show, in your own IT organization’s words  (well, numbers), the value of a shared service like a Unified Computing solution by looking at the cost envelope of the people, processes and technology to be deployed and operated COMPARED TO the existing Business As Usual or A. Competitor solution.

In fact, some people out there like Steve Kaplan makes a good distinction about the uses of TCO versus ROI.  In Steve’s view, which I share, TCO is great for comparing the costs of two product options; ROI is great at comparing the values of two platform solutions.  There is a challenge with using ROI as a comparative model is that IT today can’t quantify it’s value:  just how does IT measure the financial return of their $10m spent on a new change management system?  Perhaps it can be measured in a reduction in OpEx?  But we are back at the “how to get OpEx numbers out of IT”.

How about Payback?  This is a measure of when you recoup your outgoings: but unless IT is using a charging mechanism, like a managed service provider, then the normal practice is for IT to instead just get its costs covered at the end of the year.  So payback is guaranteed by the business within twelve months? But the business won’t just foot any old IT invoices, they will want to know the numbers.  Back at OpEx again.

NPV and IRR are used to work out which investment is best.  Like the old “what if I just put my money in the bank and collect the 6% interest in 12 months” rather than spend it on IT: unless IT can provide hard numbers then this comparison is hard to do, and back we come to OpEx numbers.

OpEx numbers are really important so how do you dervie them from the “IT Total OpEx” into something realistic for your platform solution?  Here’s a list of techniques to help identify that pesky OpEx:

  1. Be clear on what’s CapEx vs. OpEx. An example of potential confusion is the “Services and Subscription” part of a purchase.  Is software CapEx?  Is the “SnS” OpEx?  How do you do this for other solutions, follow your existing methodology if that makes sense.  Obvious OpEx is Full Time Employee (FTE) costs.  Although you might want to do a Theory of Constraints model and only put FTE in OpEx if it is a scalable resource directly related to throughput?
  2. Calculate the cost of Work Packages. These are the operations equivalent of a “Unit of Compute”.  With compute you divide total hardware/software resources into units, like a standard server of 500MHz CPU and 256GB RAM, and break down the costs by the same factor.  In the Work Package format, each Work Package is assigned to FTEs and a level of effort per FTE should be estimated.  If you know the labour rate for that FTE, and you know how many Work Packages that FTE has on your solution, you can estimate the OpEx.
  3. Allocate a portion of existing OpEx to the new solution. If, through planning work, you decide that the new solution will add 5% workload to existing IT Ops, Administration and Service Management teams, then what is 5% of their combined OpEx?
  4. Identify additional OpEx. This is the obvious OpEx, like new maintenance contracts, additional staff.  Note that when it comes to comparing these numbers to “business as usual”, there is normally additional OpEx incurred for that – costs don’t just stay static because you aren’t changing things.

It is possible to work out the OpEx for a solution, but it isn’t easy.  Even more troubling, everyone seems to work it out using a different method, both internally and externally.  That makes it hard for you to compare the cost of your solution against another inside your organization, and hard to compare yourself against another organization in your business vertical.  How do you work out OpEx in your organization?

All of this, so far, pertains to cost and not necessarily value.  Just because something is cheaper, doesn’t mean it is better – right?  In fact, Gartner has recently talked about IT Intensity that shows the negative impact that pure focus on IT costs can have, especially in a downturn.  The thinking goes that if your CFO treats IT as a cost-burden that sucks 20% of revenue, no matter what, then in bad times when revenue plummits then so does the cash available to IT.

This can have appalling effects on the business as purchases are stopped, contracts and staffing are shrunk, and all at a time when the business needs IT to make it as competitive as ever.

References

Check out these great resources:

Related posts:

  1. How real life became my worst nightmare