Posted: Fri Nov 11, 2005 4:37 am Post subject: I.T. Serivce Costing: Here's what we did...how about you?
I was wondering if anyone has had any experience in this, and is able to share their views.
I should mentioned early-on that the discussion below is about costing internal I.T. services. My SWAG approach would probably not work for providing I.T. services to "real" clients.
I've been assigned with the task of designating a service costing methodology for our company. At first I assumed that there was a lot of information for this on the web. I soon found out that it was not the case. The "pay for information" services didn't seem to have much either. Generally, it would end up being much to specific (cost metrics related to a specific type of software product for example).
The organization I'm presently a part of has viewed the I.T. department the same way for a very long time, as far as costs are concerned: Our department is just a big black hole into which they throw money. There is no measurement of how efficiently those dollars are being utilized. This being the case, you can probably imagine the difficulty of coming in right in "the middle of things" and starting up a costing initiative. In fact, I am told that there have been a number of failed initiatives to do the very same thing that I'm doing, over the past few years.
I endeavored to push on into unknown territory on my own, and my findings are as follows:
It's more or less up to you and your organization to agree upon exactly what things are going to feed your service costs. I think you can make the mistake of going too deep, thereby adding too much complexity. By having everyone agree upon what inputs to cost you are going to measure, everyone will at least be using the same assumptions. Later on, when the process gets more mature, you can re-evaluate what you are measuring.
One quote I got from Gartner that offered consolation was: "The one and only guarantee about a cost estimate is this: It's wrong." That helped me to realize that our objective here is not to come up with a "perfect" answer in I.T. service costing. I believe the most important thing is that each service is measured against the same cost components. At least that way, we can provide a realistic comparison to ourselves and our customers.
At this point, we've identified the following components of cost that are (relatively) easy to measure:
You may be noticing right off the bat two things that are missing:
In our specific environment, these two pieces - although measureable - are difficult to obtain. We determined that we should push ahead with what we can obtain immediately, in order to maintain momentum.
Following is a short blurb about each cost component:
Hardware cost (purchase)
I assumed that this would be one of the easiest things to obtain. Our organization has a separate I.T. Accounting group who orders all our hardware. However, I found out that these costs were more elusive than I thought for the following reasons:
When we would order stuff, it wasn't put into a nice clean format that an I.T. person can really understand. At first, you may assume that if we need a server of type "x" with capabilities "y", then we'd fill out one Purchase Order for it and our cost would be nice and neat right on the bottom of the PO.
The truth is that all these things are just widgets to I.T. Accounting. They neither know what items should be grouped together, nor do they care. Any given PO can include different items for a number of different systems, and in some cases, different projects.
So, next I came up with the idea that we could come up with a group of say, 5 standard server builds and apply a standard cost to them for simplicity's sake. This may in fact work in another organization (especially if they were to use a modern procurement system), but this wasn't the case. There is just too much complexity and not enough maturity in these processes right now.
So, what was the solution? Fortunately there was a parallel effort going on to better manage our capital assets on the raised floor. I was able to feed off this, and obtain a list of abotu 80 percent of our servers and their purchase costs.
The assumptions that I applied? Since we made use of a number of different channels to obtain hardware, i attempted to use the exact purchase cost of any given piece of equipment. If I couldn't do that, than I found others of the same type and capabilities and took an average. Lacking that, I would go to a vendor website and get the price there. (a lot of the major hardware vendors don't include costs on their sites, but their partners who distribute their stuff often do)
What about depreciation and such things? I made the decision that I don't care. I'm not trying to value capital assets. I'm trying to determine ongoing service costs.
Hardware cost (maintenance)
This one is a little easier to do. We had an executive here who had done some costing work elsewhere. He recommended that we use a figure of 14% of the purchase price, per year. Another factor to consider is inflation. Use whatever you think works for your locale. The standard assumption is that any system's lifespan is around 5 years, so cost out five years, and start applying 5% inflation after the first year.
Software cost (purchase)
The cost of the software, divided by 5 years. Beyond COTS stuff, this also includes Tivoli agents, virus scan, Operating systems, server products, etc. (some of the latter items may or may not fall into the realm of software licensing)
One challenging thing: how do you determine the cost of a piece of software developed in-house? Do the best you can. If you are able to determine the man-hours that it took to write, test, debug, and implement the software, you are probably OK.
Software cost (maintenance)
All things being equal, we use a figure of 18% to calculate the maintenance costs of any given piece of software. These costs are going to vary widely based on type and complexity of the solution. Once again, we do the best SWAG that we can at first to keep the momentum going.
One interesting point is that some vendor contracts will charge a flat “maintenance” fee for their software. This could replace your calculated software maintenance cost.
Software costs (licensing)
Most I.T. professionals should be familiar with the myriad ways that our vendors come up with to license software. It will be different every time, so it’s just a matter of talking to the person in your organization who manages the vendor relationship or looking at the contract / enterprise agreement.
Labor Costs (internal)
Our present method utilizes one standard figure for any associate who is involved. The idea is that anyone who costs a whole lot can be offset by someone who doesn’t. HR, or whoever does your project costing could be helpful here. I think an important thing to attempt is to get every group that does costing to use the same labor figure. That way you don’t get into apples vs. oranges comparisons later.
I’ve worked in other organizations in which there was a standard hourly rate for every job role and level. I don’t see a problem with that. Based on the job roles (defined by HR) that can potentially work on a project, you just apply the costs for that specific role and level. Sometimes this is going to be a political problem, because this information is guarded pretty closely. If so, just use a blended rate for everyone.
Labor Costs (Contract)
The political concerns really don’t apply for contractors, so you can usually obtain their exact rate that your organization is being charged. I recommend that you use that whenever possible, as it can be a statistically significant difference – either very low for offshore, or very high for an on-site contractor.
Basically I look at a couple of things: who is assigned specifically to support this service or app? How many of them are there on that team? What % of utilization do they apply to it (what % of their time in an 8-hour day or 40-hour week do they spend supporting the service or application). Using this you can come up with an easy formula:
(#FTEs) x (%Utilization) x (hourly rate) = hourly cost for this tier of support.
Another thing to consider is the service desk. How many tickets or requests do they handle related to this application or service, per day or per week? What is the average amount of time they spend on each ticket?
Obtaining access to underpinning contracts should be fairly easy. Try to see if there are any other additional costs in the contract that should be related to the service or hosted application.
All systems require data, and there is a cost associated with its storage. In our organization, we use local, NAS and SAN storage. Each of these has a different cost. SAN storage also has an additional one-time cost for the installation of a SAN card, to allow the server to interact with the SAN. NAS is considerably less expensive than SAN. We decided not concern ourselves with local storage, as that is part of the component cost of the server.
Part of storage costs can also be the costs required to backup data. Based on the criticality of the data, it could be stored redundantly in numerous locations.
That is presently all that we are measuring. We will eventually start incorporating the following into our service costing model as well:
This is a biggie, because everything uses the network. I can share my thoughts on how I think we are going to cost it:
(a small caveat here…I’m not, nor do I want to be, a network professional. If I refer to something incorrectly I hope my networking compatriots will forgive me)
My main goal is to come up with a number of different costs / MB transferred over our networks. This is probably the easiest metric to utilize. Once we know this, we can determine the number of transmissions that occur for any given service or application in a day, and the average size of those transmissions.
Different kinds of networks cost different amounts. I’m expecting to have to come up with a number of different costs based on things like physical architecture (fiber vs. some other type, etc.), location/region (different provides charge different amounts in different locations), and communication method. My understanding is that frame relay would have a different cost than say….well, some other method. I’ve yet to talk to our networking professionals, so I’m sure I’ll learn later.
The vast majority of our services and applications live in a data center somewhere, so it makes sense to apply facilities costs. The difficult part is, for a DC that hosts dozens of services or applications, it would be require obtaining a list of each one. Once that is accomplished, how do you apply the cost? Do you apply more facilities costs to larger applications, or do you just split it up evenly? There are a number of ways that you could probably apply variable costs: number of users, amount of raised floor space, amount of incident tickets, etc.
So those are our components of cost. We take all these figures, stir them up into an Excel spreadsheet, and there you have our cost template.
One thing to say about the cost template: The idea is that it’s simple enough so that the Service Owner or Service Provider can fill it out his- or herself. As you can expect, there is potentially a lot of complexity involved, so you can find yourself challenged to hide most of it in spreadsheet formulas and VBA (if you use Microsoft). I did my best to allow the Service Providers to enter simple information, having my spreadsheet do the complex stuff. This only works ok. It may be that if your organization is not too mature in I.T. Costing, you may want to shepherd your Service Owners in the process.
Another important issue from before is that you want to do the best that you can to ensure that the same things are being measured across the board. Really the only way to guarantee this is to have some QA or shepherding going on.
Generally, we use a different worksheet for each of the cost components listed above. The cover sheet gives just a rollup of the data. Presently, we show
-Implementation Cost (for repeatable services)
-Special breakdowns (for email for example, we thought it would be cool to show the cost per mailbox)
Implementation Cost is broken down by Hardware, Software, and Labor.
Ongoing maintenance is broken down by HW maintenance, SW maintenance, SW licensing, Data Storage, Operational Labor
I’ll point out that we are in the crawl stages of I.T. Costing. We made a lot of assumptions and SWAGs, based on the fact that this is an initiative that died on the vine numerous times before. One of our main objectives was to forge ahead quickly so as not to lose momentum.
As far as next steps are concerned, we are presently costing out all of our “utility” services – things like network, database, storage, server build, etc. The reason behind this is because every project and application uses them. Using the 80-20 rule, if we can cost out 80% of the utility-type of services, we can quickly apply costing to any given project or hosted application, and then spend our time working out the other 20%.
Another next step – and it will take a lot of effort – is to get the organization to capture baseline costing metrics throughout a year or so. Once that is done, we plan on coming back to our cost templates and comparing our “SWAG” assumptions with our baselines to see where we can tune things up.
Finally, as I mentioned, we are in the “I.T. is a big pit that you throw money into” phase of costing. We don’t do chargebacks or even notional costing yet. It would be good to move into notional costing first, so that we can show our leadership exactly what the financial implications of I.T. decisions are, and to help them make more informed decisions. Perhaps at the end of a year, we could give business units a notional invoice showing them how much their I.T. services cost the organization. This should raise a few eyebrows, and will once again lead to some more informed decision making.
Will we ever move forward into chargebacks? According to ITIL, it’s not necessary. We may or we may not.
As I.T. professionals, it’s our responsibility to help our business customers make decisions on the services we provide to them. By providing them with measurements of costs for the services they consume, we are better able to help support their vital business functions, which will lead to improvements in customer satisfaction.
I felt like I was charging into unproven territory in this initiative, and I wanted to share our experiences with you. Hopefully, it will allow you to spark some ideas when you find yourself in a similar situation.
In addition, if anyone has any opinions on this post, or can share there experiences as well, it would be greatly appreciated.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum