For IT Cost Recovery, When Is Good Better Than Perfect?

I suspect most people working in IT and Finance are at least a little bit OCD.

I’ve worked across both these disciplines for nearly twenty years and certainly exhibit some of these characteristics from time to time. I could argue these are two professions where the details really matter. Things are either correct or they’re not. So I should feel justified demanding a higher attention to detail both from myself and from those around me, right?

But I’ve also learned it’s possible to make a lot of progress with a “pretty good” level of detail as we make our way closer to “perfect.” As uncomfortable as it may make some of us, there are times when it’s better to keep moving forward than to wait for the ideal set of information. This axiom has proved particularly true in the adoption of shared IT services models and cost recovery implementations I’m hearing about from CIOs.

The sharing (IT services) economy

For those just getting to the shared services party, the fundamental concept is offering IT as a collection of services to internal constituencies. As central IT, you create a service catalog, set appropriate rates, and enable some form of cost recovery, either through a showback or chargeback model.

This IT strategy was first implemented in the 80’s and has been gaining significant traction in recent years, particularly among larger private sector organizations. Recently, public sector entities have also realized the potential benefits as budgets tighten and accountability increases. For example, NASA was an early adopter of this approach and has already saved taxpayers in excess of $200M.

While a lot certainly depends on the implementation, when you’re dealing with IT budgets of hundreds of millions of dollars or more, the potential cost savings of 15% or more can be very compelling. In addition to the cost savings, IT is also able to shape demand for various services by changing the price they charge. Want to slow the growth of your expensive, premium SAN environment? Charge a bit more for that service and also offer a less expensive option based on commodity hardware for non-mission critical applications.

But my data isn’t good enough

A primary obstacle faced by many organizations moving to shared services is the concern about data quality. They don’t feel they’re in a position to send out a monthly bill for IT services that is defensible at a low enough level. Is the rate appropriate? Is the usage volume right? Is the team even seeing a full accounting of all IT spend by each department?

Here is where I’ve seen the most successful organizations push forward in spite of a lack of perfect data: they use their early implementation as a tool to drive a more targeted conversation, which in turn, improves their data quality. This will usually happen during a showback phase of their cost recovery implementation before they are formally recouping costs in a full chargeback model. They simply present the charges for what they believe to true and this invariably starts the conversation about how the business owner believes accuracy in the numbers could be improved.

Case in point: A Fortune 500 retailer we work with has been in the process of mapping all IT spend to VP owners. When they started, most of the spend was put into a category titled “Other VP” because they couldn’t identify who to allocate it to. In an effort to take that category to zero, as well as the desire to drive behavior, Central IT began a showback program to all of the VPs, sharing spend in a weekly cadence. Because of this rigor and the conversations which ensued, they’re now very close to eliminating the “Other VP” category entirely.

Good, better, best in cost recovery

Once they understand you don’t need perfect data, the next question is how good does it have to be to get started? Recent conversations with IT leaders have provided some valuable context for evaluating cost recovery across a maturity model of good, better, and best.

A primary decision about how you’re going to allocate actual expenses throughout the organization is key. This is both about translating the things you transact (e.g. telecom, storage, headcount, etc.) into the applications and services that power your organization and determining how much each division consumes of those applications and services.

»Resource: See the TBM Council’s industry standard allocation model, ATUM® (Apptio TBM Unified Model®)

Here are a few incrementally sophisticated approaches to implementation:

  • Good: Use a best guess at either actual costs or budget (whichever is more readily available) and share a bill based purely on headcount for a given division. This will not provide any direct allocation to consumption of services for the division, but will certainly get the ball rolling.
  • Better: Take the cost/budget estimations from the good scenario and map them to actual service usage by the divisions/business units. This can be taken a step further by transitioning from a showback model to a true chargeback model, where the division is financially responsible for all of the expenses. With this model, keep in mind your constituencies won’t like being surprised by big true-ups at the end of the month, particularly if they’re not defensible. So make sure you have a good handle on costs before moving to this stage.
  • Best: Introduce the ability to set rates based on desired adoption of the service. This requires knowledge of your actual costs and a baseline plan for consumption of that service, as well as a strategy for shaping consumption of various services in your organization.

Know your competition

One piece of advice that is always relevant when implementing a cost recovery process is to be fully aware of the price / ease-of-procurement of any of these services from a public cloud provider.

In this day and age, it has become very easy for business units to quickly procure IT services from 3rd party vendors. After all, frustration with a slow central IT was one of the original catalysts for AWS and Azure. Don’t think that just because you are transparent about the price for a virtual machine (VM) that your customer will be happy to pay that much for it. Challenge yourself to offer a compelling value proposition, which need not be based solely on price. In fact, your fully burdened VM cost will likely not be competitive with the big cloud providers, so make sure you’re clear on the other benefits you can provide.

Why wait?

Having helped organizations solve challenges like this for over a decade, we’ve seen firsthand the positive outcomes that can result from jumping in feet first. Unilever’s shared services approach is a great example of this. When we started working with them, only about 30% of IT infrastructure costs were traceable back to the business unit that consumed them. Using TBM as the foundation of their efforts, Unilever has managed some very positive changes including:

  • A 30% increase in server-to-services relationship
  • Mapping of over 95% of servers to their respective service or application
  • Total cost of ownership (TCO) visibility into more than 70% of their services

If you look back to 2015, almost 80% of our IT cost was, frankly, a black box. We are now in a position to deliver a bill to all countries for the consumption of the services and to work with the business in relation to the value IT delivers.


Gopalan Natarajan

CFO, Global Business Services


There really has never been a better time to be involved in IT. The notion of digital transformation is real and means organizations are leveraging technology to reinvent the fundamental way they execute on their mandate. Shared services are a great example of efficiencies that can be gained as part of this process.

The opportunity for real change in IT service delivery is certainly there, but only once you’re comfortable with the notion that sometimes good is better than perfect.

Article Contents



Additional Resources