Here at Apptio, we believe that benchmarking is a critical component of IT Financial Management – it provides two important reality checks:

  • First, to ensure an accurate allocation model.  For example, if you’re allocating Desktop Support costs back to Lines of Businesses based on Revenue (instead of something more appropriate, such as head or device count), you may end up charging effective “Cost Per Desktop” rates over $3k per month! We’ve seen it happen, but it’s easily fixed once you know about it, and benchmarking can help you find anomolies in the allocation model.
  • Second, to objectively measure the actual performance of your IT towers. E.g., your UNIX servers may indeed have world class availability (five 9’s), but do they also have world class cost efficiency? Or are you paying a significant premium for that availability?

That said, there are several types of benchmarks in the IT industry:

Unit Rates – These tell you what the average loaded cost of a unit of work is. For example, that a Linux server typically costs a large enterprise about $900 a month to host, fully burdened. The financial model that produces these numbers includes elements such as facility costs, system administrator labor and software costs (such as operating system, monitoring and automation licenses). When you get down to the ‘cost drivers’ of these numbers (how much of that $900 is system administration, how much is monitoring & automation licenses?), comparisons between your company and the benchmark can get very interesting indeed.

Key Performance Indicators – These metrics are often only indirectly related to cost efficiency. For example, we know that a Help Desk employee can typically handle between five and eight hundred calls per month, or that across the industry, you typically need one dedicated storage administrator for every 120 TB of storage you have. These metrics have to be interpreted carefully – deviation doesn’t necessarily mean performance is suboptimal - there may be something about your company that drastically reduces the number of calls a help desk professional can handle. If the deviation is significant, however, it should probably be explained.

Comparative spend – These metrics are most useful when specific to a particular industry and can give you high level assurances that your spend in a particular area isn’t disproportionate to your peers. For example, that 21% of IT spend in Telecommunication companies is on Hardware. If you’re a telco and are spending 30% on hardware, you can hypothesize that your software is less efficient or that you have above average spare capacity. These are clearly long term, high level observations about your IT department – not something that is immediately actionable.

The pundits among us also enjoy comparing entire industries using these metrics. For example, in the finance industry, 30% of IT staff is doing new application development. It doesn’t surprise me that higher education institutions average half that (15%), but I wouldn’t have thought that the transportation industry did relatively more custom development (34%) – although their overall IT spend is lower, so, unsurprisingly, they aren’t as technology dependent as the financials.

Every industry does benchmarking: the advantages of such a reality check should be obvious. For example, one of our customers discovered that his storage cost per gigabyte was over twice the industry average. An investigation discovered that some questionable capacity planning and… let’s say “aggressive” vendor sales practices had led to significant storage over-provisioning.

We see a lot of innovative IT financial management practices here at Apptio, and recently we’ve had a few customers begin to continually measure performance against benchmarks, as opposed to simply doing point-in-time checks. It isn’t realistic to expect all of IT’s unit rates to meet or exceed benchmarks, but improvement targets can still be set in MBO’s to measure the results of cost cutting initiatives. “How can we reduce our support cost per desktop?”

As automated IT financial management systems become more prevalent, benchmarks will become more accurate, detailed and understandable. In the future, I believe we’ll see a large increase in the use of benchmarks to manage IT.