How Many Admins?

In responding to a question about planning your IT staff requirements in LinkedIn Answers today, I realized two things: 

1. This could make a good blog post.
2. I love talking about the mechanics of IT.

One of my favorite answers to this quandary is Mark Verber's "How Many Administrators Are Enough?" article from a 1991 edition of Unix Review magazine.  It's a great discussion of how to plan and control your IT staffing needs, but it's mostly about principles.  And the 24x7-site staffing model, which concludes that you need fourteen people may cause unneeded consternation if you're starting with an IT staff of, for instance, two.


One of the great challenges in IT management is keeping IT and non-IT folks in sync, and my impression is that most IT managers resort to very simple formulae, like "1 helpdesk staff per 50 employees".  I've found that this kind of formula breaks down very quickly as organizations and technologies change, so I've tried to use something slightly more involved.

I try to keep three goals in mind when coming up with a model to answer this question:

1. Make something IT folks can easily understand, critique and validate.
2. Make something non-IT folks can understand immediately.
3. Make a good approximation of reality.

Personally, I've used some estimates based on a few factors, and plugged them into a simple spreadsheet, to come up with a number of hours required to support a set of systems and users.*

This involves a simple classification approach for users and systems.  I used hours-per-month as a simple, easily explained number to assess the "how many admins" question.  Here are some very, very rough examples.

Non-Technical User - 4 hours/month (more assistance)
Technical User - 1 hour/month (less assistance)
Technical User - Developer - 2 hours/month (in the middle - has more interesting problems to solve)

Assign all the staff to these categories, multiply by x hours/month, and you get an end-user support workload.

Similarly, you can classify your systems based on complexity and rate of change:

High complexity, high rate of change = 40 hours/month (e.g. in-house email system)
High complexity, low rate of change = 10 hours/month (e.g. firewall)
Low complexity, high rate of change = 20 hours/month (e.g. out-sourced email system)
Low complexity, low rate of change = 2 hours/month (e.g. fax machine)

If you want a more complex and precise model, you can add in other factors in classifying your systems, like reliability, or replacement life cycle.  For instance, a system you replace once a year might have an annual transition project, which drives up the workload.  For every ~180 hours/month of work in the model, you would want to have an IT person to do that block of work.

Similarly, you may want to include projects in this model, so you could add another column, such as "Active Months".  Ongoing system support rows would have 12 active months, shorter-term projects would have fewer active months.

So, assign each system your techs own to a category, total your hours/month requirement, and you've got a system that meets the three requirements.  It's also very easy to show your work, including all of your understanding and assumptions about how IT time is being spent.  If you have a more complex environment, with many different IT staff functions (helpdesk, network infrastructure, security, storage, etc.), the model would need to be a little more complex.  For example, you might add a "specialty" or "skill" column, and assess your workload in hours/month/skill against the skills of different staff members.

I'd recommend doing no more than one year in a single set of calculations.  However, having following-year and previous-year estimates could be really useful planning tools, and can help keep IT and non-IT staff expectations aligned.

Happy planning!

*My social-media-branding advisors have suggested calling this the Sloane IT Staffing Model, but I'm not convinced.