November is DevPro's month to think about how to automate and scale service provision. Here are a half-dozen ideas to keep on top:

Trade-offs

Think in terms of costs and benefits. As tempting as it is to want to AUTOMATE ALL THE THINGS the real-world workplace is constrained. We only have time to tackle the highest priorities, and those priorities need to be the projects that pay off biggest. A lot of your value as an IT professional lies not just in the technical tasks you undertake, but your judgment about which ones are worthwhile.

One key: eliminate surprises. Any conversation that begins with someone in your organization reporting, "Our customers say that ..." and ends with you replying, "I had no idea", is something you want to prevent. A screen that shows, "Sorry, but we're working through high volume at the moment; check back in fifteen minutes" or even "Error #83--our technicians will call you ..." is tolerable (in most, but not all, industries). A blank screen or a Java stack trace should never happen, though. Figure out the worst and best IT things that can happen, prevent the former, and enable the latter.

Make routine change routine

Employees come and go. Equipment arrives, needs repair, and returns to service. Cables end up knotted and connected exotically. Treat these not as mysteries for heroic response, but as routine matters that you handle with complete assurance.

Not all automation must be turned over to the computer. While my instinct is to put everything online so I can just "push a button", sometimes the right first step is to construct a checklist for a human to read and execute. In any case, keep in mind all the different entities you need to manage expertly: hardware, software, users, licenses, certificates, domains, calendars, and so on, probably across a range of sites.

Resist special treatment

Once your routines are in good shape, they'll make such a strong impression on your supervisors that ... they'll ask for exceptions. Or you'll make the same request of yourself: you'll want to treat one special customer particularly well, and, hey, everything is on-line, so it's easy just to tweak one setting.

Resist that seductive appeal. Even in the best case, your organization will lose track of the exception and its circumstances, and you'll end up treating that special customer worse than your usual standards. You intended to extend the free trial period an extra two weeks, but a database stored procedure you didn't consider ends up crippling the trial. As strained and extreme as that example sounds, real-world experience teaches that unintentional degradation of service is nearly inevitable for "one-off" fixes.

The correct remedy for such solutions is to define a new routine: figure out an elevated class of service, co-ordinate it with Product and Support, and test it as you do everything else.

Good hygiene

Clean up your on-line environment. Take out the trash before filesystems fill up. Do your specialists have to scramble on a monthly (or worse!) basis to rotate disused logs or compact old databases? Capture that expertise in scheduled tasks. Consistent hygiene is a good job for computers.

Monitor well

IT's value to the larger organization is not just in what we do, but what we observe. Automation helps there, too: set up monitors so you know before your customers do when your Web sites slow, your databases stall, or your licenses expire. Minimize your surprises. Even when you have performance problems you can't yet solve--and you will--you'll be a step ahead to track accurately when they're happening.

Automate reports

How much of your day goes to answering questions about what you're doing, when a particular job will finish, or how to reset the three passwords that aren't connected to your SSO? Write an FAQ. It doesn't have to be static: you can teach your FAQ to pull the current data it needs. In the worst case, even if you're unable to persuade your co-workers to read the FAQ, you will have a standard location to find the answers you need quickly.

What automations have made a difference for you? Tell us in the comments below. We'll return later in the month with more tips on what to automate, and what isn't worth the effort.