We call it the “process bypass process.” Hold that thought.

In a healthy organization, IT’s business partners have a lot of ingenuity, much of which is focused on what IT can do to make their work more efficient, productive, and effective.

Far too much ingenuity, in fact, for IT to be able to take care of it all. The result is the ever-present “IT backlog” — the list of projects IT won’t be able to get to for the foreseeable future. IT has struggled to find solutions to the backlog for as long as I’ve been in this business. The problem is as pernicious as ever.

There’s no single solution — you have to nibble away at it. Here are some techniques you can apply to deal with it and, if not solve it altogether, at least make it manageable.

• Apply the rule of ten. Any organization, at any level, can deal with between three and five initiatives of any size in a given year. Any more than that and the organization’s energy is divided too thinly for anything to get done. That being the case, whenever anyone suggests a great idea (as in, “You know what this organization needs to do? It should …”) ask this question: “Great idea. And we have a lot of great ideas in front of us for what this organization needs to do. Does this make the top ten list? Because if it isn’t one of the ten most important, we’ll never get to it.”

• Insist on a business sponsor. No, not just someone willing to testify that “we need it.” We’re talking about someone who has the authority to commit to the business benefits to be had from whatever the request is for and live up to delivering them once the project is finished. That also means defining all projects in terms of business change, not software delivery, and including implementation of the business change within the project plan. No benefit, no business change, no project.

• Use a magic quadrant. Yeah, I know, but use one anyway. The x-axis is total benefit — amount of improvement combined with organizational scope. The y-axis is total investment — the combination of cost, effort and risk. These don’t have to be precise — consensus scales of 1 to 10 will do just fine. By throwing each request into the magic quadrant you can do some initial screening: Reject everything (of course!) that combines high investment with low benefit. Accept everything that combines low investment with high benefit. So far, so good.Low investment and benefit? Either make it a local problem, to be solved with Excel or MS Access, or find a way to fold it into another, larger effort.

  The high investment/high-benefit proposals are the toughest. These are the ones you carefully screen and analyze. Choose only as many as you can successfully implement, and make sure they get a lot of oversight. They’re the most likely to get out of hand, and the most likely to be disappointing once completed.

• Don’t say yes or no. Say no or when. If you can’t provide a project launch date, you might have said yes, but you didn’t mean it because you don’t know if, let alone when, you’re going to work on it. Don’t pretend. You haven’t said yes until you’ve put the project on the master schedule, with the launch date predicated on the availability of all staff required for the project to proceed.So now you have an governance process. Be careful what you ask for. Another name for well-defined governance is bureaucracy — the loss of flexibility and agility that’s the result of slow, consensus, committee-driven decision-making.

Which is why, as part of your IT governance process, you need a process bypass process — a governance process that bypasses your governance process. It’s a way to fast-track opportunities that just won’t wait. The keys to success for a process bypass process are (1) it’s only used for time-sensitive opportunities, not to placate the loudest voices; (2) it’s used to give the project the next available time-slot in the master schedule, not to interrupt work in progress; and (3) process bypasses still require approval — most likely by the CEO, CFO and CIO. And that includes approval to use the process bypass process in the first place.

Which is to say, nobody gets to bypass the process bypass process.