Most availability problems are queueing problems

When availability drops, the reflex is to look at maintenance capacity. Sometimes that is right. More often the constraint is a queue somewhere upstream that produces an identical symptom and responds to an entirely different intervention.

Mean time to repair is an average over a pipeline with several distinct segments: waiting for diagnosis, waiting for parts, waiting for a certified technician, waiting for an approval that batches weekly. Reporting them as one number guarantees effort flows toward the most visible segment rather than the most expensive.

We decompose the pipeline and cost each hold point in availability terms. It is common to find that half the elapsed time sits in two segments nobody has authority over, which reframes the problem from a maintenance investment into a policy conversation.

One installation portfolio we worked found the largest single contributor was a weekly approval cycle. No new capacity, no new parts, no new system. Just a change in batching that recovered days of turnaround.

The analysis is not difficult. What makes it rare is that it requires measuring a process as executed, including the workarounds, rather than as the procedure describes it.