DEV Community

TechnoXander
TechnoXander

Posted on

What Compliance Queues Teach You About Sprint Planning

 Engineers know Little’s law even if they never call it that: the number of items in a system equals the arrival rate multiplied by the time each spends there. It applies to alert queues as well as tickets.

Take a hypothetical team with 100 alerts arriving per day, each spending two days in the system. About 200 are open at any moment. Triple the arrivals to 300 per day with the same two days and about 600 are open. Only two things bring that number back down: more people, or less time per alert.

That is why programmes that break at scale tend to break in the same place. Hiring changes the service rate, but each new person needs weeks before reaching the old time per alert. A tool that cuts time per alert changes the other side of the equation, so it is worth measuring before the queue grows.

Cases do not all weigh the same. A cross-border payment case can take several times longer than a domestic one, so queue length alone misleads. Track time by case type.

If the tooling is third-party, add its failure modes to the plan. The ICT risk steps for financial institutions are a decent checklist for what happens when it is down.

The finance version of this argument is in the original business-case article. Here are free Stress-test sheet & Hypothetical example.

Over to you

Do you track time per ticket by type, or only throughput?
Would you accept a tool that shrinks time per item but adds a review step?
Hot take: queue length is the metric most teams read least and should read first. Agree, or fight me?

Top comments (1)

Collapse
 
technoxander profile image
TechnoXander •

If you only get one number out of your ticketing tool, I would pick median time in system by type. Which would you pick?