How can we implement 'Service Level Expectations' (SLE) for our Kanban board?

All answers to this question.

To calculate an SLE, you need a Scatterplot of your past Cycle Times. Look at your data from the last 3-6 months. Find the point where 85% of your tickets were completed. For example, if 85% of your tasks were done in 10 days or less, your SLE is "10 days." When communicating this, don't give a single date. Say, "There is an 85% probability that this item will be finished within 10 days of starting." This is much more honest than a traditional estimate because it accounts for the natural variability and "outliers" in your historical process.

   Answered 2025-11-16 by Mary Jones

That probability approach is very scientific, but how do you handle the 15% of tickets that go over the SLE? Do you have a specific "aging" alert on your board to flag tickets that have exceeded the 85th percentile, so the team can swarm them before they become 99th percentile outliers?

   Answered 2025-11-18 by Steven Taylor

• Steven, we absolutely do. Our Kanban tool turns the card red once it hits day 9 of our 10-day SLE. In our daily stand-up, we prioritize "red" cards over everything else. It’s reduced our "tail" significantly and made our overall delivery much more predictable for the stakeholders.

     Commented 2025-11-20 by Jessica Thompson

Using Monte Carlo simulations based on your Throughput data is another level up. It can tell you the probability of finishing a whole project (e.g., 50 items) by a certain date, which is great for roadmapping.

   Answered 2025-11-22 by Christopher Lee

• I love Monte Carlo! It takes the guesswork out of planning. I used it to show my boss why a December launch was only 20% likely, which allowed us to move the date to January with zero drama.

     Commented 2025-11-24 by Mary Jones