A Service Level Expectation (SLE) is a data-driven forecast of how long items typically take to pass through the system. It replaces deterministic deadlines with probabilistic statements and thereby creates more realistic expectations for stakeholders and customers.
Instead of promising that a feature will be done in ten days, a team using an SLE might state: 85 percent of our items are completed within twelve days. This statement is based on historical lead time data and becomes more reliable the more data points are available. When a stakeholder asks when a particular item will be finished, the team can offer different probabilities: 50 percent likelihood in seven days, 85 percent in twelve days, 95 percent in eighteen days. This is more honest and more useful than a single estimate, which typically turns out too optimistic.
The SLE was introduced as an official practice in the 2020 Kanban Guide. It requires the team to systematically capture its Lead Time data. Without data, there is no reliable forecast.