How a steady trickle can still produce a burst
Suppose a help desk receives, on average, four urgent calls an hour. That average does not mean one call arrives every 15 minutes. Calls may come close together, followed by a quiet stretch. The Poisson distribution describes the probabilities of those different counts when each call occurs independently and the underlying rate stays roughly steady.
The expected count sets the curve
For a chosen time window, the average number of events is called lambda. If lambda is four calls per hour, the model assigns a probability to getting zero, one, two, or more calls in an hour. The chance of exactly k events is calculated as e⁻λ λᵏ / k!; in plain terms, the expected count determines how likely each count is, while the factorial makes very large counts increasingly uncommon.
A useful consequence is that the model predicts both ordinary variation and occasional surprises. Even when the average is four, an hour with many more calls is possible—it is simply less likely than an hour near four. The distribution gives a way to judge whether a busy interval is consistent with random fluctuation or unusual enough to investigate.
When the model stops being useful
The assumptions matter. If one outage triggers a flood of calls, events are no longer independent; if demand rises at lunchtime, the rate is not steady. In either case, a single Poisson curve can misrepresent the real pattern. Used where its assumptions fit, though, it turns an average into a full set of probabilities—not a promise about what the next hour will bring.
