Poisson Distribution
Your organization's donation page receives about 26 online gifts per day. It's been steady for months. Then yesterday, 38 donations came in. The development director credits a social media post that went moderately viral. But two weeks ago, a day with 34 donations passed without anyone noticing, and nobody could point to a cause. Is 38 genuinely unusual for a page that averages 26 per day, or does this kind of clustering just happen?
The Poisson distribution was built for exactly this kind of question. It describes how many times something happens in a fixed window of time when events occur randomly and independently at a known average rate. Think of it as a close relative of the binomial distribution, but for a different situation. With the binomial, you have a specific number of attempts, each with a yes-or-no outcome. With the Poisson, there's no fixed number of attempts. You're just watching a window of time and counting what arrives. Donations per day. Phone calls per hour. New email subscribers per week. Complaints per month. The only thing you need to know is the average rate.
The entire distribution is determined by a single number, the average rate, often called lambda. If your donation page averages 26 gifts per day, lambda is 26. From that one number, the Poisson distribution tells you the probability of seeing 15 donations, 26 donations, 35 donations, and so on. The most likely counts cluster around the average, but the distribution always has a tail stretching to the right. With an average of 26, seeing 35 or more donations on any given day happens about 4% of the time. That's uncommon but not extraordinary. Seeing 40 or more happens less than 0.5% of the time, and that would be genuinely remarkable.
One of the most counterintuitive properties of the Poisson distribution is how it handles variation. The standard deviation equals the square root of the average rate. When the average is 26, the standard deviation is about 5.1. When the average is 100, the standard deviation is 10. This means that as the average rate gets larger, the absolute variation grows, but more slowly relative to the total. A crisis hotline that averages 100 calls per day will regularly see swings of 20 calls in either direction, and nobody should worry. For that donation page averaging 26 gifts per day, the standard deviation of about 5 means daily counts between roughly 16 and 36 are perfectly ordinary. That day with 38 donations sits just beyond the upper edge of normal variation, which suggests something real may have changed.
The Poisson distribution shows up wherever your organization counts events over time. In digital campaigning, if your advocacy page normally gets 15 petition signatures per hour and suddenly jumps to 30, this distribution tells you whether that's within normal fluctuation or a genuine spike worth investigating. In program management, if your youth center averages 6 incident reports per month and one month produces 12, you need to know whether to launch a review or accept it as randomness. The answer depends on the math, not on how alarming the number feels.
It also helps with capacity planning. If you know your average rate, the Poisson distribution tells you how much to prepare for. An after-hours crisis line averaging 3 calls per night should plan for handling 7 or 8 in order to cover roughly 99% of nights without overwhelming staff. A volunteer coordinator who processes an average of 5 new applications per week should expect the occasional week with 10 or more, and the occasional week with none, even if nothing about the program has changed.
Random events are lumpier than they look. The Poisson distribution puts a number on how much lumpiness is normal, so you can separate a genuine signal from the natural clustering of chance.
See It
The shaded region shows normal variation. Drag the red "Today" marker to see how unusual any count would be. Use the slider below to change the average rate.
Reflect
Think about something your organization counts over a fixed time window. Online donations per day, new volunteer signups per week, complaints per quarter. Do you know the average rate? The next time that number spikes, will you be able to tell whether it's a real change or just the lumpiness that randomness produces?
When your team sees a slow day with only 15 donations, do they treat it as a problem to solve? For a page that averages 26 gifts per day, a day with 15 or fewer has only about a 2% chance of occurring. But over a full year, that still means roughly 7 slow days are completely expected. Is your organization reacting to signals or to noise?