The average customer spends 40 dollars a month. It is one of the most reassuring sentences a founder can write, and one of the most misleading. That single number feels like it sums up the whole business, when most of the time it describes nobody in particular.
I run my own small businesses, and I have been fooled by my own averages more than once. I have read an average order value, planned around it, and only later found out that one big customer was holding the whole number up while everyone else sat far below it. So here is the plain version of a distinction that quietly decides whether your dashboard tells you the truth: the difference between the mean and the median.
What the mean and median actually measure
The mean is the average you already know. You add up every value and divide by how many there are. Add up what all your customers paid, divide by the number of customers, and you get the mean revenue per customer. It is quick, it uses every number you have, and for tidy data it is exactly the right summary.
The median takes a different route. Instead of adding anything up, you line all your values in order from smallest to largest and take the one sitting in the middle. Half your customers are above it and half are below it. It does not care how big the biggest value is or how small the smallest is, only where the middle of the pack sits. That one difference is what makes it so useful when the data is not tidy, which for a small business is most of the time.
Why the two numbers disagree
Here is the whole idea in one line. The mean gets dragged toward extremes, and the median does not. One unusually large value pulls the mean up toward itself, while the median barely moves, because the middle of the line is still the middle of the line.
When your data is lopsided, with a few big values off to one side, the mean and the median drift apart. The size of that gap is itself a warning that an average is hiding something. When they sit close together, either one tells the truth.
A worked example
Treat these numbers as an illustration, not a benchmark. Say nine of your customers pay 20 dollars a month and one pays 1,000. Add it all up and divide by 10, and the mean is 118 dollars. It sounds like a healthy business full of 100 dollar customers.
But the median, the middle of the line, is 20 dollars, because nine of your ten customers pay exactly that. The mean here is describing a customer who does not exist. If you priced, staffed, or forecast around that 118, you would be planning for a business you do not actually have.
The average customer who does not exist
That is the trap worth remembering. An average can land in a gap where nobody actually sits. If half your buyers pick a cheap plan and half pick an expensive one, the mean lands neatly in the middle, on a price point that not a single customer chose. You can spend a month designing for that imaginary average person, when your real business is two different groups who want two different things, and the mean has smeared them into one blurry face.
When to reach for each one
None of this makes the mean bad. When your values cluster gently around a center, without wild outliers, the mean and the median sit close together and either one is honest. Heights, delivery times that rarely go crazy, ratings on a tight scale. The mean is perfect for those, and it reacts to small shifts the median would shrug off.
The median earns its keep exactly where the mean struggles, on money and time and anything with a long tail. Revenue per customer, order values, how long people take to reply, days to first purchase. All of these have a floor at zero and a few enormous values stretching far to the right, and that shape drags the mean upward every time. For numbers like these, the median is usually the more honest headline, and quoting it saves you from believing your own biggest outlier.
The mode, and the spread the average hides
There is a third average worth a mention, the mode, which is simply the value that shows up most often. It is the odd one out because it works even when your data is not numbers at all. You cannot take a mean of plan names, but you can ask which plan customers choose most, which reason for cancelling comes up most, which day of the week is busiest.
And even the right average only tells you the center, never the spread. Two products can share the exact same average rating while one pleases everyone mildly and the other splits the room into lovers and haters. The single number looks identical; the reality could not be more different. The cheap way to see that spread is percentiles. The median is just the 50th percentile, the middle. Add the 90th percentile, the value 90 percent of your data sits below, and you get a feel for the tail without letting one freak value hijack the summary. Better still, sketch a quick histogram: bars showing how many customers fall into each range reveal one gentle hump, two separate humps, or a long thin tail at a glance, and that picture tells you which average to trust faster than any formula.
Traps that quietly break an average
Outliers deserve their own warning, because in a small business they are often not real customers at all. A test account you forgot to delete, a refund logged as a sale, a bot, a single enterprise deal ten times your normal size. Any one of these can yank the mean somewhere silly, and the smaller your data, the harder it yanks. Before you trust an average, glance at your biggest and smallest values just to check they are genuine.
Small samples make it worse. With a few dozen numbers, one new large value can swing the mean noticeably from week to week, so a moving average looks like a trend when it is really just one big customer arriving and leaving. The median holds much steadier on tiny data.
And do not average an average. If a group of 100 customers converts at 2 percent and a group of 10 converts at 20 percent, the honest overall rate is not the 11 percent you get by averaging the two percentages. It is the four conversions spread over all 110 people, under four percent, because the big group deserves far more weight. Whenever you are tempted to average rates, go back to the raw counts.
How to check it in one cell
The practical version of all this is almost embarrassingly simple. In any spreadsheet, right next to the average function you already use, there is a median function that takes the exact same range. Put them side by side. If the two numbers are close, relax, either one is fine. If they are far apart, trust the median as your headline and go look at why, because that gap is your data telling you an outlier or a lopsided shape is in the room.
There is one honest exception where the plain mean is exactly right: whenever a total is the thing you actually care about. For budgeting, the sum of what everyone paid is what lands in the bank, and the mean is just that total shared out. It is only when you are trying to describe a typical customer, rather than add up all of them, that the median usually deserves the microphone.
The honest limit
No single number can describe a whole group, and pretending otherwise is the quiet original sin of every dashboard. The mean, the median, and the mode are three different honest answers to three slightly different questions. The skill is not memorizing formulas, it is picking the summary that fits the shape of your data and saying out loud which one you used.
Get this right and your averages stop lying to you. Every metric and method like this one is explained in plain English at dataresearchanalysiscollection.com, so you can read it again slowly with your own customer list open. No hype, no promises about your results, just the idea explained until the gap between your mean and your median starts telling you things you were missing.
Get new guides and videos first — join the Telegram channel.