Most revenue plans contain a line like this: new accounts times average revenue per account. It looks harmless. It’s usually the weakest number in the model.
Revenue per customer is almost never bell-shaped. A handful of accounts produce most of it and a long tail produces very little. In a distribution like that, the average describes nobody. It sits far above the typical account and far below the big ones.
Plan on the median and size the tail separately
If you want to know what the next ordinary customer is worth, look at the median. In skewed data it can be a small fraction of the mean. A plan that assumes the mean is quietly assuming you’ll land more very large accounts, without saying so.
So split it. Model the typical account on the median, and treat large accounts as their own line, with their own count, their own sales cycle, and their own odds.
Look at concentration before anything else
Before trusting any total, sort the accounts and ask what share comes from the top ten. If it’s most of it, your “trend” is really a few customers’ behavior, and a single departure can pass for a market shift.
Report the spread
One number invites one conclusion. I’d rather show the median, the 90th percentile, and the count behind them. It takes one more line on a slide, and it prevents the worst meetings, the ones where everyone argues about an average none of them could find in the data.
The habit is cheap. Every time you’re handed an average, ask to see the distribution it came from. If nobody has looked, you’ve found the first thing to do.
Originally published at malekdamiri.com. I'm Malek Damiri, Head of Revenue Operations in Atlanta; more of my writing is at malekdamiri.com.
Top comments (1)
Checking the top ten share before trusting any total is a habit most plans skip. When you split the large accounts into their own line, how do you set their odds if you only have a handful of past deals to go on?