Grow Without Breaking
The number that lies
Your follower count climbs every week and your total sales since day one only ever rise — yet the cash in the drawer keeps shrinking. A number that can only go up is a number that can never warn you.
Cumulative totals — total sales, total customers, total followers — have one fatal flaw: they can only rise. Add one sale to a million and the line still ticks up. They flatter you on the exact day the business starts to rot.
A vanity metric makes you feel good and tells you nothing to do; an actionable metric can go down, ties to a decision, and is measured per group over time. Ask of any number: 'If this changed, what would I do differently?' If the answer is nothing, it's vanity.
- Tag every customer by the month they first bought from you — that group is their 'cohort'.
- For each cohort, track one behavior over time: did they buy again within 60 days?
- Line the cohorts up side by side — January's rate, February's, March's — and read the trend down the column.
- A falling column means each new group loves you less — a leak no rising total will ever show you.
Split customers by first-purchase month. Of January's 100 new buyers, 42 bought again within 60 days (42%); February's group came back at 35%; March's at 27%. Your monthly revenue still rose — you kept adding newcomers — but each fresh group returns less. That falling column, not the rising total, is the alarm.
The average lies too. One 'whale' — a single huge order — can drag the mean up while nine out of ten customers spend almost nothing. The median (the middle customer) barely moves. Watch the median, not the average, or one big fish will fool you into thinking the whole pond is thriving.
Groupon's dashboards told a triumphant story: total revenue and subscriber counts rocketing, powering one of the biggest tech IPOs of 2011 at around a $13 billion valuation.
→ Within roughly two years the stock had fallen about 90%, as the metric that actually mattered finally came due.
They optimized signups and gross billings — vanity totals — while the real number, whether a merchant ran a profitable second deal, was collapsing; many merchants lost money on the first deal and flatly refused to return.
Open your records and pick two cohorts: everyone who first bought three months ago, and everyone who first bought last month. For each, count what share came back at least once. Write the two percentages next to each other. If the newer one is lower, you've found your leak.
A rising total can hide a dying business. Trust only numbers that can fall, tie to a decision, and are read per cohort over time — and watch the median, not the average.
Grow Without Breaking