Two accounts on your books, both at $890/month.

Any churn tool you own will flag Account A. Account B probably isn't on the list at all.

That's backwards. Account A is a problem. Account B isn't yours anymore — it just isn't official yet.

The one that dropped had value once. The one that never used it never did.

Account A onboarded, found something that worked, ran it 40 times a week for months, and cooled off. That's a relationship. It has a history and a discoverable reason: the team changed, the internal champion left, one of your deploys broke the flow they lived in. You call and ask. Sometimes they come back.

And they know what they're losing. That's the strongest retention argument there is, and you don't have to build it — only remind them of it.

Account B has none of that. There's no relationship to recover because there never was one. Calling them isn't reactivating a customer: it's making a fresh sale to someone who already said no with their behavior and kept paying out of inertia.

Account B isn't about to churn. It churned mentally months ago. The cancellation is just the formality of the day someone notices the line on the statement.

A drop makes noise. Being forgotten makes none.

A drop is a ratio:

usage(last 7 days) ÷ usage(previous 7 days)

Account A: 9 ÷ 40 = 0.225. It crosses the threshold, the alert fires, the line falls neatly on the chart.

Account B: 0 ÷ 0. The division doesn't exist. It crosses no threshold — not because it's healthy, but because the question doesn't apply to it.

That's why Account B disappears from the dashboard: by construction, not by oversight. A flat line at zero and a flat line at 40 are both flat lines. If you measure change, they're identical. One is your most stable customer, the other is revenue that never existed.

There's the asymmetry that explains the rest of the market. Measuring a drop needs a baseline — which means it needs the customer to have used the thing. Measuring neglect takes an entirely different question: how many days between the first charge and the first meaningful use? Another query, against another table, with a definition of "meaningful" somebody has to decide. That's why almost nobody runs it.

Forgotten revenue doesn't decay. It vanishes all at once.

Account A, left alone, goes 40 to 9 to 4 to cancelled, over months. There's time.

Account B goes from $890 to zero in a day. No ramp. It survives until the first event that forces someone to look at the invoice:

• the quarterly budget review
• the new CFO, asking for every recurring line
• the expense audit
• the card expiring

That last one is the cruellest. When an active customer's card fails, they update it — they need the product tomorrow. When Account B's card fails, nobody updates anything. You didn't lose a customer to a billing problem; you found out, seven months late, that the revenue was never yours. Dunning won't win that account back, because there's nothing on the other side that wants to be won back.

It's the churn that leaves telling the worst story

The customer who dropped and cancelled leaves with something lukewarm: "we used it for a while, it stopped making sense." Nobody repeats that in a founder group chat.

The one who paid for seven months without using it leaves with this: "I paid for seven months and never used it." That's $6,230 and a story that tells itself.

Look at what that sentence does. It doesn't say the product is bad. It says the product delivered nothing and charged anyway. And the person feels stupid — which nobody tolerates for long. Within a week the blame turns into "they sold it to me and disappeared." That's the version that travels.

One account that dropped is an incident. Five forgotten accounts are a symptom.

This is the part that changes what you do on Monday morning.

Take five accounts that dropped. They're five different stories: one switched tools, one cut budget, one lost its champion, one had its flow broken by a deploy of yours. Five causes, five conversations. Account work.

Now take five accounts that pay and never activated. Add it up: 5 × $450 = $2,250 a month, $27,000 a year.

The money isn't the worst part. The worst part is that it's almost never five causes. It's one:

• onboarding stalls at the same step every time, and nobody comes back after stalling
• activation depends on an integration that needs an engineer the buyer doesn't have
• sales promised one thing and the product does it another way, which the customer discovers on day one

An account that dropped gives you work. Five forgotten accounts give you a diagnosis — free, quantified, already grouped by cause. If five customers pay and never activated, you didn't have five pieces of bad luck: you have a manufacturing defect, and it is paying for itself while you stare at the drop chart.

And it compounds. As long as it exists, every new sale enters the same funnel and a predictable fraction becomes another Account B. You're not retaining badly — you're selling into a leaky pipe and calling the leak churn.

The right order: never used > gone quiet > dropping

Three states, worst to least:

Most tools rank it the other way round, and not out of stupidity: they rank by size of signal, which is what the system knows how to compute. 90% > 60% > 40% is obvious, defensible and easy to draw. The catch is that the first state has no percentage: it doesn't lose the ranking, it never enters it.

Severity isn't magnitude. A 90% drop and an account that never opened the product aren't on the same scale; measuring both with the same ruler is the error at the source.

Severity breaks ties; money does the ranking. A 90% drop on a $90 account can wait; a 40% drop on a $4,200 account cannot. And a $4,200 account that never opened the product is the first call of your week — even though no chart you own has pointed at it once in seven months.

---

That's why, at Keep, "never showed up" is a risk reason of its own, and it ranks above going quiet and above dropping.

But the tool is a detail. Run the query yourself, today: list every active customer, find each one's first meaningful use, separate the ones who never had it. Add up that MRR.

It's the most uncomfortable number on your books — and the one almost nobody has calculated.