Red Account

When one person owns sixty customer accounts, treating all sixty equally isn't an option. The trouble is that sixty rows on a screen all look the same size. Each one shows a company name, a contract value and a renewal date, and none of that tells you which one to open on Monday morning.

So people worked from noise. Customers who filed a lot of tickets and called often got answered first, and customers who quietly used the product week after week never came up. The quiet ones were frequently the dangerous ones. When an account with no tickets for three months gave notice at renewal — the point where a contract term ends and both sides decide whether to continue — the post-mortem always produced the same question: how did nobody see this coming?

The red account exists to answer that by cutting the list in two: accounts that need action now, and accounts that don't. The point is to turn sixty rows into eight, and those eight get opened by someone this week. It isn't a ranking. It's a single line drawn across a list, and where exactly the line sits matters less than the fact that it exists and that everyone reads it the same way.

The classification is built by layering an automatic rule and a human judgment. Four parts.

1. A score threshold does the first pass. At Pagedeck, any account whose health score — several usage signals weighted into one number between 0 and 100 — drops below 45 is flagged automatically. The threshold is a single fixed number rather than each person's instinct, because instinct means the whole list changes shape when the account changes owners. The specific number came from looking backward: Pagedeck checked the scores of accounts that had cancelled over the previous year and found most of them spent about a month in the mid-40s first.
2. People can move accounts in and out by hand. There are always facts the score cannot see. That the customer was acquired, or that next year's budget was cut in half, appears in no usage log; it comes out of someone's mouth, usually in a quarterly business review. The reverse happens too — an account at 38 with company-wide training booked for next week gets moved back to green.
3. Actions are attached to the flag. A label with nothing bound to it turns the list into an observation log. Pagedeck assigns an owner to every flagged account, puts it at the top of the weekly agenda, and activates an escalation path — an agreed route for who gets pulled in — once the account has been red for two weeks.
4. The exit condition is written down with the entry condition. Without a rule for what returns an account to green, accounts stay flagged for six months, the list grows to twenty rows, and you are back to not being able to choose. Pagedeck clears the flag when the score holds above 55 for three weeks.

Two states, or at most three. Add more middle grades and most accounts pile into amber, which dissolves the whole reason for drawing the line.

First, it sets the order of the weekly meeting. The customer success meeting at Pagedeck starts by reading the flagged list top to bottom, and green accounts aren't discussed. Because nobody has to decide in the room which accounts to cover, that meeting went from 40 minutes to 20.

Second, it allocates a person's week. Eight red accounts out of sixty means half of this week goes to those eight. The other fifty-two get only whatever was already scheduled.

Third, it feeds renewal forecasting. Counting how many of next quarter's renewals are currently flagged produces the bottom end of a revenue forecast, and finance gets that bottom end alongside the top.

Fourth, it sends a signal to the product team. If several accounts went red for the same reason, that isn't a problem any account owner can fix. Pagedeck collects the usage signals — behaviors like login frequency or feature use that hint at account health — behind each flag, and once more than five accounts are stuck on the same feature it goes to Mia. The last time that happened the blocker was permission settings, and the answer was a product change rather than sixty separate conversations.

"Eight red this week and three of them renew next quarter. Start with those three."

Inside the flagged list, renewal timing decides the order.

"Score says 38, but they have company-wide training next week, so I moved it back to green."

An automatic classification overridden by a person, with the reason recorded.

"Don't ask us to reduce the number of red accounts. Move the threshold and the count drops on its own."

Pushing back on someone trying to use the classification as a performance metric.

At Pagedeck, Sofia opens this list first thing Monday. For anything newly flagged she unpacks which signals pushed the score down, and when the usage data doesn't explain it she calls the customer and asks whether something changed on their side. She also checks what left the list, because an account that dropped off because its score recovered and an account that dropped off because the contract ended look identical on the screen.

Once an account has been red for four weeks, Grace reviews it too. Those conversations aren't only about saving the account — they also cover what the company promised badly when it first signed the customer.

The most common misreading is that red means this customer is leaving. The phrase "likely to churn" invites it. But the flag is a priority marker, not a prediction. About one in four of Pagedeck's flagged accounts actually cancels within the year; the other three return to green after someone works on them. Read as a prediction, the flag makes people give up on flagged accounts and spend their time where renewal looks safe, which is the exact opposite of what the classification was built for. It also makes the flag self-fulfilling, since the accounts nobody works on are the ones that go on to cancel.

Treating the count of red accounts as a number to drive down is the other frequent mistake. A smaller count looks like healthier customers, so it's an easy goal to write. But moving the threshold from 45 to 35 halves the count by itself. A team handed that target tends to adjust the threshold or apply the exit condition loosely, and two quarters later there are three red accounts on the board and more cancellations than before. The number worth reducing is cancellations, not flags.