A VIP watchlist: catching churn before it happens

Top customers rarely complain loudly before they leave. They get a little quieter, order a little less, and one day the account is just gone. By the time you notice, the useful moment passed weeks ago. I manage VIP relationships with 95% retention, and the most important part of that is not charm. It is a watchlist that tells the right person at the right time.

Why VIP churn is quiet

Two things make high-value customers different from the average:

• They have less patience and more options. One bad experience weighs more.
• They often have a personal contact, which hides the data. The relationship feels fine to the owner right up to the point it is not.

So a feeling is not good enough. I wanted a short list, updated automatically, that says who needs attention today and why.

What goes on the watchlist

I use three inputs per customer, all already in HubSpot or one export away:

• Order frequency: compared to the customer's own pattern, not to an average.
• Spend: a drop against their usual level.
• Complaint history: open or recent tickets, and whether they were resolved to the customer's satisfaction.

From those, a handful of simple triggers decide who appears:

1. A gold-tier member has gone quiet for longer than their usual gap.
2. A VIP has complained twice in a month.
3. A VIP's last ticket was escalated and closed with no follow-up contact.
4. Spend has dropped for a sustained stretch.

I deliberately kept these as plain rules rather than a score. A rule can be explained to the account owner in one sentence, and that is what makes them act on it.

The alert

When a customer meets a trigger, the account owner gets a Slack message with the context they would otherwise have to hunt for:

• Who the customer is, their tier and how long they have been with you.
• Which trigger fired, in plain words.
• The last two interactions and the status of any open ticket.
• A suggested next step, drafted by Claude, for the owner to accept or ignore.

One alert, one owner, one suggested action. If an alert goes to a channel of ten people, none of them feels responsible.

The gate: a person reviews before any outreach

This is the rule I will not bend. The system proposes, a person decides. Alerts are reviewed by the account owner before anything goes to the customer, because the wrong message to a valued customer costs more than a late one. The draft may be well written and still be wrong: the customer might have just spoken to someone on the team, or be going through something that makes a cheerful message tone-deaf.

A worked example

An illustrative alert posted to the key accounts channel:

Watchlist: gold member, quiet Customer: M. (gold, long-standing). No order for longer than their usual gap. Last contact: a delivery complaint, closed, no follow-up. Trigger: quiet after unresolved-feeling ticket. Suggested step: personal call from the owner, apologise for the delivery issue, ask if the schedule still works. No discount in the first message.

The owner reads it, knows the customer, and decides to call rather than message. They note the outcome on the contact record in HubSpot. If the customer orders again, the trigger clears. If not, it escalates to a second look after a set number of days.

What went wrong the first time

The first watchlist was too long. Every small dip in orders produced an alert, and owners learned to ignore it within days. An alert nobody trusts is worse than no alert, because it creates the feeling of being covered.

I fixed it by narrowing the triggers to the four above, capping the number of alerts per owner per day, and asking owners to tell me which ones were useless. Their answers shaped the rules more than any analysis did.

The second mistake was suggesting discounts as a default next step. A VIP who is annoyed about a late delivery does not want a coupon. They want to be heard. The suggestions now start with contact, and offers need explicit approval.

Closing the loop

An alert is only half the system. Every alert ends in a recorded outcome on the contact: called, replied, ordered again, no answer, or customer asked to be left alone. I review these outcomes with the owners on a regular rhythm. They show which triggers lead to a saved account and which only create work. Triggers that never lead anywhere are removed. Triggers that regularly do are tightened, so they fire earlier. Over time the list gets shorter and more accurate, which is the opposite of how most alerting systems age.

Trade-offs

• Fewer, better alerts mean you will miss some quiet churn. That is acceptable if the alerts you send are acted on.
• Rules need tuning per business. A weekly pattern and a monthly pattern need different definitions of "quiet."
• Personal relationships will always beat automation for the final step. The automation is there so the personal step happens at the right time.
• The watchlist depends on clean ticket data. If tickets are closed without a resolution note, the complaint trigger is weaker.

The VIP list feeds into the lifecycle campaign design, where VIPs are excluded from mass win-back sends. And when a VIP complaint becomes an escalation, the owner and clock rules apply.

What I would build for your team

I would sit with the people who own your top accounts and ask which quiet signs they have missed in the past. Those stories become the first triggers. I would build the watchlist in HubSpot, send alerts through Slack to the owner only, and require a human review before any outreach. After a short period I would ask the owners which alerts were useful and delete the rest.

Email: kodirovrukhillo@gmail.com