- Suppression is an operating model, not a data feed
- Use two states, a soft hold and a confirmed hold, so you are never choosing between acting and waiting
- Every suppression needs an owner, a review clock and an exit path
- The evidence trail is what turns this into a defensible control
The failure mode nobody plans for
Most teams buy death data, wire it into a nightly job, and produce a list. Then the list lands with an operations team that has no defined next action, no authority to close anything and no owner for the exceptions. Within a month the list is being ignored and the original problem is still live.
The match is the easy part. What makes suppression work is deciding, in advance, exactly what happens to an account when a deceased signal arrives, who does it, how quickly, and how somebody gets un suppressed if the match was wrong.
Two hold states, not one
A single deceased flag forces a bad choice. Set the bar high and you miss real cases. Set it low and you freeze living customers.
Two states solve it. A soft hold triggers on a moderate confidence match and does one thing only: it stops outbound contact. No collections call, no marketing, no retention campaign. It does not restrict the customer, it does not change their balance, and it is invisible to them. The downside of a false positive is that somebody does not get a phone call.
A confirmed hold triggers on a high confidence match or on a soft hold that has been validated by a human. This is the state that stops payments, restricts new credit and starts the estate process. It carries consequences, so it needs a higher bar and a name attached to the decision.
The workflow, end to end
The shape below works for servicing, collections and retention with minor changes. Write it down before you turn on the feed.
- Match runs on a defined cadence against the live book, with a full sweep and a daily delta
- Moderate confidence match applies a soft hold and suppresses all outbound contact immediately
- High confidence match, or a validated soft hold, applies a confirmed hold and opens an estate case
- Every hold is assigned to a named owner within one business day
- Confirmed holds route to the estate or bereavement team with the date of death attached
- A review clock runs on every soft hold so nothing sits in limbo indefinitely
- A documented release path exists for false positives, with the reason recorded
- All state changes, the evidence behind them and the person who approved them are logged
Getting the estate handover right
The moment an account is confirmed deceased, the counterparty changes. You are no longer dealing with a customer, you are dealing with an estate, and often with a grieving family member who has been asked the same questions by four different departments.
Two things make this materially better. First, carry the date of death through to the estate team so they are not asking the family for information you already have. Second, collapse the number of contact points. A single named handler beats a queue every time, and it is the difference between a complaint and a thank you letter.
This is also where the commercial argument lives. Handled badly, a death in a customer household costs you the surviving partner's business too. Handled well, it is frequently the moment a family consolidates with you rather than away from you.
Evidence, audit and the annual review
A suppression control that cannot be evidenced is a suppression control that does not exist as far as an examiner is concerned.
Log the match itself, including which identity elements agreed and at what confidence. Log the state change and its timestamp. Log the human decision on any confirmed hold, with the approver. Log releases and the reason. That record is what lets you show a regulator, an auditor or your own board that the control operated, not just that it was purchased.
Review the thresholds at least annually against outcomes. If soft holds are almost never escalating, your moderate threshold is too loose. If confirmed holds are producing releases, it is too tight. Both are fixable, but only if you kept the data.
- Match evidence retained with the matched elements and confidence
- Timestamped state transitions with the triggering rule
- Named approver on every confirmed hold and every release
- Threshold performance reviewed on a defined cadence
Common questions
How often should we run the portfolio sweep?
Run a daily delta against newly published death records and a full portfolio sweep on a slower cadence, typically monthly or quarterly depending on book size. The daily delta catches the deaths that just happened, which is where the exposure is. The full sweep catches anything that was missed or that matched later because the record was corrected or enriched.
What confidence threshold should trigger suppression?
There is no universal number because it depends on the identity data quality in your own book. Start by running the match in observation mode for a few weeks, plot the confidence distribution against manual review outcomes, and set the soft hold threshold where the cost of a missed call is clearly lower than the cost of a wrong call. Set the confirmed hold threshold materially higher and require human validation below it.
Do we need to freeze the account as soon as we get a match?
No, and doing so on a moderate confidence match is usually the wrong call. Stopping outbound contact carries almost no downside if you are wrong. Freezing an account carries a significant one. Separate the two decisions and give them different thresholds.
See what this looks like in your portfolio
Upload a sample file or call the API and get confirmed deceased matches with date of death, age and last known address. You are only charged for records we confirm.