The Renewal Reminder That Went Out Six Weeks Late
David Selva · · 4 min read
Data Quality, Insurance Agencies, Custom Fields
Automation does not clean up a bad contact record — it runs it, faster, and in front of more people.
Automation does not clean up a bad contact record. It runs it, faster, and in front of more people.
I watched an agency send a renewal reminder to a commercial client six weeks after the policy had already lapsed. The automation worked perfectly. The date it was reading came off a row somebody had pasted in from a spreadsheet in 2023, and nobody had touched it since.
The problem before the automation problem
Most agencies I sit down with want to talk about sequences. What should go out, when, on which channel. It is the fun part.
Then I open a contact record and ask what the fields mean. Renewal date is filled in for maybe half the book. Policy type is a free-text box, so it contains "auto," "Auto," "personal auto," and "PA" — four values that the system reads as four unrelated things. Carrier is in the notes, in a sentence, where nothing can find it.
You cannot build a renewal sequence on that. Not because the builder is not capable of it, but because there is nothing dependable to trigger on. Every rule you write will be right for the half of the book that happens to be filled in correctly and silent for the rest.
Free text is where automation goes to die
The single most expensive habit in an agency database is typing information into a notes field.
Notes are for context — the client's kid is starting to drive, they mentioned a boat. Anything you ever want to filter, sort, or fire an automation on has to live in a field with a fixed set of choices. A dropdown with six policy types is not bureaucracy. It is the difference between a list you can act on and a pile of sentences.
The test I use: if I asked you to pull every commercial auto policy renewing in November, could you do it in one click? If the answer is "I'd have to go through and check," the field does not exist yet, no matter what is written in the record.
The three fields that carry almost all the weight
For most books of business, the automation you actually want depends on a very short list:
- Renewal date, as a real date field, on every single policy
- Policy type, as a dropdown with a fixed set of values you agreed on once
- Carrier, also a dropdown, because you will eventually need to message a whole block of clients when a carrier does something
Everything else can stay messy for now. Those three, filled in consistently, unlock renewal reminders, cross-sell targeting, and carrier-change notices — which is most of what an agency wants automation to do in the first place.
Nobody wants to do this part
The cleanup is genuinely tedious. It is a person, or a few people, going record by record for a couple of weeks.
I am not going to pretend there is a clever way around it. There are import tools and there is some pattern-matching you can do on obvious cases — "PA" and "personal auto" can be mapped in bulk. But the records where the renewal date is simply absent cannot be inferred by anything. Somebody has to look it up.
What I will say is that the cleanup is one-time and the alternative is permanent. An agency that skips it does not get less work; it gets the same work forever, done by hand, plus a set of automations that misfire often enough that the staff stop trusting them.
Where the discipline actually breaks
It is not the cleanup. Agencies get through the cleanup.
It breaks about four months later, when a producer is in a hurry and types a new client's policy type into the notes instead of picking the dropdown. Then another one does. The field stops being reliable, quietly, and nobody notices until a renewal batch goes out short.
Whatever you clean up, you need one rule about how a record gets created going forward, and one person who checks it. Making the fields required at entry is the version of this that does not depend on anybody remembering.
Try the November question
Before you buy or build anything, open your system and try to pull one list: every policy of one type, renewing in one specific month, with the client's mobile number attached.
If that list comes out in a few seconds, your data is in better shape than most and you are ready to automate against it. If it takes you an afternoon, you have just found the actual project — and it is not the sequence you thought you were going to build.