CRM Fields for Insurance Agents: What Actually Belongs on a Contact Record

Every CRM an agent has ever been handed comes with a contact record that is either almost empty or absurdly full. The generic sales CRM gives you Company, Job Title, and Deal Stage, none of which mean anything when you are selling a $12,000 final expense policy to a 68-year-old. The upline-provided platform gives you sixty fields, forty of which nobody has ever filled in.

Both failures cause the same behavior. You stop looking at the panel. And once you stop looking at the panel, the four or five fields that would have saved you a bad call go empty too.

This is the field list I would build from scratch for an agent whose day is dialing their own lead list. It is short on purpose. Every field on it has to earn its place by answering one of three questions: can I legally call this person right now, what happened last time, and what am I selling them.

The test a field has to pass

Before the list, the rule. A field belongs on the contact record only if something reads it. Not “it would be nice to know” — something. A filter, a queue order, a compliance block, a merge tag in a follow-up, a report you actually open.

If nothing reads it, it is a note. Notes are fine. Notes are where “wife passed in March, do not open with the spouse question” lives, and no structured field would hold that better. The mistake is turning every observation into a dropdown and then wondering why nobody fills anything in.

The corollary: if a field is empty on more than half your records, it is either the wrong field or it needs to be captured automatically. Fields that a human has to remember to type during a dialing block lose. Fields the dialer writes by itself win.

Group 1: Identity and reach

The boring ones, with two that agents routinely get wrong.

• First and last name, separate. You need the first name alone for openers and merge fields, and a single “Name” column turns into “Hi MARIA GONZALEZ” on a text.
• Primary phone, normalized to a single format on import. Mixed formats are the number one cause of duplicates that a dedupe pass misses.
• Alternate phone. On aged lists this is worth more than almost anything else on the record, and most agents drop the column at import because their CRM only maps one phone.
• Line type — cell or landline. It changes which consent rules apply and whether texting is even on the table.
• Email. Not for blasting. For the one follow-up that keeps a slow prospect alive between dials.
• State. The single most under-rated field on this list, and the reason it gets its own section below.
• Date of birth or age. Underwriting depends on it, and it lets you sort a list by product fit before you dial it.

Street address is optional for a telesales agent and mandatory for a field agent. If you never drive to a kitchen table, ZIP and state are enough, and a full address you never verify is just another empty field.

Why state beats area code

Calling hours are set by the called party's time zone, and the area code is a bad proxy for where somebody lives. Numbers moved with people for two decades. A 305 on a phone that now sits in Portland means an 8:30 a.m. Eastern dial is a 5:30 a.m. local dial, and no compliance system will save you if the only geography you stored was the area code.

So: capture state at import, from the lead form if it has one and from the ZIP if it does not. Let the area code be the fallback when both are missing, and flag those records so you know which ones are guesses. This matters more than usual if you dial across states with mini-TCPA laws that cut calling hours shorter than the federal window.

Group 2: The compliance fields

These five are the ones that decide whether a complaint is an afternoon of annoyance or a real problem. Every one of them should be written by the system, not typed by you.

• Lead source. Specific, not “Facebook.” Which form, which vendor, which campaign, which upload. When you need to answer “where did you get my number,” this is the field you read off the screen.
• Consent record. The actual artifact: form URL, the disclosure language shown, IP and timestamp, or the recorded permission. A checkbox that says “consented: yes” with nothing behind it is worth nothing.
• Consent timestamp. Separate from the record so you can filter by age. Consent does not expire on a schedule, but a two-year-old form on an aged list changes how you dial it.
• DNC scrub date and result. Per number, with the date. “We scrub monthly” is a policy; this field is the proof.
• Internal do-not-call flag. One switch that blocks every dial, text, and email on the record, immediately, permanently, and across every list the number ever appears on.

That last one is where homegrown setups fail. An agent marks a contact “DNC” as a disposition, then re-imports the same person from a different vendor six weeks later, and the flag does not follow because it lived on the old record instead of on the phone number. The block has to be keyed to the number, not the row.

If you want the full picture of what a defensible file looks like, the consent records post goes deeper on retention and revocation than I will here.

Group 3: The dialing fields

These drive the queue. If they are wrong, you dial the wrong records in the wrong order, which is the most expensive kind of wrong because it is invisible.

• Attempt count. Written by the dialer on every connect and every no-answer. Hand-counting attempts is how a lead gets nine dials in four days.
• Last attempt timestamp, stored in a real time zone so you can see you have called this person at 10 a.m. five times running.
• Last disposition, from a short fixed set. Ten codes is plenty; I laid out the set I use in the disposition codes post.
• Next callback — a date and time, not a “follow up soon” tag. This is the field your morning queue is built on.
• Best time to reach. Fill it only when the prospect tells you. A guessed value is noise; a stated one turns a fourth attempt into a conversation.
• Number dialed / caller ID used. Rarely offered, quietly valuable: when answer rates fall off a cliff you want to know which of your numbers went bad.

Notice what is not here: a “lead score.” For a solo agent working their own list, a score is a number you invented that hides the two fields you already had — attempt count and last disposition. Queue order should come from rules you can read out loud, not from a weighting nobody can explain.

Group 4: The sales fields

Only the ones that change the next call.

• Product interest — final expense, mortgage protection, IUL, term. Determines the opener and the carrier shortlist.
• Coverage amount discussed and budget stated. Two numbers that make the second call five minutes shorter.
• Health flags. A short list, not a medical history: tobacco, major conditions, any knockout answer. Enough to know which carrier you are quoting before you dial back.
• Beneficiary or spouse situation, when it came up. This is the field that keeps you from asking a widow about her husband.
• Stage — new, contacted, appointment set, quoted, application, placed, lost. Seven values, no more.
• Carrier, policy number, issue date, premium once something is written. This is what makes your book searchable a year later for annual review calls.

Field vs. note vs. tag

Agents mix these three up constantly, and the mix-up is what produces a CRM nobody trusts. The distinction is mechanical, not stylistic.

Tag rot is the most common of the three failures. Tags are cheap to create, so agents create one per campaign per month, and two years in the tag list is unusable and half the filters point at tags that stopped meaning anything. Cap yourself and prune during your monthly pipeline audit.

The other reliable failure is the date-in-a-note. “Call her back Tuesday after 4” typed into a note is not a callback. Nothing reads it, nothing surfaces it, and Tuesday goes by. If it has a time attached, it is a field.

Where the fields come from

Half of this list should be populated before you ever see the record. The lead form or the vendor file already carries name, phone, state, age, product interest, and the consent artifact — and the place those get lost is the import. A CSV mapped by hand at 9 p.m. drops the alternate phone, mangles the consent column into a note, and puts the whole thing in your list with no scrub date.

So the field list and the import process are the same project. Set up the mapping once, per vendor, and save it. The import checklist covers the order; the point here is that the fields you defined only exist if the import fills them.

The other half — attempts, dispositions, callbacks, scrub dates, caller ID used — should come from the dialer automatically. Any field on that list that depends on you remembering to update it after a call will be about seventy percent accurate on a good week, and seventy percent accurate is worse than absent, because you will trust it.

A quick audit you can run today

Open your CRM, export your contacts to a spreadsheet, and do three things.

1. Count the fill rate on every column. Anything under fifty percent is a candidate for deletion or automation. Decide which, per field, and do it.
2. Pick five contacts you dialed last month at random and try to answer, from the record alone: where did this number come from, when was it scrubbed, how many times have I called, and what happened last time. If you cannot answer all four on all five, that is the gap.
3. List every field you would have to check before a call to know it is legal to place. If that check is more than a glance, it will not happen at dial 60, and it needs to be a block the system enforces instead of a habit you maintain.

That third one is the real dividing line between a CRM that is a filing cabinet and a CRM that is a guardrail. A record that merely stores the scrub date is a filing cabinet. A record where a stale scrub date, a DNC hit, or an out-of-hours state stops the dial from going out at all is something else.