Insurance Agency CRM Consultant: A 20-Point Hiring Scorecard

Score an insurance agency CRM consultant on 20 checks: policy dates, beneficiaries, pipelines, lead forms, bilingual agents and call rules. Includes a pass mark.

By Carlos Familia, CRM & AI automation expert. Updated October 16, 2026. 7 min.

A life insurance agent in Utah came to me with three requests. Capture the insured and every beneficiary at intake. Let the beneficiaries know when a policy issues. And put birthdays, follow-ups and the pipeline on one screen.

Intake now creates the insured as the main contact, adds each beneficiary as a linked contact of their own, opens two opportunities and writes a consent note. The screen became a console embedded in GoHighLevel, showing birthdays, holiday reminders, a kanban view of the pipeline and a tree of each policy's beneficiaries. Its automatic texts sit behind two kill switches. Then I counted what the book could support. Of 527 contacts, 126 carried the "client" tag and just 10 had a policy start date recorded, so most date-based reminders had nothing to fire on.

To hire an insurance agency CRM consultant, score them on the 20 checks below. Each is worth one point, and the point goes only to proof. 17 or more means hire on milestones. 13 to 16 means pay for one test task first. Below that, keep looking. And a zero on check 19, the test log, fails a candidate whatever the total.

Group 1: Do they build around policies and people?

• Check 1. They count your dates before designing reminders. Policy start, renewal and birthday: how many contacts actually have each one? The Utah gap (10 policy dates among 126 clients) is common, and a good consultant shows you your own number in the first week.
• Check 2. Beneficiaries get their own records. Each one is a linked contact, so a policy issuing can reach them. A name typed into a notes field can't receive anything.
• Check 3. Shared details live in one place. The agency name, sender, booking link and review link sit in custom values, so changing one updates every message. My build for a California agency uses 20 of them.
• Check 4. Every automatic text has an off switch you can reach. The Utah console has two.

Group 2: Do they structure pipelines the way agencies work?

• Check 5. One pipeline per line of business, with a folder per stage. That California life insurance agency runs 3 pipelines (11, 4 and 5 stages) and 46 live workflows sorted into 23 stage folders. Open a folder and you should know what fires when a policy moves.
• Check 6. Each objection reason and each call outcome gets its own sequence. One generic "not interested" follow-up throws away what the agent just learned on the call.
• Check 7. They know the one-opportunity limit. GoHighLevel won't open a second opportunity for the same contact in the same pipeline, even if the first was abandoned. For an agency in Puerto Rico, the workflow that advances the auto-insurance opportunity after a confirmed appointment can only push it forward. Ask how the consultant handles a client who comes back for a second policy.
• Check 8. Each workflow has a plain-language card that matches what is live. On the California build, comparing the roadmap cards with the live workflow definitions found 47 places to correct. A consultant who never makes that comparison hands you documentation that drifts from day one.

Group 3: Do they get leads and bookings right?

• Check 9. Every lead form is mapped, and the backlog has a plan. At the Puerto Rico agency, two Facebook forms had never been mapped. Mapping them pulled in 39 stuck leads in about 3 minutes, and none of those leads fired a workflow, so they needed a one-time follow-up of their own. My post on a GoHighLevel Facebook lead form not working covers the mapping.
• Check 10. The bot says "booked" only after the booking exists. The same agency complained twice that its WhatsApp bot was confirming appointments that never reached the calendar. It took a booking step pointed at the correct calendar, plus a rule against confirming before the reservation exists.
• Check 11. They open every calendar link the way a lead would. A calendar I created by API for the California agency had no slug, so its share link opened another company's calendar. A team member's personal work-hours schedule took precedence over the calendar's own hours, and the meeting type told leads they would be the ones calling. Ask the consultant to open each share link in a private window.
• Check 12. QA never touches a real calendar. GoHighLevel's trial chat books a real appointment when the scripted lead accepts a slot. During my QA at the Puerto Rico agency, one landed on the owner's calendar. Test against a calendar nobody uses.

Group 4: Do voice, chat and language work for your clients?

• Check 13. The chat agent is wired to a channel. A chat agent set to auto-pilot can still sit silent. On the California build, it ignored website visitors for a day because routing had no channel assigned, and the trial chat passed the whole time. A second website ran a different widget that routing didn't include at all.
• Check 14. One language per contact, proven on a real Spanish call. Mixed language tags send reminders in the wrong language, so each contact carries one. The California agency's bilingual voice agent passed a full Spanish phone test on a 143-second call. Ask who on the candidate's team speaks Spanish well enough to run one.
• Check 15. A disclosure and a recording notice. A compliance review of that agency's voice agent added both, citing California's two-party consent rule. Ask who does that review for your state.
• Check 16. A balance alert and a backup number. The agency's voice line sat on a prepaid account that carried eight client numbers in all. When its credit hit zero, all eight went quiet at once, and three inbound calls to the agency in that gap may have been real prospects. Your consultant should set a low-balance alert and a number that rings your office.

Group 5: Do they follow the rules and show proof?

• Check 17. They know the calling window and the consent rule. The federal rule 47 CFR 64.1200 puts telephone solicitations to residential subscribers off limits outside 8 a.m. to 9 p.m., measured at the lead's location. Telemarketing with an artificial or prerecorded voice needs prior express written consent first. The consultant should work out each lead's time zone before anything dials. This isn't legal advice: your compliance person decides what applies to your calls.
• Check 18. Nothing waits on a text before texting is approved. A2P registration comes first. Until it's approved, a flow that cancels or escalates because a text got no answer will misfire. The GoHighLevel A2P guide has the full checklist.
• Check 19. They show a test log. Fake contacts move through every path, with each step's status visible. For the California agency, the integrity scan came back with 0 findings, and all 10 execution tests passed. How I test GoHighLevel workflows describes the method.
• Check 20. You keep control of access and data. You hold the admin login and pay the bills, and the consultant gets a user account with only the permissions the job needs. A dashboard that caches your contacts is a second copy of your client data. If they embed one, they show you that a signed-out request gets nothing back.

Where does your candidate land?

What does a passing score actually tell you?

It tells you the person can show work on the parts of an agency that generic CRM builders miss: dates, beneficiaries, language, booking and call rules. It doesn't promise they'll finish yours, so even a 20 gets a paid first task and milestones.

I don't check licenses, carrier appointments or filings. Those belong to your agency's compliance person. And this is my scorecard, not an industry standard, so rerun it after any big change to a form, a calendar or a workflow.

For the system itself, read insurance agent CRM or look at the insurance Automation Pack, a $22 GoHighLevel snapshot an agency can import on its own. My insurance page shows the builds behind this list. Bots that book are covered in GoHighLevel Conversation AI setup and how AI receptionists work, and the longer general version is my GoHighLevel expert hiring checklist. Want a second score on a candidate? Use the calendar below.

About the author

Carlos Familia, CRM & AI automation expert. GoHighLevel, n8n, AI agents, servers. Full-stack developer.

I build CRM systems in GoHighLevel, n8n automations on my own servers and AI voice and chat agents for service businesses. Full-stack developer: I write the workflow, the API integration, the database and the server it runs on, and I test every build with real journeys before handoff.