Insurance is a business where the phone is the pipeline. Quote enquiries, coverage questions, claims and renewals all arrive as calls. And unlike a restaurant booking, a lost quote rarely comes back: the prospect has usually called three agencies and buys from whichever answered.
Why the loss is structural
The best time to call is the worst time to answer. Quote enquiries cluster around lunch, after work and Saturday morning — the same pattern as other consumer businesses. An agency open 9–5 misses the majority of consumer intent.
Quotes take time, so the phone backs up. A commercial quote can be a fifteen-minute conversation. While it happens, the line is busy, and the next caller hears an engaged tone and calls someone else.
Renewal season is a wall. When a book renews in the same month, the phone volume multiplies for two weeks and then returns to normal. Staffing for the peak is uneconomic; not staffing for it costs the renewals.
Claims cannot wait. A claim call at 21:00 is a person whose house has flooded. That call must reach a human, and it must reach them fast.
Sorting the calls before buying anything
This exercise is free and decides whether any system can help.
Bucket A — answer from policy, no advice given:
- office hours and location
- what documents are needed for a quote
- how to make a payment
- how to submit a claim and what happens next
- whether a specific product is offered (not whether it covers something)
Bucket B — collect, structure, route to a licensed person:
- new quote enquiries (personal and commercial)
- mid-term changes: address, vehicle, additional driver, new employee
- renewal questions where the answer depends on the policy wording
- complaints and billing disputes
Bucket C — never automated, route immediately:
- anything that is advice: "am I covered if…", "should I claim?", "is this excluded?"
- an active claim or a loss in progress
- cancellation decisions
- anything involving vulnerable customers
Bucket C is not a technical limitation. It is a regulatory one. Advice given by an unlicensed automation is a compliance event, and it does not stop being one because the wording came from your own website.
The arithmetic, framed honestly
``
quote enquiries per month: 120
share lost to busy lines or after hours: 25% -> 30
share that would have converted: 22% -> 6.6 policies
average first-year commission per policy: 320
= monthly commission at risk: ~2,100
``
Two adjustments matter:
Renewal value compounds. A policy kept for four years is worth several times the first-year commission, so the lifetime value of the lost quote is considerably higher than the monthly figure.
Commercial lines are lumpy. One lost commercial account can exceed a month of personal-lines leakage. If your book is commercial, the arithmetic is dominated by a handful of calls.
Where agencies actually lose the call
1. The engaged tone. During a long quote conversation, nobody can take the next call. There is no queue on a single line.
2. After 5pm. Consumers shop for insurance in the evening. The agency is dark.
3. The "let me get back to you" quote. Details taken on paper, sent to an underwriter, callback in two days. The prospect has already bound cover elsewhere.
4. Renewal spike. Nobody can absorb a 3× volume fortnight without dropping calls.
What a system must do to be worth it — and what it must refuse to do
Requirements:
- Capture structured quote data — name, contact, product type, key risk details, preferred callback window — so the licensed person calls back with the information already assembled.
- Answer administrative questions from your own written policy: documents needed, payment methods, claim process.
- Route claims immediately to the claims path, with the caller's details, and never attempt triage.
- Refuse advice, explicitly and consistently. "I can't advise on coverage, but I can have a licensed agent call you in ten minutes" is the correct answer, every time.
- Handle the renewal spike without a queue.
- Log every interaction for audit.
Refusals are the design constraint:
- It must not confirm coverage, exclusions, or whether a claim would be accepted.
- It must not discuss pricing that is not published.
- It must not take a cancellation instruction.
Compliance questions to settle first
Before any deployment, get a written answer to these:
1. Which jurisdictions do you operate in, and what counts as advice in each? 2. Are calls recorded, and how is consent obtained? 3. Where are transcripts stored, and for how long? 4. Who audits the agent's refusals, and how often? 5. What is the escalation SLA for a claim call at 21:00? 6. How are vulnerable customers identified and routed?
None of this is a software feature. It is your governance, and the vendor should be able to support it.
Where this is the wrong purchase
- A single-producer agency with fewer than 40 enquiries a month. A callback discipline and an after-hours message will do more.
- No CRM integration possible. Without writing into the book, the agent creates paperwork rather than reducing it.
- A book that runs on personal relationships. Clients who call and expect their broker may read automation as a downgrade.
- No appetite for auditing refusals. In insurance, an unaudited agent is a compliance risk, not a saving.
What to measure
- Quote enquiries captured per week, and share converted
- Callback time: enquiry to licensed contact
- Engaged/busy rate before and after
- Claims routed within the SLA
- Refusal audit pass rate — sample monthly, this is the compliance metric
The last one is the only one that matters if the others conflict.
Disclosure: I work on RingBot, an Agent-as-a-Service platform whose voice agents are built with an AI Builder, bound to your own knowledge base, and escalate to a person with a structured summary. It is a general platform, not a licensed advisory system, and the governance questions above are the agency's to answer. See also the same arithmetic for clinics, hotels and homestays, letting agents, restaurants, gyms and dental practices.
Top comments (0)