The Insurance Renewal Automation Playbook

TL;DR: Insurance renewal automation is a four-touch, multi-channel outreach system triggered 60, 30, 14, and 7 days before each policy's expiration. Built correctly on top of your AMS (Applied Epic, HawkSoft, EZLynx), it lifts retention 15-20% without adding headcount. This playbook covers the cadence, personalization rules, AMS integration patterns, cost model, and metrics that prove it's working.

Most independent agencies lose 8-12% of their book every year. The work that stops that loss — touching every renewing client, flagging at-risk policies, personalizing outreach by policy history and life event — is exactly the work that gets cut when the team is stretched. AI automation closes the gap: it runs the outreach the team keeps meaning to run, at the scale the team can't hit manually, and it does it without the brittle drip-campaign feel that clients instantly tune out.

This playbook is how we build it for agencies — the trigger dates, the channel mix, the content templates, the handoff to producers, and the metrics that tell you whether it's actually working.

What is insurance renewal automation?

Insurance renewal automation is a software system that triggers personalized outreach to policyholders ahead of their renewal date, using AI to draft the messages, segment the book, and decide which renewals need a human touch versus which can confirm without one.

A good renewal automation system has five components:

1. A scheduled job against your AMS that pulls every policy expiring in the next 60-90 days.
2. A classifier that scores each renewal on retention risk using claim history, engagement signal, rate delta, and life events.
3. A message drafter that produces personalized outreach in the agency's voice.
4. An orchestration layer that delivers messages across email, text, and voice and tracks responses.
5. A write-back channel that records every interaction back in the AMS as an activity so producers see the full history.

It's the opposite of a generic CRM drip campaign. Drip campaigns send the same three emails to everyone on the list. Renewal automation writes a different message to every household, every renewal, every year.

Why proactive beats reactive (and why most agencies are still reactive)

Reactive renewal work looks like this: a producer gets a carrier-generated expiration list, works the top of it, the shop runs out of time, and the bottom of the list quietly re-ups (or doesn't) on autopilot. The clients who actually need a conversation — the ones with rate shock, the ones who just had a baby, the ones who opened six emails about a competitor — get the same treatment as the ones who were always going to re-sign.

Proactive retention flips the default. Every policy gets outreach. The system decides how much, in what channel, and at what depth, based on what it knows about the client. Producers are pulled in by the AI, not the other way around, and only for the 10-15% of renewals where a human conversation moves the number.

Agencies that ship this well see 15-20% retention lift in the first full year. That's the number we reliably quote because it's what shows up in the Operations Audit data across the agencies we've worked with.

The four-touch renewal cadence

The spine of any AI renewal system is a trigger-date cadence. Ours is built around four touches, each with a specific job:

Touch 1 — Day 60: "We see you"

Sent 60 days before expiration. The goal isn't to sell; it's to reset the relationship. The message reviews what the client has, mentions any changes since last renewal (a teen driver turning 16, a home value reassessment, a new business location), and asks one open question: "Anything changing on your end before we put the renewal in motion?"

Why 60 days? Because that's when competitors start their quote-farming runs. If we're not top of mind at Day 60, we're fighting from behind at Day 30.

Touch 2 — Day 30: Personalized preview

Sent 30 days out, after the carrier has issued preliminary renewal numbers. This is the highest-stakes message — it's where rate shock happens. The AI drafts the message with the rate context built in: if it's a 4% bump, the copy reassures; if it's a 14% bump, it acknowledges the jump, explains the industry context, and offers a remarket conversation proactively instead of waiting for the client to ask.

Agencies that get Touch 2 right convert 70%+ of at-risk renewals directly. Agencies that don't watch those renewals shop and lose half of them.

Touch 3 — Day 14: The decision helper

Two weeks out. At this point the client has either confirmed, gone silent, or started asking questions. The AI routes by state:

• Confirmed — one-sentence thank-you, no sales copy, no upsell. Don't burn the goodwill.
• Silent — switch channels (if Touch 1 and 2 were email, send a text). Tone softens: "Wanted to make sure you got these. If something's off, easier to sort it now than the week of."
• Questions — a producer gets paged immediately with the context.

Touch 4 — Day 7: The hand-off gate

Seven days before expiration is the last automated moment. After this, everything is a producer call. The AI tees up the call list ranked by risk: clients still silent, clients who opened but didn't reply, clients with large commission exposure. The ranking matters — a team with capacity for 25 calls shouldn't waste them on low-risk re-signs.

Channels: when to email, when to text, when to call

The default stack is email + text + voice (AI-driven voice for first contacts, human voice for conversion calls). Rules of thumb:

• Email for context, text for urgency, voice for objections. Don't send a 200-word email and expect a reply — use it to deliver documents, rate detail, and the "we see you" relationship check-in. Use text for anything time-bound: "Heads up — carrier needs your signed form by Friday." Use voice when you need a real yes/no.
• Start with email for anyone over 55. Flip that for anyone under 35. Test the cutoff in your own book; it will drift.
• Stop sending when you hit the mute signal. Two unopened emails in a row or one unanswered text = move that client out of the email cadence and into a single producer touchpoint. The worst outcome is not losing the renewal — it's losing the client's trust because you spammed them at the worst possible moment.

Personalization that actually lifts numbers

The difference between AI automation that works and AI automation that gets clients asking to be removed from your list is how deeply the system personalizes. Five dimensions, in order of impact:

1. Policy-specific rate context. "Your auto premium is up $180, which tracks with the 8% state average for your ZIP." Clients hate rate bumps. They tolerate them when they understand them.
2. Life-event awareness. If a renewal client had a baby, bought a house, or got married in the last year, the message acknowledges it — not with saccharine copy, with relevance: "Now that you're at the new address, a quick coverage review makes sense before renewal lands."
3. Engagement history. A client who hasn't opened an email in 18 months doesn't get another email. They get a producer phone call, period.
4. Book-driven references. "You've been with us since 2019, so I wanted to personally look at this renewal." Say it when it's true; say nothing when it isn't.
5. Producer voice. The system drafts the message, but the producer's name and voice are on it. No "from the team at" — from Sarah, from Mike.

The build: what it actually looks like in your AMS

Technically, the renewal engine is a scheduled job that runs against your Agency Management System. Daily, it:

1. Pulls every policy with an expiration date 60, 30, 14, or 7 days out.
2. Looks up the client's record — policy history, household members, claim activity, last N interactions.
3. Runs a classifier to decide: send, escalate to producer, skip.
4. For sends, drafts the message using a prompt tuned to your house voice, with the rate and context data injected.
5. Queues the message in your outbound system (email via Gmail/Outlook, text via Twilio, call via an AI voice engine).
6. Writes the outcome back to the AMS as a note and updates the renewal pipeline view.

The integration work depends on your AMS. We've written a full integration guide for Applied Epic, HawkSoft, and EZLynx that covers the exact endpoints, export schedules, and webhook patterns — if you're evaluating whether this will actually work on your stack, start there.

The metrics that matter

Most agencies track renewal rate and stop. That number hides whether the system is doing anything — a renewal rate can stay flat while the composition of your book quietly degrades (you keep the easy renewals, you lose the profitable complex ones). The real scoreboard:

• Touch completion rate. Of renewals with an expiration in the window, what % received all four touches? Target: 98%+. If you're under 90%, something is broken in the data pipeline, not the messaging.
• Engagement rate by touch. Open rates, click rates, reply rates per touch. Touch 2 engagement is the most predictive number on the dashboard.
• At-risk conversion. Of the policies flagged at-risk by Touch 2, what % retained? This is where the AI earns its keep. Target: 70%+.
• Producer-handled share. What % of renewals ended up needing a human call? 10-15% is the sweet spot. If you're under 5%, you're letting clients churn silently. If you're over 25%, the classifier needs retraining.
• Net revenue retention. Not just policy count — premium dollars retained year over year. This is the number that matters for the P&L.

How AMS integration actually works for renewal automation

The system only works if the AMS layer is clean. Each major platform has a different integration pattern for renewal workflows:

Applied Epic

The Applied API exposes clients, policies, activities, and claim history. For renewal automation specifically:

• Daily batch pull: use Applied DataBridge to export policies expiring 60/30/14/7 days out. Batch is faster and doesn't count against API throttling.
• Event feeds: AppliedNet event subscriptions detect claim events and policy changes in near-real-time so the classifier can re-score in-flight renewals.
• Write-back: create an Activity on the client record for every AI-generated message. Use a dedicated Activity Type ("AI Contact") so producers can filter.

Timeline to production integration: 2-3 weeks on a clean Applied tenant.

HawkSoft CMS

HawkSoft's API (modernized through 2024-2025) covers clients, policies, and notes. Renewal workflow specifics:

• API pull: hit the policy endpoint nightly with a window filter. HawkSoft's export builder is a reasonable CSV fallback for full-book pulls.
• Write-back: write AI interactions as Notes on the Client record, never on the Policy. HawkSoft's policy data comes from carrier downloads and overwriting it creates sync nightmares.
• Opt-out enforcement: HawkSoft has a native consent field; the automation must check it on every send.

Timeline: 2 weeks on a modern HawkSoft setup.

EZLynx

EZLynx Management System API covers clients, policies, and accounts. Watch the client vs account vs household distinction — pulling "clients" without understanding EZLynx's hierarchy surprises people.

• API pull: policy endpoint with expiration window filter. Reporting exports as fallback.
• Write-back: Notes + Tasks on the Account.
• Custom fields: EZLynx custom fields are available but require a small config project inside EZLynx's support team.

Timeline: 2-3 weeks.

Full integration specifics including authentication, rate limits, and write-back patterns are in the AMS + AI Integration Guide.

How much does insurance renewal automation cost?

Done-for-you renewal automation typically runs $6,000-$8,000 for the initial build and $2,000-$3,000 per month for ongoing managed operations. Cost depends on three things: book size, AMS complexity, and whether you want us to run the system or you'd rather run it yourself.

Three cost models to consider:

1. DIY with AI tools ($0-$200/month in tool cost)

If your team has the time and enough technical comfort, you can build most of the four-touch cadence yourself using an AI email tool + your AMS's export features + a workflow tool like Zapier or Make.

• Realistic time investment: 40-80 hours of initial setup, 5-10 hours a week ongoing for monitoring and tuning.
• Best for: agencies with a principal who likes to tinker and a book under 200 policies.
• Limitation: DIY setups usually stop at Touch 1 and 2 (email) and never add the multi-channel, classifier, or write-back layers — which is where the retention lift actually comes from.
• Resource: the AI for Agent Retention course walks through this approach step by step.

2. Done-for-you Build & Launch ($6,000 one-time)

We build the whole system — four-touch cadence, classifier, multi-channel, AMS integration, write-back, opt-out enforcement — inside your environment in 2-3 weeks. Full pricing is on the Build & Launch offering.

• Includes: Applied Epic / HawkSoft / EZLynx integration, producer training, operations dashboard, and go-live monitoring.
• Best for: agencies with 200+ policies who want the outcome without the build.
• What the $6,000 buys you: a finished system. Not a template. Not a CRM seat. A system that's running messages through your book from Week 3 onwards.
• Credit: the $1,500 audit fee is fully credited toward this if you continue.

3. Managed Operations ($2,500/month)

Once the system is live, Managed Ops is the ongoing layer:

• Weekly classifier tuning as your book evolves (new carriers, new coverage types, seasonal patterns).
• Prompt optimization as your voice and offers evolve.
• Monthly strategy reviews with the agency owner.
• Priority support and new-automation requests within the same week.

Month-to-month. Cancel with 30 days' notice.

The economics for a representative agency

For a $2M personal lines book, the typical 12-month economics:

The full 12-month ROI math across all four revenue lines (retention, lead response, cross-sell, producer time freed) typically lands at 4-8×. See the Pacific Agency Group case study for the composite numbers on a $2.5M book.

A real-world example: 350-policy P&C agency on Applied Epic

For a worked example of the full rollout — starting state, what we built, 90-day outcomes, 12-month projection — read the Pacific Agency Group composite case study. It covers:

• 82% → 94% retention inside 90 days.
• 47 hours → 38 seconds median lead response time.
• $187K in recovered annual premium projected at month 12.
• What the principal wishes they'd known before the build.

For the commercial lines equivalent on HawkSoft, see the Ridgeline Commercial case study.

Frequently asked questions

How long until I see retention lift?

The first full four-touch cohort completes around month 4-5, which is when retention numbers start moving meaningfully. First 90 days: system is live and running; renewal outcomes still reflect pre-automation workflow. Month 4-6: first real signal. Month 12: full signal.

Will my team resist this?

They might, at the start. Two patterns reduce resistance: (1) frame it as capacity reclamation for the work they actually want to do (service, relationship) rather than replacement of what they do; (2) include CSRs in the first-100-message tone review so the system learns their voice and they feel ownership of how it sounds.

What if my AMS data is messy?

A book with 40%+ missing mobile numbers or 20%+ blank emails needs data cleanup before the automation layer. That's the honest answer. The audit surfaces this and we'll tell you whether to budget a cleanup sprint before the build. Some agencies do the cleanup themselves after the audit and come back ready.

Does this work for commercial lines?

Yes, with a different cadence and an account-level classifier (see the commercial case study). The four-touch structure extends to 90/60/30/14 days out because commercial renewal conversations require more prep time from the client.

Can I do this without changing AMS?

Yes. Applied Epic, HawkSoft, and EZLynx all support this via their APIs. The only situation that forces a change is if you're on a legacy system without any API access at all, and even then CSV-export workflows bridge the gap.

What happens if I cancel?

Managed Ops is month-to-month. If you cancel, the automation stops sending new messages. The historical interaction data stays in your AMS (we wrote it there). You keep the system architecture — we just stop running it for you.

What to build this quarter if you're starting from zero

You don't need the whole stack on day one. The 30-60-90 we recommend:

• Month 1: Run Touch 1 and Touch 2 via AI-drafted email for your top 200 renewals. Measure. Learn the rhythm.
• Month 2: Add Touch 3 (multi-channel) and the producer hand-off queue. Start running the classifier to rank the call list.
• Month 3: Add text as a channel, extend to the full book, and wire the write-back to your AMS so producers see the full interaction history.

At that point you have a real system, not a marketing gimmick. Agencies we've walked through this tend to see the retention lift land in month 4-5 as the first full cohort of four-touch renewals completes.

Where this breaks (and how to avoid it)

Three failure modes we see most:

1. Tone drift. The system starts drafting in "email marketing" voice instead of "your voice." Fix by having a producer review the first 50 messages weekly for the first month, then spot-check.
2. AMS data staleness. Garbage in, garbage out. If the AMS says the phone number is still the office number from the client's previous job in 2019, you're sending to dead contacts. Budget 1-2 weeks of data cleanup before you scale.
3. No kill switch for upset clients. The single worst outcome is a client who asked to stop getting emails getting another one. Build an explicit opt-out table that every send checks against. Every send.

Next step

If you want the cadence templates, the classifier prompt, and the exact AMS field mapping we use — that's the core of the AI for Agent Retention course, or we build and run the whole thing as a service via Build & Launch. If you want to see what this would look like for your specific book before deciding, a free audit surfaces the retention gap with the actual numbers from your agency.

Renewal automation isn't a new idea. What's new is that the AI is finally good enough to personalize at the per-policy level — which is the only version of this that actually lifts retention. Until you've built that, you don't have a retention system, you have a mail merge.