Insurance agency renewal automation: a 90/60/30/15 cadence that protects retention

Most independent agencies do not lose renewals because nobody cares. They lose hours because the work starts when the carrier notice is already loud. Documents are requested in a panic. Producers stack conversations into the last two weeks. CSRs cannot tell which high-value account has had no activity. A cadence is a way to put dates on that work before it becomes a scramble.

A cadence is not a retention guarantee. Clients still shop. Markets still move. Files still have open claims or nonpay issues that change the path. The operating win is earlier visibility: owners see what is unowned, what is waiting on documents, and what still needs a licensed conversation.

Scrambles hide E&O risk as well as service risk. Rushed messages skip facts. Rushed checklists miss last year’s driver. Rushed producers skip notes. Spreading the same work across 90, 60, 30, and 15 days does not make the agency slower. It makes the work eligible for a human review instead of a Friday-night guess.

If the book is small, you still need dates. Memory does not scale across vacation, a busy storm week, or a producer who also owns intake. The cadence is a staffing plan the AMS or CRM can display.

Ninety days is account review. Confirm the renewal date, the owner, the product, and whether the file looks complete enough to start. Create the internal brief. Do not blast the client with every possible form on day one. Decide what this account actually needs.

Sixty days is documents and marketing prep. Request loss runs, driver lists, vehicle or property updates, and any inspections the agency already knows are required. Start carrier or market prep the way the agency already works. AI can draft the document request. A person confirms it is the right list.

Thirty days is the client conversation. A licensed producer reviews coverage, answers questions, and decides what advice is appropriate. Automation can schedule the appointment task and draft an administrative reminder. It should not hold the coverage conversation.

Fifteen days is close-out. Outstanding items, last administrative reminders, bind or rewrite logistics the agency already uses, and a clear disposition: in progress, waiting on client, waiting on market, or not proceeding. If the file is still silent at 15 days, it should be at the top of the owner list, not discovered at expiration.

Stage the asks so clients can complete them. At 90 days, list what you will need and gather what is already in the file. At 60 days, send the named outstanding items. At 30 days, confirm what is still missing before the producer meeting. At 15 days, chase only what still blocks a human decision.

Keep checklists by product. Personal auto, homeowners, commercial auto, and packages do not share one “send everything” paragraph. Named items with a short reason tied to the renewal review get better responses than a homework dump.

Mark received only after a person confirms the document is usable. A blurry photo of a declarations page is not a completed 60-day item. Extraction tools can flag unreadable pages. They do not get to check the box alone.

If a client cannot produce an item, log that and change the next action. The cadence should flex. It should not keep mailing the same request after a person has already explained the gap to a producer.

Every client-facing renewal message is a draft until a person approves it. Document requests, appointment asks, and administrative reminders can use templates. Coverage recommendations cannot. If the draft mentions limits, exclusions, or “you should,” a licensed reviewer writes or approves that sentence.

Show the reviewer the file context: last activity, outstanding documents, open service work, and who owns the account. Approval without context is how wrong letters go out. Phone-friendly approve-or-edit matters because service work does not only happen at a desktop.

Record what was sent, by whom, and what the client was asked to do. That record is part of the file, not extra bureaucracy. It also stops duplicate outreach when AMS and CRM both think nobody has written yet.

Stop the sequence when the client replies, books, or asks not to be contacted. A cadence that cannot pause will talk past an angry or already-handled account. Pause is a feature.

Report operating facts. Count upcoming renewals in each window. Count unowned files. Count files with no logged activity. Count outstanding document tasks. Count drafts waiting for approval. Those numbers run a weekly meeting. They do not predict who will stay.

Avoid vanity retention percentages built from a new reminder sequence. Retention depends on price, advice, service, and the client. A cleaner cadence can support better work. It cannot honestly be sold as a guaranteed retention lift.

Give producers the same definitions. If “at risk” means no activity inside the agency’s chosen window, everyone should use that meaning. Private owner charts that the service team never sees will not change Thursday behavior.

A useful dashboard is a list, not a scoreboard. Names, dates, owners, and next actions beat a red-yellow-green widget that hides the files.

Keep notes that show what was requested, what was received, what was reviewed, and what remained outstanding when a licensed person spoke with the client. Those notes support the file. They are not a substitute for the coverage conversation.

Do not let automation write advice into the activity log. “Discuss garage ZIP change with producer” is an operating note. “Client is fine without updating vehicles” is a conclusion a person must own. Mix those up and the file becomes hard to defend and hard to service.

If the client declines a meeting or a document, log it in the agency’s normal way and surface the gap on the dashboard. The cadence should not invent a reason around the decline or keep sending as if nothing was said.

Align note standards with how the agency already thinks about E&O: facts, dates, owners, and outstanding items. AI summaries can help a reviewer scan. The source note should remain available.

Late starts with a 90-day label and a 12-day habit. If tasks are created at 90 and ignored until 15, you still have a scramble. The cadence only works if someone works the 90- and 60-day lists.

Bad dates. Midterm changes, rewritten policies, and messy AMS downloads will fire the wrong reminders. Fix the date source before you add more messages. Duplicate policies create duplicate letters.

Advice in templates. A single sentence about limits in the default 30-day email will be sent a hundred times without a coverage review. Strip templates to operations language.

No owner and no stop. Unowned tasks and sequences that ignore replies are how clients hear from the agency twice or never. Both are failure modes. Weekly list review is the correction.

Do not launch 90/60/30/15 on the entire book on day one. Pick one segment: one producer’s personal lines, local commercial auto under a vehicle-count threshold you already use, or accounts with clean renewal dates. A small segment teaches you whether dates, checklists, and templates are true.

Run at least one full cycle of tasks on that segment, even if some accounts are already inside 60 days when you start. You will learn more from real outstanding documents than from a theoretical calendar.

Hold a short weekly review with the people who touch the files. Skip reasons, heavy draft edits, and unmatched documents are the curriculum. Expand only when the team uses the list without a workaround spreadsheet.

If you want a deeper map of how renewals connect to intake, CRM, and reporting, the free 15-minute workflow audit is a first conversation. A paid Agency AI Audit is a written workflow map and build-priority roadmap; public pricing for that deeper audit starts at $250+. Neither is a retention promise.