The AI retention manager for Australian pest control

What the Retention Manager does

• Working the back catalogue, by text and email

  Lapsed customers are found, checked against the ladder and messaged — with the drafts waiting for you until you decide that sequence has earned auto-send. Win-back is marketing under the Australian rules, so it needs consent on file for that person, and Hunter shows you exactly who it held back rather than quietly sending anyway. Calling the back catalogue exists in the product but is not sold on this site: it needs a person to approve every call, refuses any number with no relationship on file, and is not published until it has been verified against a real Australian number.

  What we do not claim

  The offer originally promised: An autonomous outbound calling agent working the back catalogue. An AI-voiced commercial call is a telemarketing call under the Australian rules, with no carve-out. Hunter cannot wash a list against the Do Not Call Register, and a runaway routine dialling a thousand people is the failure mode that ends a product.

  The offer originally promised: Batch calling the back catalogue. Batch dialling bypasses quiet hours, do-not-contact, the trust dial and the calling-hours rules — every guard the product exists to enforce.

• Re-treatment reminders that escalate

  Ninety days out, then sixty, then a call on somebody's list — with each rung a step in one sequence rather than three separate reminders that do not know about each other. The last rung is a task for a human, not a call Hunter places on its own.

• Next year's inspection booked at the end of this one

  Recurring templates put the next visit on the planner with the lead time you want, so the annual termite inspection is scheduled work rather than an act of memory. Applying a book-ahead recommendation moves only the day the job appears on your planner: the visit date does not move, nobody is contacted, and nothing is invoiced.

• Anniversary campaigns that keep running

  A campaign is a standing audience rule rather than a one-off blast, so a customer who reaches their anniversary next March is enrolled next March without anyone building a list. It reports what was sent, what was drafted for approval, what was held back and why, what replied and what converted.

• Between-visit comms about what is actually in season

  A seasonal calendar with five broad activity windows — termite season, spring crawlers, bird nesting, summer flyers, rodents — by climate zone, so the message goes out when the pest is actually about. The copy talks about what to look for and what to do. It names no chemical, no rate and no standard, and it says the prompt comes from a calendar rather than from anything observed at their property, because telling a customer their house may have termites when nobody has looked is a claim nobody can defend.

• The post-visit summary, with the findings and the photos

  A private link carrying what was done, what was found, what is recommended, the photos and any warranty. It is transactional and stays that way: nothing may append an offer, an upsell or a review request to it. That is not caution for its own sake — one promotional sentence makes an entire message commercial under the spam rules, and the penalties for exactly that mistake have been large and repeated.

  What we do not claim

  The offer originally promised: A service summary with the photos, emailed on completion. Hunter's outbound email does not carry images, deliberately — an allowlist that admits image tags is an allowlist that admits tracking pixels.

• The review request, with the direct link, to everyone

  Everyone is asked, unconditionally — there is no sentiment argument to the request and no predicate that could become one. Your review link is the one you copied out of your own business profile and pasted in; Hunter stores it exactly as pasted and never constructs or guesses a URL. Reviews that come in are recorded by hand, so the technician scorecard's review figure is real rather than a count of requests sent.

• You hear about a bad visit before the internet does

  Separately from the review request, and equally unconditionally, every customer is asked how the visit went. A poor answer raises a task with the job's photos, chemicals and checklist already attached, so somebody can actually fix it. If you are mid-repair you can hold that customer's review request for at most a fortnight against a named open problem — and then it sends anyway. Deferring during a repair is legitimate; never asking is not, and the difference is whether the request eventually goes to everybody.

  What we do not claim

  The offer originally promised: Pre-review feedback capture that intercepts unhappy customers first. Selectively soliciting positive reviews is against the review platforms' own policy and has been litigated here. The penalty lands on the operator's business profile — the lead flow this product exists to protect.

• Churn risk, scored from eight things that actually happened

  Unresolved payment failures, an overdue balance, days past that customer's own observed visit cadence, cancelled visits, open complaints, callbacks, an agreement ending, no payment method on file. Every factor's contribution is shown, so the score survives the question 'why is this one flagged?' in front of somebody. It is observation, not prediction: there is no trained model, no accuracy figure and none is published. Hunter needs three completed visits before it will claim a customer has a cadence at all, and with fewer it switches that factor off rather than guessing.

• Dunning that escalates in tone and channel

  A polite reminder, then a firmer one, then a text, then a task — stopping the moment the invoice is paid. It refuses to chase an invoice whose balance is covered by a payment already in flight, which matters more than it sounds: a direct debit takes a couple of business days to land, and chasing somebody who has already paid is how a good customer becomes a former one.

• Failed payments retried, expiring cards replaced before they fail

  A failed payment is retried on a schedule and raised if it keeps failing. A card about to expire is dealt with two months out, before it declines in front of a customer. A bank debit has no expiry, and Hunter deliberately never counts one as expiring — so the claim is that it asks for a new card before the old one fails, not that it keeps your payment details current.

• The pre-purchase buyer becomes a homeowner in six weeks

  Record the settlement date on the job and Hunter follows up after it — which is the single highest-intent moment in this trade and the one nobody has a system for. The buyer who read your timber pest report is now the owner of the property it described.

• A save desk that offers your discounts, not ours

  When a cancellation is requested, Hunter can offer a retention deal — but only something from a catalogue you wrote, and by default it still asks you first. That constraint is the reassurance rather than a limitation: an AI that hands out discounts is the most common objection to this whole category, and the catalogue is the answer to it.

• Plans, card on file, and direct debit

  Monthly or annual plans billed as ordinary invoices charged against a stored authority — card, or Australian bank debit. Every cycle produces a tax invoice the customer can be sent and query, and what they owe lives in one place rather than in a payment provider's parallel universe. A bank debit lands about two business days later, and Hunter shows it as in flight in the meantime rather than pretending the money has arrived.