Auto-generated service leads are the rare automation whose value nobody disputes — the argument is always about readiness: "our contract data is a mess", "the advisors won't work a queue". Both are solvable, in order, in about a quarter. This is the playbook.
Stage 0 — make the generators possible (weeks 1–3)
- Contracts must exist as records. Going forward that is automatic — invoicing a contract item creates the contract on the asset. The backlog is the work: load the live warranty and AMC book against assets, even if only with start/end dates and type. Don't wait for perfection; a contract with approximate scope still generates a lead on time.
- Schedules need their rhythm. Define maintenance schedules per model — days, usage hours or readings, in sequence — and set reminder offsets deliberately: too early wastes calls, too late loses the visit to the corner workshop. Start with the OEM book and adjust.
- Readings need a habit. The schedule engine is only as good as the odometer and hour readings feeding it — which technicians capture on every job once capture is a required step, not a favour.
Stage 1 — pilot one generator, one branch (weeks 4–8)
Start with contract-due reminders at a single branch: highest signal, cleanest data, easiest story. Run the campaign action on a weekly rhythm; let the leads land in two named advisors' queues; book straight into capacity-backed slots. Resist the urge to also switch on schedules, campaigns and win-backs simultaneously — a pilot that changes one thing can be measured.
Stage 2 — adoption is queue design (weeks 6–10, overlapping)
Advisors abandon queues that mix gold with noise. Three rules keep the queue worked: cap daily volume to what the branch can actually call; rank by value (contract expiry beats routine reminder); and close the loop visibly — a lead booked into an appointment leaves the queue, a lead called-no-answer schedules its own retry. The advisor's experience must be "this queue makes my target easier", or the pilot dies in week six.
Stage 3 — scale and layer (weeks 9–12)
Add maintenance-schedule leads, then campaign sweeps (lapsed customers, recall-adjacent outreach), branch by branch. Each layer inherits the pilot's queue discipline. This is also where AI earns its place — ranking, drafting outreach, flagging the lead that is also carrying an open complaint — with humans approving anything outbound.
The metrics that prove it
Four numbers, before versus after: service retention (vehicles returning within the due window), lead-to-appointment conversion, advisor outbound productivity (booked appointments per calling hour), and revenue per generated lead. If retention doesn't move in two quarters, your reminder offsets are wrong — tune them; the machinery is fine.
The machinery itself: service contracts & automatic service lead generation; the wider story: the service lead that writes itself.