Reserve roughly 8–10 elapsed weeks for the controlled pilot described here, with overlapping stages and approximately 60–120 internal hours. These are bottom-up planning allowances, not market averages. A native reminder configuration may require substantially less work; a bespoke voice-and-scheduling integration may require more.
| Stage | Suggested timing | Internal effort allowance | Main owner | Required output |
|---|---|---|---|---|
| 0. Define the problem | Start of week 1 | 2–4 hours | Owner/MD | One-page pilot charter |
| 1. Measure the baseline | Weeks 1–2 | 8–12 hours | Office manager and finance | Baseline workbook |
| 2. Map rules and clean records | Weeks 2–3 | 8–16 hours | Service manager | Workflow map and exception register |
| 3. Check software, permissions and privacy | Weeks 2–4 | 8–16 hours | Systems owner and advisers | Capability and approval pack |
| 4. Configure or build | Weeks 3–5 | 12–30 hours | Internal implementer | Restricted test workflow |
| 5. Test and rehearse failure | Weeks 5–6 | 10–20 hours | Office manager and technical lead | Test evidence and launch decision |
| 6. Run a limited pilot | Weeks 7–10 | 8–16 hours | Process owner | Outcome and exception log |
| 7. Decide and hand over | End of pilot | 4–8 hours | Owner and finance | Continue, revise or stop decision |
These allowances exclude a major data migration, prolonged supplier approval, substantial legal work and new infrastructure deployment. For installation quotations, keep following the cohort after the pilot if jobs have not yet reached completion; do not present accepted quotes as realised contribution.
Who should you contact before buying anything?
| Contact | What to ask for | Evidence to retain |
|---|---|---|
| Existing job-software support or customer-success team | A demonstration of the exact workflow in your subscription tier; available APIs; restrictions; additional fees | Written capability confirmation and relevant documentation |
| Telephone provider or telecoms administrator | Conditional diversion, caller-ID behaviour, transfer charging, outage fallback and reversal procedure | Routing diagram and a test-call record |
| Office/service manager | The awkward cases, customer exceptions and current workarounds | Approved rules and exception list |
| Competent technical/safety lead | Approved emergency routing and prohibited advice | Versioned escalation procedure |
| Privacy adviser or responsible internal lead | Data flows, lawful basis, recording/transcription, marketing classification and DPIA screening | Documented review and unresolved issues |
| MSP or developer, if necessary | Restricted credentials, validation, monitoring, recovery and support cover | Architecture review and operating runbook |
| Accountant or finance controller | Contribution definition, internal time valuation, tax treatment and cost comparison | Cost model with assumptions |
| Human answering provider | The same enquiry-handling brief used for AI | Like-for-like quote and exclusions |
Send the workflow description before requesting a demonstration. Ask suppliers to show your difficult cases, not their preferred scenario.