TG Guest Talk
Migration playbook · Source system: legacy on-premise

Migrating from a legacy on-premise PMS — the on-property switch.

This is the playbook the Talkguest migration team runs when a Portuguese property moves off an on-premise system — Oracle Opera, Protel Air's on-prem instances, Amadeus PMS, or a custom Windows application running on a desktop behind reception. Twenty-one to twenty-eight working days, one migration engineer, one on-property visit, and a physical retirement of the old server.

Who this playbook is for

The shape of the property, the shape of the system, the shape of the migration

This playbook is written for a Portuguese property still on an on-premise PMS in 2026. The typical set-up is a Windows-based application from Oracle Opera, Protel, Amadeus or a boutique local vendor, installed on a desktop or a small server behind the reception counter, with a printer wired directly to the certified factura sequence, a separate channel manager licensed by month, and a WhatsApp workflow that lives on a receptionist's personal phone. The estate is stable, the property knows it, and the switch is a real project — not a Saturday-afternoon subscription swap.

Because the system runs on-property, the migration includes an on-property visit by the migration engineer during cutover week. This is included in the platform price — there is no travel line billed. The visit is typically two full days: one to run the data-export session and one to be physically present for go-live. The physical server or desktop is not touched until forty-eight hours after go-live, at which point it is powered down, archived to a read-only virtual machine at Talkguest's data centre, and the physical hardware is either handed back to the property or, if the property prefers, removed and recycled by a certified vendor.

Week-by-week

Four weeks, in order

  1. Week 1, day 1 — Kick-off, remote. Migration engineer, general manager, accountant, and if possible the IT contractor who maintains the on-premise system. Current-state stack documented: PMS vendor and version, database engine, on-premise integrations (payment terminal, key-card system, phone-billing pass-through, POS), factura sequence, VAT profiles, WhatsApp arrangement.
  2. Week 1, day 2 — Remote database access set up. The migration engineer establishes a read-only remote connection to the on-premise database, typically over a VPN with two-factor authentication or an SSH tunnel to a jump host. If the on-premise system does not permit remote access, day 2 slips to the on-property visit in week three and the timing extends.
  3. Week 1, day 3 — Data schema review. The legacy system's data model is documented — table names, key relationships, custom fields the property has added over the years. This is the day that turns a legacy migration from art into engineering.
  4. Week 1, day 4 — Talkguest tenant provisioned. Property record created, users invited, rooms and room types mapped, VAT profiles configured, factura series carried from AT registration.
  5. Week 1, day 5 — Historic data extraction, dry run. A full read of the historic guest CRM and 24-month reservation history is pulled from the legacy database and loaded into Talkguest as a staging dataset. Data quality issues surfaced.
  6. Week 2, day 6 — Data cleaning pass. Guest profiles de-duplicated, guest tags normalised, rate plans cleaned up. Migration engineer and general manager work through the discrepancy list one row at a time.
  7. Week 2, day 7 — Reservation load, historic. The cleaned historic reservations loaded into Talkguest read-only.
  8. Week 2, day 8 — Reservation load, upcoming. Every upcoming reservation loaded and cross-checked against the legacy system.
  9. Week 2, day 9 — OTA re-mapping. Booking, Expedia, Airbnb and any other active OTAs re-mapped in Talkguest's channel manager. If the property is on an external channel manager, the property's OTA property IDs are moved to Talkguest.
  10. Week 2, day 10 — WhatsApp setup. WhatsApp Business API provisioned or ported. Templates loaded end-to-end tested.
  11. Week 3, day 11 — Cross-check pass one. Every reservation, every folio, every deposit verified.
  12. Week 3, day 12 — Migration engineer on property. Physical visit to the property. Payment terminal, key-card system, phone-billing pass-through and POS integrations tested end-to-end against Talkguest.
  13. Week 3, day 13 — Cross-check pass two, on property. Every upcoming reservation, every deposit, every open folio walked through with the reception team. Final go/no-go decision at end of day.
  14. Week 3, day 14 — Channel-manager cutover, afternoon. OTAs switched from the legacy channel manager to Talkguest. Test booking placed and traced.
  15. Week 3, day 15 — Go-live, morning. Talkguest becomes master of record. Legacy system moves to read-only. Migration engineer physically on the property from morning through end of day.
  16. Week 4, day 16 to 20 — First week live. Reception, housekeeping and reservations run entirely on Talkguest. Migration engineer on-call remotely.
  17. Week 4, day 21 — Legacy server retirement. The on-premise server or desktop is powered down. Data is archived to a read-only VM at Talkguest's data centre for the statutory retention period. The physical hardware is either handed back or recycled by a certified vendor.
Legacy-specific gotchas

Four things that are unique to on-premise migrations

The database schema is undocumented. Legacy systems typically ship with a documented core schema and years of ad-hoc additions on top. Custom fields, ad-hoc tables, and stored procedures the property added over the years are usually not documented. The day-3 schema review takes longer for a legacy source than for a cloud source, and the migration engineer sometimes finds himself reverse-engineering a stored procedure to understand what a field actually represents.

The payment terminal is wired to a specific PSP. On-premise PMSs are frequently coupled to a specific payment terminal and a specific PSP through a physical wire or a serial connection. Cutover to Talkguest means moving payments to the property's PSP account through Talkguest's PSP integration. The day-12 on-property visit is the moment this transition is tested end-to-end with real cards.

Factura printer configuration. Portuguese hotels running on-premise systems often have a factura printer wired directly to the PMS with the AT-registered series stored in the printer's memory. During cutover the printer is either re-configured to accept factura output from Talkguest, or retired in favour of Talkguest's PDF-and-email factura workflow. Either way is fine — the property decides at day 3.

Key-card system integration. Boutique hotels frequently run a key-card system tightly coupled to the on-premise PMS. Talkguest supports the major key-card vendors (Assa Abloy, Salto, Onity, Dormakaba) through direct integrations. During the on-property visit the migration engineer verifies that check-in on Talkguest triggers key encoding on the property's existing key-card system.

Post-go-live

The two weeks after the server goes off

The migration engineer stays on the account remotely for fourteen working days after go-live and returns to the property for a half-day at day 30 to run a physical review with the reception team and the accountant. During the post-go-live window the typical surprises are practical rather than technical — a receptionist who has been using the legacy system for eight years takes a week to adjust to a new interface; the housekeeping supervisor discovers that the mobile app is easier than the printed round; the accountant queries a SAF-T (PT) line and needs a walk-through of the export.

At day 30 the migration engineer runs the final walk-through, hands the account to Talkguest's standard support queue, and the property is fully self-sufficient. See the Solar do Chiado case study for a real legacy migration, and the pricing page for what running Talkguest costs after go-live. Ready to start? Head to get access.

Schedule your legacy migration