Oriental Nippon Development
Back to blog
Insights

Process First, Software Second: Lessons from Replacing Legacy Operations

Restaurants, maintenance crews, delivery fleets, a debt-recovery business stuck since 2013. The software differed every time. The order of work never did.

We have replaced legacy operations for a restaurant chain, a fuel-network maintenance contractor, a delivery fleet and a debt-recovery firm whose workflows hadn't changed since 2013. The software was different every time. The order of work that made them succeed was the same.

Four projects, one sequence

ProjectBeforeThe connected chain we built
Restaurant chain6 separate toolsOrder → kitchen → recipe stock → purchasing → ledger
Maintenance contractor4 disconnected stepsWork order → parts → cost → invoice
Delivery fleetWeekly paper planningTour plan → driver app → live record
Debt-recovery firmSuite fixed since 2013Intake → allocation → agent work → reporting

1. Map the process with the people who run it

Not head office alone. Branch and kitchen staff for the restaurants. Ride-alongs with drivers for the fleet. The people doing the work know where the double handling is, and they know which "requirements" are really habits the old system forced on them.

2. Make the process the design object

When replacing a legacy suite, the temptation is to rebuild its screens. Don't. Design around the work, such as account handling, work allocation or dispatch-to-cash, and let the screens follow.

3. Connect the chain end to end

The win is rarely one feature. It is that completing a job moves the stock, books the cost and raises the invoice, and that selling a dish depletes its recipe and posts the ledger. One flow, not six integrations.

For the restaurant chain, the most important design sessions happened with the chefs, not in head office.

4. Pilot for a full cycle

A pilot branch for a full week, including a physical stock count. One region through a full billing cycle. One depot for a week. Short pilots hide month-end problems.

5. Keep the old system until the new one is proven

Functional areas are signed off one at a time, with the legacy system still running beside them. Rollback should be boring.

What the timeline looks like

This is the restaurant chain's plan. Notice that the menu and recipe phase, which is process work with the chefs, is the longest fixed phase:

Delivery timelineIndicative total: 10–14 weeks
  1. 01Fit review

    1–2 wks

    Menu and recipe scope, branch topology, hardware check

  2. 02Foundation

    2 wks

    Entities, accounts, tax, environments

  3. 03Menu and recipes

    3–5 wks

    Recipe structures, yields, waste rates

  4. 04Channel wiring

    2–3 wks

    Storefront, POS, payments, kitchen routing

  5. 05Pilot branch

    2 wks

    One branch live through a full cycle and a stock count

  6. 06Roll-out

    ongoing

    Branch by branch

Questions to ask before you buy any software

  • Who touches each transaction, and how many times?
  • Where is information typed in a second time?
  • Which reports are reconstructed by hand at month end?
  • What would a full-cycle pilot look like for one branch, depot or region?
  • How would you roll back if the pilot failed?

The result

Accounting becomes a consequence of operations instead of a monthly reconstruction. Read how this played out in our collections platform replacement.

Sources and further reading

  1. Strangler fig application, Martin Fowler
  2. Odoo 17 documentation, Odoo

Related case study

Replacing a 2013-Era Collections Suite with a Multi-Tenant Operations Platform