Kayu Lestari — from 40 spreadsheets to one source of truth

Kayu Lestari makes beautiful furniture. Their teak dining sets ship to buyers in Rotterdam, Melbourne, and Osaka. What their customers never saw was how the orders got made: forty spreadsheets, three versions of the truth, and a production planner who kept the real schedule in his head.

The company had grown from a workshop into a 300-person exporter without ever replacing its original tools. Every department had built its own spreadsheet kingdom — sales quoted from one price file, purchasing ordered from another, and the warehouse counted stock that both of them disagreed with. It worked, in the way that things work right up until they don't.

When we mapped the pain points in the first workshop, the team named dozens. They collapsed into four — and every one of them traced back to the same root: no shared, trusted record of what the business actually knew.

We don't configure software in week one. We walked the floor — from log intake to finishing to the loading dock — and shadowed the people whose spreadsheets we were about to retire. Discovery produced a process map with 61 steps, of which 14 turned out to exist only because a spreadsheet needed them.

The most important finding wasn't technical. The planner's "mental schedule" was actually a sophisticated set of routing rules — which machines, which crews, which drying times per wood batch. Our job was not to replace his knowledge but to write it down so the system could use it.

The design principle was ruthless: standard Odoo wherever possible, customization only where the furniture business genuinely differs. We ended with just two custom pieces — a wood-batch drying tracker and an export documentation generator. Everything else is configuration, which means every future Odoo upgrade stays cheap.

The routing rules from discovery became actual MRP routings: 22 work centers, per-batch drying constraints, and crew capacities. The planner reviewed every one of them. It was his knowledge, finally legible to everyone.

We ran the rollout in phases, each one live-tested by the people who would own it. Not everything went to plan — the first inventory migration imported 4,000 duplicate SKUs, and we spent an unglamorous week cleaning master data. We're including that here because pretending implementations are smooth helps nobody.

Go-live was March 2, 2026. Our whole team was on site at 6 a.m. expecting firefighting. By 10 a.m. the most dramatic event had been a forgotten barcode-scanner password. The parallel run in December had done its job: the system held no surprises because everyone had already lived in it for three months.

There's a short video of that morning in our library — including the moment the planner approved his first fully automated schedule.

We report impact at day 90, not day one, because honest numbers need a settled system. All figures below were pulled from Odoo itself and verified with the client's finance team.