How to Roll Out IRIS Module by Module Without Disrupting the Plant
DBR77 Team7 min read
On this page
Roll out IRIS in waves: one pilot line, one or two modules, and a short period where the new system runs next to the old way of working. Start with machine data, MES and KPIs, add the next module only when the first is used on every shift and meets measures you agreed in advance, then copy the pattern to the next line. The plant never changes more than it can absorb in one wave.
A rollout that stops shipments loses the floor's trust for years.
Why big-bang rollouts stop plants
A big-bang go-live switches production, maintenance, quality and the warehouse to a new system on the same day. Every department learns new screens at once, a data error in one module spreads to the others, and there is no old process left to fall back on.
In McKinsey's 2018 survey of more than 700 manufacturers, less than 30% had rolled out solutions company-wide. The authors warn against treating the effort as an isolated IT implementation and advise starting from bottom-line value, not from the technology. A wave plan does both: each wave is chosen for a loss it should reduce, and each is owned by operations, not only by IT.
A plant with three shifts and six lines starts wave 1 on line 2.
How to choose the pilot line
The pilot line should teach you something that transfers. Five criteria:
- Representative, not the worst. Pick a line with typical products and machines. A line in crisis will hide what the system does.
- A clear pain. Unrecorded stops, retyped work orders or late material, something people already complain about.
- A willing shift lead on each shift. One skeptical shift can stall a pilot.
- A measured baseline. Record two to four weeks of current numbers before go-live, even if they come from paper.
- A stable way of working. The Lean Enterprise Institute calls standardized work a baseline for improvement activities. If the line has no agreed way of working, write it down first, or you will not know what the system changed.
A module order that keeps the plant running
Each wave should change one main habit on the floor. The order below suits most mid-size plants; a plant whose main loss is in the warehouse may bring WMS forward.
| Wave | Modules | What changes on the floor | Go criteria for the next wave |
|---|---|---|---|
| 1 | Machine data, IRIS MES, IRIS KPI | Stops and counts are recorded live, orders run in the system | Stop reasons stable for three weeks, counts match the line's real output |
| 2 | IRIS CMMS, IRIS Tasks | Stops above a limit open work orders and tasks with an owner | Most tasks closed on time, no parallel whiteboard |
| 3 | IRIS QMS, IRIS WMS | Inspections and material moves run on the same order and lot data | Inspections done in the system, picks confirmed by scan |
| 4 | IRIS APS, IRIS MRP | The schedule and material plan read live status | Planners use the system plan as the plan of record |
| 5 | AI recommendations, IRIS Gemba | Suggested tasks and priorities reviewed by people in daily meetings | Accept and reject rates stable, every rejection has a reason |
The first 30 days on the machine data side are covered in The First 30 Days of IoT in a Brownfield Factory.
How to go live without a stop
Plan go-live the way you plan a changeover.
- Run in parallel for one to two weeks. The old form stays as a backup. Retire it on a fixed date, or it will never stop.
- Go live on a low-load shift, never in the week of a big customer order or an audit.
- Name a super user on every shift, an operator or setter who knows the screens and answers questions on the spot.
- Train at the line, in short sessions. Ten minutes at the terminal beats two hours in a meeting room.
- Keep a fallback. If the network or a terminal fails, people need a paper form and a rule for entering it later.
- Hold a 15-minute review each day for the first two weeks: what was hard, what was entered wrong, what to change.
Change management weighs heavily on whether a wave works. Prosci's research with more than 2,600 change practitioners found that 88% of projects with excellent change management met or exceeded objectives, against 13% with poor change management.
Illustrative example: a plant with three shifts and six lines starts wave 1 on line 2. Weeks 1 and 2 record a baseline on paper. Week 3 connects the machines and runs orders in MES next to the old shift sheet. In week 5 the shift sheet is retired. Wave 2 starts in week 8, after stop reasons have stayed stable for three weeks. Line 4 copies the setup in week 10.
Success measures for each wave
Measure use before results. A module nobody uses on the night shift changes nothing.
- Use: share of orders, stops or work orders recorded in the system on every shift.
- Data quality: share of downtime minutes with a known reason, counts that match the line's real output.
- Double entry removed: number of forms, whiteboards or spreadsheets retired.
- Loop speed: time from a stop to a response, and from a task to its closure.
- The target loss: the one problem the wave was chosen for, against the baseline.
Write the go criteria down before the wave starts. How to Review and Scale IRIS After the First 90 Days explains how to review the results after the first quarter and decide where to scale.
How this works in DBR77 IRIS
DBR77 IRIS is sold by module. A plant can start with one module and add more at any time, and data and configurations carry over. All modules share one data model, so a new module reads the asset list, orders and shifts already set up in wave 1 instead of needing a new interface. DBR77 also offers a guided pilot program that sets up IRIS on a single production line, so the first wave runs on your own data.
AI can wait until the base is in use. When you add it, start with the IRIS assistant, which only reads data, and with Data & AI recommendations that a person applies or rejects one by one. For how to connect existing systems along the way, and which ones to keep, see How to Unify MES, WMS, QMS and CMMS Without Replacing Everything.
FAQ
How long does one wave take?
On one line, often six to ten weeks: baseline, go-live, parallel run, and a few stable weeks before the go criteria are checked. Later lines go faster because the setup is copied.
Do we need to stop production to go live?
No. Go live on a low-load shift, run in parallel with the old forms, and keep a paper fallback.
Who should lead the rollout?
Operations, with IT. The plant director owns the target losses and the go criteria, and the project manager runs the plan.
When should we switch on AI?
After the modules it reads from are used on every shift and the data is stable. Start with suggestions that a person approves.
Conclusion
Before wave 1, pick a representative pilot line, record two to four weeks of baseline and write down the go criteria. Name a super user on every shift, run the new system next to the old forms for one to two weeks, and retire the old forms on a fixed date. Start the next wave, or copy the setup to the next line, only when the go criteria are met; if they are not, fix the wave on the pilot line first.
Sources
- McKinsey & Company, How digital manufacturing can escape "pilot purgatory", 2018
- Prosci, The Correlation Between Change Management and Project Success
- Lean Enterprise Institute, Standardized Work