A water-sector delivery centre: two projects distilled into twelve checklists

The short version: from November, water-sector work is split out of our general delivery system into a dedicated water-sector delivery centre: one sector lead, two implementation consultants and one architect on a standing team, scheduled around the flood season. The centre's first task was not hiring — it was distilling the Yellow River platform and river sand-mining projects into twelve reusable checklists.

Why a separate sector line? Because water projects are not manufacturing projects. A factory system runs inside a walled site: when something breaks you call, walk in and look. A water project lives on embankments, at sluices and in dredging zones; construction stops during the flood season, acceptance is usually squeezed in before it, and a problem may require being on site the same day. One delivery process for both makes both worse.

From two projects to one method: what is captured, what is reused Delivered projects Yellow River platform 13 sections · six flood steps three trace channels River sand mining 8 modules, 24 pages one weighing, three ledgers 12 checklists · four groups Discovery 3 · roles, forms, sites Build 4 · data, APIs, cutover Acceptance 3 · tie-out, traces Operations 2 · flood duty What changes on a new project 1 · reviewable plan in 30 days discovery starts from a checklist 2 · narrower quote variance person-days estimated per line item 3 · acceptance pack, first time tie-out sheets are deliverables Checklists are not templates: each carries a pass criterion and a failure path, mapped to the gates in the delivery handbook.
What accumulates is not a slide deck but checklists that can be ticked, signed and reused

1 · Three things that make water projects different

  1. The site is on an embankment, at a sluice, in a dredging zone. Devices are scattered, networks are unreliable and some points have no mains power. Design must assume disconnection and offline back-fill — otherwise crews revert to paper within a week, which is exactly what the sand-mining project showed us.
  2. Acceptance tests whether the numbers are right, not whether the screens look good. Clients reconcile tonnage against sales records, check that contract balances deducted correctly, compare water-level curves with field readings. A beautiful interface with numbers that do not tie out fails.
  3. Inspections look for traces. When supervisors or auditors ask "who did what, when", every record needs operator, timestamp, location and photo — and the timestamp must be written by the server, never accepted from the client.

2 · What the twelve checklists cover

GroupChecklistThe problem it settles
Discovery (3)Roles and permissions matrixWho may view, edit and export — fixed before work starts
Document flow mapHow many posts one form passes, and where it queues
Monitoring point inventoryHow many cameras, level sensors and weighbridges, and where
Build (4)Data definition sheetTonnes, money and volume — whose formula governs
Interface agreementField contracts with hydrology, video and barrier systems
Cutover planHow long old and new systems run in parallel, and the rollback trigger
Training matrixWhich role trains for how long, and who is assessed
Acceptance (3)Data tie-out sheetSampling ratio between system figures and source documents
Trace checklistAre the required fields present on every record
Drill recordPower loss, network loss, device failure — rehearsed or not
Operations (2)Flood-season duty rosterWho is on duty, how often patrols run, who gets the alert
Device asset registerModel, install date, warranty, replacement history

3 · How the centre is staffed

We reuse the eight-person project group, with one substitution: the business consultant becomes a sector consultant who knows water level, discharge, weigh tickets and mining boundaries — otherwise they cannot ask the right questions. A typical team is one consultant throughout, one architect on design and critical modules, two back-end and one front-end developers, one on-site implementation consultant, and half a tester plus half an operations engineer.

The real difference is scheduling: water projects are planned backwards from the flood season. Everything that can go live must be live, or at least rehearsed, before it starts; during the season only duty and small fixes; the second phase waits until after. We learned this twice the hard way — going live in flood season hands the risk to the one month that cannot absorb it.

4 · What it means for a client

  • Faster discovery. Arriving with the checklists means roles, documents and sites are covered in the first meeting; a reviewable plan within 30 days.
  • Tighter quotes. Person-days estimated per checklist item instead of "this feels like a four-month job"; anything outside the list is explicitly marked as excluded.
  • Smoother acceptance. The tie-out sheet and trace checklist are themselves deliverables; ticking them beats "looks about right".
  • Coverage in the flood season. The duty roster is written into the operations contract, with response times and named contacts.

Planning a similar production line or enterprise system?

Start with our pricing logic, then let's talk requirements — cost structure, technology comparison and acceptance criteria, all laid out in the open.