The project

Structure

Multi-level car park, structural steel

Steel ordered

1,850.29 MT across 11 approved phase lists

Distinct part marks

1,174

Period measured

28 March – 20 August 2026 · 4.8 months

Team on the system

12 — eight site engineers, two admin, one executive, one production floor

What went through it

These are the denominators. They are published so the derived figure further down can be recalculated rather than taken on trust.

Crew-days recorded

122

Deliveries received

103

Asset-days of machinery logged

281

Days with erection logged

92

Submissions approved

542 of 556

What changed

Each line carries the identifier of the claim behind it, so it can be traced to the measurement that produced it.

1. C-09

   Site log to approved record

   Under an hour, median across 642 submissions. The process it replaces — a hand-compiled report reaching the office — was measured in days.

2. C-10

   Bad entries stopped at capture

   20 in the first three weeks of counting. Three were deliveries being received against the wrong vehicle.

3. C-11

   Double-counted steel suppressed

   42.36 MT that the revision filter kept out of a bill.

4. C-13

   Planned versus actual

   Now compiled every day. Previously it was rarely or never done — a capability that did not exist, rather than a faster version of one.

5. C-14

   Locating a part mark

   "Where is this mark, and is it up yet?" was often unanswerable without walking the site. It is now a search.

6. C-07

   Time returned

   Roughly 302 hours over 4.8 months — about 1.5 working weeks of engineer and project-manager time per month, on this one project.

What this does not show

Listed here rather than left to be discovered, because a case study that reads as flawless is read as marketing.

1. 01This is one project. Nothing here demonstrates behaviour across a portfolio, because there has not been one.
2. 02The time figure is derived, not stopwatched: measured volumes multiplied by the operator’s own confirmed timings for their prior process. The assumptions are published alongside it.
3. 03Three workflows contribute zero hours on purpose — planned-versus-actual, part-mark lookup and delay evidence. Nothing is saved by speeding up a task nobody performed by hand; those are capability claims instead.
4. 04Dispatch was not run through the system on this project. The site-receipt half of that workflow is heavily used; the factory-dispatch half is proven in testing, not in the field.
5. 05Figures counted out of an audit trail are current, not lifetime — a row disappears with the entity it describes.