Seeing margin fade while there is still time to recover it
A monthly margin review that shows each project's forecast margin against tender, and every unapproved variation by age, so directors can act while a project is still running.
The problem
A builder found out about margin fade at project close, after the money was gone. Variations sat in inboxes and site diaries, and nobody had a single list of what was unapproved or how old it was.
What the numbers say
Medical centre fit out has lost the most margin at 3.4 points. Unapproved variations are the main reason margin is not yet recovered.
How FDX uses it
From the data to an approved draft
- 01Connect the data
We bring together the job costing system, the variations register and progress claims, refreshed each week.
- 02Draft, then approve
AI drafts variation letters and tender documents from plans, diaries and contracts. Project managers check and send them.
- 03Keep it private
Commercial data stays inside approved tools, with onshore hosting available, and it is not used to train AI models.
Key numbers on the dashboard
- Active projects8Contract value A$46.2M
- Projects fading over 2 points3 of 8Worst is medical centre fit out, -3.4 pts
- Unapproved variationsA$1.40MA$0.60M older than 60 days
- Forecast margin, all projects8.9%Tender margin 10.6%
Next steps it points to
Variations. Chase the A$0.60M older than 60 days. AI drafts each variation letter from the site diary and the contract, and the project manager signs it off.
Medical centre. Hold a cost to complete review with the site team before the next progress claim.
Safety. SWMS drafts can be prepared with AI help, but a competent person reviews and approves all safety content.
ExampleExample case study with illustrative sample data. Not a real client or real results.
Next exampleLawGetting cash in faster without adding admin to lawyers' days
The dashboard is one part of the work.
It gives the team one trusted view of the numbers. The bigger gain is AI working from that same data to draft the follow ups, with a person approving every step.
When AI drafts from your data, personal details are masked before the model sees them. Onshore hosting is available, and your data is not used for training.