Module 6 · Lesson 9
How Much Contingency Should You Keep?
10% for new construction, 15–20% for conversions and hillsides. Why contingency isn't pessimism — it's how experienced builders plan.
Contingency is the money you set aside for the things nobody can see yet. It isn't pessimism, and it isn't padding — it's the statistical honesty of construction.
The benchmarks:
- New construction on a flat lot: 10%. Fresh ground, new materials, few unknowns. Surprises happen (soil conditions, price changes, a design tweak) but they're usually contained.
- Garage conversions and remodels: 15–20%. You're marrying old construction, and old construction keeps secrets — the slab's real condition, what's inside the walls, how the previous owner "fixed" things. Opening walls is an archaeology project.
- Hillsides and complex sites: 15–20%+, driven by soil and foundation unknowns until reports come back.
How contingency behaves in real life: most of it gets spent early — demolition and foundation stages are where the ground and the old building reveal themselves. If you clear foundation and framing with contingency intact, the odds shift strongly in your favor.
Three rules for managing it:
- Keep it separate and silent. It's not the finish-upgrade fund. If it survives to the end, then it buys the better counters — or stays in your pocket.
- Require written approval for every draw against it. Each use gets a change order with a cost, so the money leaves consciously (Module 11 covers change orders).
- No contingency, no start. A budget with zero buffer isn't brave; it's a stalled half-built project waiting to happen. If contingency doesn't fit, the project needs to shrink until it does.
Key takeaways
- 10% contingency for flat-lot new construction; 15–20% for conversions, remodels, and hillsides.
- Most contingency gets spent at demolition and foundation stage — surviving those with buffer intact is a great sign.
- Contingency is not the upgrade fund; every draw deserves a written change order.
- If the budget can't fit contingency, shrink the project — never delete the buffer.