Ordering framing lumber by hand vs. the takeoff tool's cutting-stock order

Once the cut list exists, you still have to decide which board lengths to buy and which pieces to cut from each. The usual hand method is first-fit-decreasing: sort the pieces longest first, put each in the first board with room. This page sets that against the tool's order (an exact dynamic program on small jobs, an LP-rounded search with a printed lower bound on whole houses), from bench.mjs in the tool's repository (run 2026-11-05, seeded and reproducible).

Get the cheapest order for your walls

On the sample house the gap is a few percent, and it is real money

The sample house is the invented example in the tool: 345 pieces, priced at 1 per linear foot of board so cost equals lumber bought. Cost is in those units.

Copies of the houseShipped orderLower boundFirst-fit-decreasingOne board per piece
1 (253 boards)2384.162384.122450.00 (+2.8%)2915.04 (+22.3%)
2 (504 boards)4768.884768.234901.72 (+2.8%)5830.08 (+22.3%)
4 (1004 boards)9532.609531.459786.00 (+2.7%)11660.16 (+22.3%)

When board prices are not proportional to length, the hand method falls behind

Lumber is rarely priced per foot: a 16 ft stud can cost more or less than two 8 footers. On 150 seeded small cases per row, where an exact solver finishes:

PricesShipped equals the cheapestFirst-fit-decreasing equals the cheapestMean excess, FFD
Flat per foot100%64.7%1.89%
Priced by length100%29.3%9.50%

The shipped column is 100% because the tool uses the exact answer whenever the job is small enough for the solver; on whole houses it prints the gap to the lower bound (0.002% to 0.014% above on the table before).

What this does not show

One invented house and seeded random cases, not a builder's real order. Prices are the benchmark's, not a lumberyard's quote, and real yards sell only some lengths and bundle by the unit. A framer who eyeballs the offcuts as they go can beat plain first-fit-decreasing, and the order assumes you cut exactly to the list. The tool does not check the lumber against any span table or code.

Common questions

Is first-fit-decreasing, the usual hand method, good enough for a lumber order?

It is close on a whole house, not exact. On the sample house (253 boards, price 1 per linear foot) it cost 2.8% more than the shipped order, which sits 0.002% above its lower bound. On 150 small flat-priced cases it matched the exact cheapest order 64.7% of the time (mean excess 1.89%); with length-dependent prices it matched 29.3% of the time (mean excess 9.50%).

What if I cut every piece from its own board?

On the sample house that costs 2915.04 against 2384.16, 22.3% more, because short pieces never come out of the offcuts of long boards.

How is the shipped order checked?

Against an exact solver on 120 seeded small cases it matched the cheapest order every time the solver finished, and on whole houses it is within 0.02% of the linear-programming lower bound. It also prints that bound so you can see the gap.

Can I trust these numbers for my own house?

Only as a guide. They come from seeded random cases and one invented sample house, not from a builder's real order or a lumberyard's real price list. A real plan with the actual order would settle it.

Run the takeoff on your own walls

Related: the per-foot rule of thumb, measured and how to count the studs for a wall by hand.