Where the margin actually went

00Summary of findings

Redbud reports a 38% contribution margin. That figure is calculated against a standard cost roll dated January 2026. Steel has risen 11.1% since that roll was set.

Restating material at replacement cost puts actual contribution margin at 34%. Redbud has been quoting, and congratulating itself, on four points of margin that no longer exist. On $38.4 million of revenue that gap is $1.53 million a year.

The second finding is more useful than the first. The shortfall is not spread evenly. Work quoted before December 2025 earns 29.8%; work quoted since earns 40.4% — a 10.6-point gap that has nothing to do with customer, volume or negotiation, and everything to do with when the price was last set.

Forty part numbers out of 640 — 6% of the catalogue — carry 47% of the shortfall. Repricing those forty recovers $1.29 million a year and touches 27% of revenue. Getting the same result with an across-the-board increase would take 7.2% on every part and every customer Redbud has.

01The question

The March Constraint Analysis valued lost throughput using Redbud's stated 38% contribution margin, and flagged the figure as untested: it came from a standard cost roll that predated a sharp move in steel. The July Market Analysis then established that the whole industry is under-repriced — fabricated metal output prices rose 5.3% over the past year against 14.2% steel inflation — and recommended requoting the cost roll before pursuing new work.

This analysis is that requote, extended into the question it raises:

The emphasis is on smallest. Every price increase spends customer goodwill. The object is to spend as little of it as possible.

02Method and data

Twelve months of closed orders were extracted at line level — 2,174 orders across 640 active part numbers — with revenue, quoted price, material issued, and labor hours booked against each. Quote dates came from the quoting system and were matched to orders by part number and revision.

Material was then revalued two ways: at the January 2026 standard cost roll, which is what Redbud's reporting uses, and at replacement cost as of June 2026. The gap between those two numbers is the whole of section 03.

Replacement cost movement is not estimated. It is taken from the Bureau of Labor Statistics producer price indexes for iron and steel (WPU101) and nonferrous metals (WPU102), which are public, monthly, and free.

Labor and overhead were held at standard throughout. Both have moved, but far less than metal, and holding them constant keeps this analysis about the thing that actually changed.

03What the cost roll misses

Redbud's standard cost roll was set in January 2026. Since then:

Applied to Redbud's material mix, that understates cost of goods by a specific and calculable amount:

Two things follow, and they pull in opposite directions. Redbud is less profitable than it believes, which is bad. And Redbud has been winning work at prices that looked comfortable and were not, which means the pricing lever has more room in it than the sales team currently thinks — because competitors are quoting off the same stale assumptions.

A note on method: this top-down restatement puts actual margin at 34.0%. The bottom-up rebuild from 2,174 individual orders in section 04 puts it at 34.06%. Two independent routes to the same number within a sixteenth of a point is the check worth running — if they had disagreed by two points, everything after this section would be unreliable.

04The spread

A 34% average is not a useful management number, because almost none of Redbud's work earns 34%. Revenue by realized margin band, with each part colored by when it was last quoted:

The distribution runs from 10% to 55%, with the tenth and ninetieth percentiles 20.1 points apart. That spread is the finding. A business with a 34% average and a 20-point spread is not one business — it is a portfolio of good work subsidising work that should have been requoted a year ago.

Only $79,743 of revenue — two tenths of one percent — sits below a 15% margin. This is not a business giving work away. It is a business whose prices are, in specific and identifiable places, roughly a year out of date.

05The mechanism

The colors in the chart above are the answer. Splitting the catalogue by when each part was last quoted:

This is worth stating plainly because it changes who owns the fix. Nothing here suggests the estimators are pricing badly or that salespeople are discounting too freely. The failure is that Redbud has no process which ever revisits the price of an active part. Work quoted in a low-steel period keeps its low-steel price indefinitely, and the longer a part runs, the more out of date it becomes.

It also explains why the problem is invisible in the monthly reporting. Blended margin drifts down a fraction of a point at a time as stale parts take a larger share of the mix. No month ever looks alarming.

06Where the money actually is

Bringing every stale part up to the 40.4% that current quoting achieves would be worth $2,782,215 a year. That is the theoretical maximum, and it is not the recommendation — some of those parts are priced where they are for competitive reasons that a spreadsheet cannot see.

The practical question is how much of that $2.78 million sits in how few parts.

Forty is the recommended stopping point, and the reason is the shape of the curve rather than the size of the number. Between 40 and 60 parts, each additional part repriced returns progressively less while adding another customer conversation. The curve has flattened by then; the cheap recovery is already taken.

07What repricing forty parts does

The across-the-board column recovers more money on paper. It is still the worse option, for a reason that does not appear in the table: it raises prices on the 40% of revenue already earning 40.4%, which is work Redbud is winning at healthy margin and would rather not put in play. A 7.2% increase on a well-priced part invites the customer to requote it elsewhere. That risk does not show up in a margin calculation and is the most likely way this exercise destroys value rather than creating it.

The targeted increase averages 12.5%, which is higher per part and will be noticed. It is defensible in a way a blanket increase is not: these are parts whose prices were set before an 11.1% steel move, and the increase can be explained with a published index the customer can look up themselves. That conversation is much easier than justifying why a part quoted last month also went up.

08How to challenge this

09Recommendation

The last row is the one that matters in two years. Everything above it is a correction; that row is the change that prevents the next correction. A standing list of parts whose price has not been reviewed in nine months costs nothing to produce and would have caught this in March.

10Scope and fee

Against $1.29 million of recovered annual margin, the full engagement returns roughly sixty times its cost in the first year. That ratio is unusually high and is a function of how long the problem went unnoticed rather than any cleverness in the analysis — the arithmetic here is not difficult. Nobody had run it.