Research & Development · Roadmap step 13
You are not selling fabrication. You are reselling steel.
Prepared for Redbud Fabrication, Inc.
—Why this step matters
Step 13 · Research & development on the roadmap
Everything a business sells has a shelf life. The question is not whether to develop something next, it is whether the thing you are drawn to is worth building — and that decision is usually made on what customers have asked for recently, which is a sample of one.
There is a cheaper screen available first. Published price data will tell you which adjacent lines hold their price on their own merits and which are only passing an input cost through. Skipping this step is how a business spends two years developing a product whose price it will never control.
What follows is one worked example. The company figures are invented; the method and the published series are not.
00Summary of findings
Of twelve adjacent product lines, Redbud's current one — fabricated structural metal — has the highest steel exposure on the board, beta +0.37, correlation +0.74. No candidate you could move into is more tied to the steel price than what you already do.
Your ten-year price growth of +102.5% looks healthy until you set it beside steel at +89.4%. Almost all of it is the input passing through. The value you add has barely repriced in a decade.
Metal heat treating has an almost identical ten-year price rise — +105.0% — with a steel beta of −0.08. Same price growth, none of it explained by steel. That is what value-add looks like in this data.
Do not build a heat treating operation. The screen is telling you a principle, not a destination, and the honest version of the principle costs far less to test. Section 07.
01The question
Redbud has roughly 2,880 press brake hours a year of spare capacity and a decision to make about what to put through them. The question is not which product feels most in demand today: it is what should Redbud develop, so that the next decade of price growth is something it earns rather than something steel hands it?
R&D at a 38M fabricator is not a laboratory. It is the decision about which capability to add next, and it is usually made on instinct about what customers have asked for. There is a cheaper screen available first.
02Sources, and what is real
Real, and checkable. Twelve BLS industry price indices and the steel commodity index, pulled 5 August 2026. Every beta, correlation and price change in this document is computed from those series.
| Series | Title | Jun 2026 |
|---|---|---|
| WPU101 | PPI Commodity: Metals and metal products: Iron and steel | 366.341 |
| PCU332312332312 | PPI Industry: Fabricated structural metal manufacturing | 290.382 |
| PCU332811332811 | PPI Industry: Metal heat treating | 321.469 |
| PCU332710332710 | PPI Industry: Machine shops | 212.612 |
The remaining eight series appear with their identifiers in the table in section 03.
Invented. Redbud, its revenue, and its 2,880 hours of spare brake capacity.
Deliberately absent. Plate work manufacturing (PCU332313332313) is the most obvious adjacent line and it is not in the screen: BLS stopped publishing it after December 2025. A missing series is not a neutral fact — it means this screen cannot see the one candidate closest to your current work, and that gap should be filled with quotes from your own customers rather than assumed away.
03The screen
For each candidate line, the beta of its year-over-year price change against the year-over-year change in steel. A beta near 0.4 means the price is largely steel wearing a different hat. A beta near zero means the price is set by something you control.
| Product line | Steel beta | r | 10-year price | Year over year | Series |
|---|---|---|---|---|---|
| Metal heat treating | -0.08 | -0.29 | +105.0% | +2.9% | PCU332811332811 |
| Turbines & generators | -0.02 | -0.15 | +31.1% | +5.4% | PCU333611333611 |
| Machine shops | -0.01 | -0.16 | +24.6% | +1.6% | PCU332710332710 |
| Construction machinery | +0.00 | +0.01 | +48.4% | +2.4% | PCU333120333120 |
| Precision turned products | +0.01 | +0.04 | +48.6% | +5.9% | PCU332721332721 |
| Mining machinery | +0.01 | +0.07 | +73.7% | +2.5% | PCU333131333131 |
| Industrial trucks & tractors | +0.06 | +0.41 | +60.1% | +5.6% | PCU333924333924 |
| Metal coating & allied services | +0.09 | +0.63 | +37.6% | −4.2% | PCU332812332812 |
| Other fabricated metal | +0.13 | +0.74 | +60.3% | +6.6% | PCU332999332999 |
| Fabricated pipe & fittings | +0.22 | +0.80 | +68.4% | +9.4% | PCU332996332996 |
| Heavy-gauge metal tanks | +0.23 | +0.69 | +109.0% | +6.8% | PCU332420332420 |
| Fabricated structural metal — Redbud today | +0.37 | +0.74 | +102.5% | +1.8% | PCU332312332312 |
| Iron & steel — the input | — | — | +89.4% | +14.2% | WPU101 |
04What the screen actually says
Low steel exposure on its own is not the prize. Machine shops (PCU332710332710) have a beta of −0.01 — almost none — but ten-year price growth of only +24.6%. That is a business insulated from steel and unable to raise prices for any other reason either. Independence from your input is worth nothing if nobody will pay for what replaces it.
The combination worth having is low beta with real price growth, and on this board exactly one line has both: Metal heat treating, at −0.08 beta and +105.0% over ten years. Customers have paid steadily more for it for a decade for reasons that have nothing to do with the steel market.
05Why the obvious conclusion is wrong
The obvious conclusion is that Redbud should get into heat treating. It is wrong, and it is wrong in a way worth being explicit about, because this is exactly where a screen like this gets misused.
- It is a different business, not an adjacent one. Furnaces, atmosphere control, metallurgical staff and customer qualification. The price behavior is attractive because the barrier is high — which is the same sentence read from the other side.
- The capital is the wrong shape. The spare capacity you are trying to fill came from recovering setup time, not from buying a machine. A heat treat line is a far larger cheque than anything spent to get here.
- The screen cannot see demand. A price index says what the industry realized, not whether anyone within a hundred miles of Tahlequah wants to buy it from you.
What the screen supports is the principle: move revenue toward operations whose price is not set by tonnage. It does not name the destination.
06How to challenge this
- Run the same beta on your own realized prices by product family. If one of your existing families already has a low beta, the answer is to sell more of what you have rather than develop anything.
- Check whether the betas hold outside 2020–2022. If the whole spread comes from one shock, the ranking is fragile.
- Ask what the missing plate work series would have shown, and get quotes rather than assume it sits with structural.
07Recommendation
Spend the 2,880 spare hours moving up the value chain on parts you already make, before adding any capability that needs capital.
- Pick the three highest-volume parts you currently ship raw and quote them finished — machined, coated or assembled. Machining and coating both sit near zero beta; you are buying independence from the steel price without buying a furnace.
- Set the decision rule now. If finished parts do not hold at least 3.7 points more gross margin than the raw equivalent after two quarters, stop and the answer is no. Write that number down before the first quote goes out.
- Re-run this screen annually. It costs nothing, the data is free, and a beta that moves is the earliest signal that a line is turning into a commodity.
The honest summary: this document does not tell you what to build. It tells you that what you build should be chosen on whether its price is yours to set — and it gives you a free, repeatable way to check that before spending anything.