The discount is argued over the machine. The money is in the options, parts and agreements that follow it for a decade, and the option set that leaves the factory decides them. Revomo holds machine, configuration and install base on one record, and turns what a configuration returns into a published architecture.
The first is the model-year book: base list, option adders, dealer schedule, programme calendar. The second is what the machine consumes after it ships, and the configuration decides its size.
The government measures the split. BLS publishes separate price indexes for machines and for their parts inside one six-digit industry. In construction machinery manufacturing, calendar 2025, machine indexes rose 0.2% to 3.1% (213.190 to 213.595 on graders and rollers) while parts sold separately rose 44.1%, 212.470 to 306.085. Read it narrowly: one industry, one year, agency levels and percentages we derived, in the year Section 232 duties moved to full customs value, so tariff pass-through on imported parts is as plausible as commercial intent, and packaging and farm machinery show no such gap over six years. What travels is that a machinery business runs two price systems and manages one.1
The first four are governed. The rest decide what the machine returns, on calendars that do not meet.
Once the machine ships, the option set is fixed for life: what it wears, how often it is serviced, who can service it. Nothing later undoes it.
AGCO tells the SEC that most of its products "can be economically maintained with parts and service for a period of 10 to 20 years", and puts replacement parts at 19% of 2025 net sales against 13% in 2023.2
Annual price moves here are low single digits. Stream mix is not: parts, labour and agreements carry a higher margin than the machine.
Kadant attributes its 2025 consolidated gross margin, 45.2% against 44.3%, to parts and consumables rising from 66% to 71% of revenue.3
Bonuses, incentives, floor plan support and rate buy-downs settle after the machine leaves, reconstructed quarters later from systems never joined.
Deere lists sales incentives among its critical accounting estimates: "There are numerous programs available at any time, and new programs may be announced after we record the equipment sale to the dealer."4
A trade-in over-allowance becomes an inventory value, absorbed freight and startup land in cost of sales. Neither reaches a discount field.
Astec names four in one sentence of its revenue recognition policy: installation assistance, "service work to be performed in the future without charge", floor plan interest reimbursed to dealers, extended warranties.5
The enterprise problem here is rarely a missing system. A machinery group has ERP, a configurator, PLM, field service, dealer systems and a data platform, and the question can still be unanswerable: which configurations reprice, by how much, where, on whose approval? Examples, not an integration list.
Orders, cost roll, lists, and where the architecture lands. Examples:
Buildability rules and the commercial rules resolving against them. Examples:
The engineering truth the price rules track. Examples:
The records that never had a system, and usually decide the outcome:
Data is ingested through scheduled connectors into Revomo’s governed store, and governed outputs are written back to the systems where transactions occur. Warehouses are sources, not the store of record. The configurator decides what can be built; Revomo governs the architecture it resolves against.
A distributor’s spine runs from supplier cost to pocket margin. A builder’s runs through an object neither end contains: the configuration, fixed at shipment, governing a decade.
One foundation the chain resolves against, so a parts line ten years out answers to the option code that was discounted.
Explore Commercial ContextRanked by how directly the decision moves margin and how fast a builder can act.
Why it is hardThe configuration decides the economics and few analyse at that grain: the order system holds the option set, the service system what it consumed, and they do not join.
Explore Customer & Product ProfitabilityWhy it is hardA per-family policy treats two machines with one badge and a 2.44x difference in lifetime contribution identically, and both book the same margin.
Explore Price Lists & Customer-Specific PricingWhy it is hardOrders configured on the old architecture but delivered after the new one get honoured, repriced, or honoured because nobody chose.
Explore Price Increases That StickWhy it is hardEach lands somewhere other than a discount field: inventory, cost of sales, an accrual settling quarters later. The deal record shows a clean discount; the P&L does not.
Explore Margin Leakage & Price RealizationWhy it is hardParts run off a matrix treating criticality and captivity as one thing, and one option sourced from two countries can sit in two duty tiers on one machine, so a uniform percentage under-recovers.
Explore Cost, Tariff & Freight RecoveryOne of four equipment businesses in a group turning $840M across three regional model years, 196 dealers in 24 markets and an option book in five currencies. This one turns $156.0M: machines $68.0M at 20.0%, options $28.0M at 44.0%, parts, service and agreements $60.0M at 45.1%, so 38.5% of revenue carries 51.1% of gross profit. The aftermarket margin is calibrated to a filer: Kadant, 71% parts and consumables, reports 45.2% consolidated. The decision is the annual republication of lists, adders, schedule and multipliers on one effective date, on ten-year contribution per configuration.
The file lands as a governed dataset rather than a spreadsheet in an inbox, so everything below computes on the same records the business runs on.
Real money on the wrong variable. Nothing changes about which configuration earns what, and 4.1% unit loss breaches the −1.0% tolerance. Each machine lost takes its decade: about $1.5M of aftermarket gross profit against $2.00M gained once.
More from 214 codes than the uniform move takes from 1,180, because it moves the variable that decides the decade rather than the one the customer watches. It costs two families held out of scope, and it alone changes what next year’s machines are worth.
Attractive, easier, and a different page’s decision: it prices the annuity without touching the act that creates one, so the 2019 architecture goes on mispricing next year’s machines.
| Segment | Accounts | Lines | Move | Recovery | Why this action |
|---|---|---|---|---|---|
| Wear-critical multipliers banded 2.2x / 2.9x / 3.4x | 214 codes | 6 of 8 families | 2.6x → 3.1x avg | $1.33M | They decide the parts stream and cannot be substituted later. |
| Configuration-aware discount corridor | 148 high-duty | 6 of 8 families | 14.8% → 12.1% | $1.47M | Puts the guardrail where 56.1% of the discount sits. |
| Three-year agreement re-priced at the order | 148 high-duty | 22% → 41% attach | Re-price | $0.52M | Priced against the configuration’s consumption, not a schedule. |
| Import-exposed families | 2 of 8 | 96 orders | Hold at +2.0% | $0 | Deliberate. $0.55M not taken; share defence outranks one cycle. |
| Orders configured before the effective date | 118 orders | of 386 | Hold to the quoted architecture | $0 | Repricing a committed configuration is a conversation, not a publish. |
| Total | 3 taken · 2 held · 214 codes · 148 orders | +$3.13M | Annualized, net of modeled volume response | ||
| Rule | When | Then | Approval |
|---|---|---|---|
MY·BAND | Wear-critical code below its band | Reprice to the band | Auto |
MY·CEIL | Proposed multiplier above 3.4x | Block | VP Product |
MY·CORR | Contribution above the family median | Apply the high-duty corridor | Auto |
MY·FLOOR | Discount breaches the configuration floor | Route with contribution attached | Regional GM |
MY·MKT | Family flagged import-exposed | Hold list at or below +2.0% | Auto |
MY·BOOK | Order predates the effective date | Hold to the quoted architecture | Auto |
Backtested against eight quarters of shipped configurations and shadow-run beside the live MY2026 book, three model years and five currencies each on their own date. Rules participate before an action is proposed, so global policy is set without centralizing every decision.
The agent reasons. It does not publish. Reasoning does not confer authority: the state change is a governed transition, and policy decides if it is legal.
Day 52 leaves 22 days for 196 dealers to load the book, which is why the reset is dated backwards from the effective date. The seller gets a published corridor with ten-year contribution attached.
Three causes, on the record. $0.18M from 19 of the 214 repriced codes reverting at two accounts, and they were the higher-value ones: 8.9% of codes, 13.6% of the line. $0.19M from attach at 34% not 41%. $0.06M from exceptions above the floor.
The next architecture starts from what this one realized, per configuration.
An adoption order, not a catalog. Each tier produces an artefact the next depends on.
In most builders this record exists nowhere.
Machine, configuration, option codes, dealer, warranty, agreement and service history on one foundation at serial and option-code grain. The reconciliation alone surfaces codes with no price.
Per configuration. Per order averages the effect away.
Lifetime contribution by configuration and option code, each concession attributed to the record that caused it.
An annual book with a hard start date and a backlog straddling it.
Lists, adders, schedules and corridors as governed objects: decision tables with rule IDs, precedence as a chain, effective-dated resets, backtest and shadow-run.
From what a configuration returned to what next year’s should cost.
Contribution and volume-response models by family and configuration band, each cell declaring its state.
Actions arriving inside the ceilings, floors and holds, with the reason attached.
A scheduled event is the natural agent trigger.
Cost roll completes, condition evaluates, the agent drafts, the publish crosses a governed transition.
Answers "why did this option code move" against the record, not a memory of the last reset.
Where the corridor is consumed and programme economics settle.
Seller guidance consuming the published corridor.
Programme money and incentives settled against the machines they attach to. Currently in early access.
Multi-year service agreements as entitlements read before publishing. Currently with design partners.
Not with a pricing programme. With one question a builder can rarely answer: what was each configuration you shipped last year worth, once its service history is attached? Commercial transformation decomposes into decision systems, and for a group the first is one business or model year.
Start with one configuration question. Expand on the same platform.
Four motions weight the same platform differently. The first phase follows the motion.
You publish a schedule and someone else sets the retail price. Programme money, floor plan support and incentives are where the concession lives.
No dealer layer, but acceptance runs through factory test, site test, commissioning and run-at-rate, and every day of overrun is unpriced labour.
Option content dominates the price, and an engineering change reaches the bill of materials long before a multiplier.
The machine places a consumption stream, so its price is close to an acquisition cost and the configuration sets the rate.
Start with the option file, a year of shipped configurations, and the service history behind them, joined at serial grain, so each configuration’s return reads against its discount.
Externally sourced figures on this page, in order of appearance. Every other number is specimen data inside the decision story and is labelled illustrative where it appears.