The servitization bet

Equipment-as-a-Service (EaaS) means an original equipment manufacturer (OEM) sells a machine's availability, output or result instead of the machine itself. Under EaaS, as under any outcome-based contract, the OEM keeps the performance risk for the life of the contract. The margin case is real. Deloitte's 2026 manufacturing outlook, citing its own aftermarket research, puts aftermarket service margins at more than twice those of equipment sales alone [1]. Aftermarket margin is not the same as EaaS margin, which still has to be proven contract by contract.

The same move turns each wrong assumption into a recurring cost.

If usage is higher than planned, a site is harder to service than expected, or the customer reads the outcome differently from the OEM, the OEM pays for it every month until the contract ends.

Research on servitization has a name for the result when this goes wrong. Gebauer, Fleisch and Friedli, working with more than 30 equipment manufacturers, described a "service paradox": companies invest in extending their service business, costs rise, and the expected returns do not follow [2].

This paper covers the parts of an EaaS bet that decide whether it pays and that a launch plan usually leaves untested: how the outcome will be measured and disputed, how serviceability varies across the installed base, what a renewal tells you that a signature does not, how the transition affects the OEM's own revenue and people, and how to test each of these with customers and field teams before the legacy model winds down. Who approves the purchase on the buyer's side is covered in The committee nobody mapped, which also covers how capital and operating approvals differ.

Under EaaS, the performance risk moves from the buyer to the OEM for the life of the contract. For the OEM's finance and operations leaders, that has four concrete parts.

Usage risk. A per-hour, per-unit or per-output price assumes a usage level. If the customer runs the machine harder than planned, wear, parts and site visits rise while the price stays fixed.

Serviceability risk. An uptime or availability promise assumes the OEM can reach the machine, diagnose it and repair it within a set time. That varies by site, as the next section explains.

Measurement risk. An outcome contract pays on a number. If the OEM and the customer calculate that number differently, the difference becomes a dispute, and disputes are expensive to settle inside a long-term relationship.

Renewal (churn) risk. The business case for EaaS usually depends on renewals. A three- or five-year contract that does not renew leaves the OEM with a used asset at the customer's site and no margin to recover the cost of the transition.

Each of these can be tested before launch. Most OEMs test the first, the pricing, and discover the other three in the field.

An uptime promise is priced for an average site. The installed base is not average.

Machines at different customers sit in different conditions. Some are in clean, climate-controlled plants near an OEM service center. Others are in dusty, hot or remote locations, run three shifts, and are maintained by a customer team with limited skills. Some are connected and can be monitored remotely. Others have no network connection, or the customer's security policy does not allow one.

Each of these conditions changes the true cost of the promise:

• Distance and access. Travel time to site, and whether the customer allows remote access, decide how quickly a failure can be fixed.
• Operating conditions. Duty cycle, environment and the materials processed decide how often failures happen.
• Customer maintenance. Whether the customer's own team does first-line maintenance well, or at all, decides how many issues reach the OEM.
• Spare parts. Whether critical parts can be held near the customer decides the length of an outage.
• Machine age and configuration. Older units, or units modified by the customer, can be harder to diagnose and repair.

An EaaS price that does not vary with these conditions is subsidizing the hard sites with the easy ones. That can be a deliberate choice. It should not be an accident.

The practical test is to sort a sample of the installed base by these conditions before pricing, estimate the cost to serve for each group, and ask the field team which sites they already know to be expensive. They usually do know.

Outcome contracts pay on metrics such as availability, output or overall equipment effectiveness (OEE). These sound precise. In practice, they depend on definitions and data that the two parties may not share.

ISO 22400-2 defines a set of key performance indicators for manufacturing operations, including OEE as a product of availability, effectiveness and quality, with defined elements for each [3]. A standard definition helps. It does not settle the questions that cause disputes:

• Whose instrument is the record? The OEM's telemetry, the customer's manufacturing execution system (MES), or a manual log?
• What is the baseline? Performance before the contract, the machine's rated performance, or a figure agreed at signing?
• What is excluded? Downtime caused by the customer's operators, by raw material problems, by power failures, by planned maintenance the customer postponed?
• Who decides when the data is missing? Connected machines lose connection. Gaps in the data need a rule.
• What happens after a disagreement? A review process, an independent measurement, a cap on penalties?

The best test is to rehearse a dispute before signing the first contract. Take a realistic month of data from a pilot or a connected customer, and ask the customer's operations and finance people (at a plant, the plant manager and the plant controller) and the OEM's service and finance people to calculate the payment independently. If they arrive at different numbers, the contract needs work. If they cannot agree on how to calculate it, the contract is not ready.

This exercise is quick and cheap compared with the first real dispute, which usually arrives when the relationship is already under pressure.

Capital-heavy industries have sold service by the hour for decades, and they show which part of the cost moves. Airlines own or lease their aircraft: 53% of the world's commercial fleet is leased from lessors that own it [9]. What moved to an hourly price is the engine maintenance. One aero-engine maker charges its long-term care program at a fixed rate per flying hour and takes on the maintenance risk [10]. Railways work the same way. A European rail vehicle rental company contracted the maintenance of 70 of its locomotives to a train maker for eight years [11], and a mining railway in Guinea signed a multi-year services agreement worth more than $700 million for maintenance, parts overhauls and related support of its fleet of new locomotives [12]. Power plants own their gas turbines and buy long-term service agreements. A 2001 trade-press survey of them reports examples of 12 years and of more than 20, priced per fired hour, per start, per planned maintenance event, per calendar year or per megawatt-hour [13].

In each case the asset stays on a balance sheet, the operator's or a lessor's. What the OEM sells by the hour is the maintenance, which the operator was already paying as an operating cost. That is why it sells: it turns an unpredictable operating cost into a predictable one and moves the risk of failure to the party best placed to manage it, without asking the buyer's finance team to give up an asset.

Your customer's controller will ask where the fee lands against EBITDA, who takes the depreciation deduction, and whether the contract contains a lease. Have the answers before the quote goes out. The accounting also explains why the asset itself rarely moves.

• Capital spend sits below EBITDA. A machine bought outright is capitalized and depreciated over its useful life: years for most plant machinery, decades for turbines, locomotives and aircraft. Depreciation is excluded from EBITDA. A service fee that replaces the purchase is an operating cost and lowers EBITDA, which matters to any company whose lenders, bonus plans or valuation run on it.
• US tax now favors owning. The One Big Beautiful Bill Act, signed on 4 July 2025, permanently restored 100% bonus depreciation for qualifying equipment acquired after 19 January 2025, and added temporary 100% expensing for new US production buildings: construction must start after 19 January 2025 and before 1 January 2029, the building must be in service before 1 January 2031, and only the part used for qualified production counts [14]. A US buyer with taxable profit can deduct the whole purchase in the first year. Under a subscription, that deduction goes to whoever owns the machine.
• Leases are on the balance sheet, and under US GAAP they can lower EBITDA. Under ASC 842 and IFRS 16, nearly all leases go on the lessee's balance sheet, and a contract that conveys the right to control the use of an identified asset can contain a lease even when it is not labeled as one [8]. Under IFRS 16 the cost appears as depreciation and interest, below EBITDA. Under ASC 842 an operating lease is a single straight-line cost inside operating income, so it reduces EBITDA [15]. A service agreement that gives the customer the right to control a specific machine can contain a lease. An EaaS pitch built on keeping the asset off the customer's books is a pitch the customer's controller will not accept.
• A pure service contract is an operating expense, as maintenance and repair already are. A price that replaces an existing operating cost (labor, maintenance, parts, consumables, energy, downtime) does not ask finance to change how anything is booked.

Where the whole asset does go as a service, the output is usually metered and the asset is not core to the buyer. One compressor maker designs, installs, operates and maintains the compressed-air station on the customer's site and sells the air at a pre-agreed price per cubic foot, an operating expense to the customer [16]. The plant needs the air, not the compressor. Robots offered as a service follow the same logic: they replace labor, another cost the buyer already expenses. The subscription fleet is growing quickly from a small base: more than 24,500 units in 2024, up 31%, against almost 200,000 professional service robots sold that year. In transport and logistics, the largest application, traditional sales remain the main channel [17].

Evidence on how far full EaaS has spread is thin. IoT Analytics estimates that under 1% of equipment sales in 2023 were executed under EaaS contracts. The firm describes its method as surveys, secondary research and interviews with EaaS vendors and end users, and does not publish the calculation [7]. Treat it as a Benchmarked estimate, not a measure.

Two mechanics decide how the buyer's finance team reads an offer.

The operating budget is the faster path, and the less stable one. A subscription paid from the operating budget can skip the capital appropriation request, the step where a plant's capital project waits for the capital board. That is why opex offers sell faster. But a purchase from the capital budget is decided once. A subscription is decided again at every renewal, by a budget owner who may have changed since the first signature.

The cost the price replaces decides the sale. An offer that replaces labor, maintenance, parts or downtime competes with a cost the plant already carries. An offer that asks the plant to stop buying the machine competes with depreciation, the tax deduction, EBITDA and the plant's control of its own equipment.

So the test is which cost line the offer replaces and how the buyer's finance team books it. A survey answer about preferring opex is Benchmarked at best. A plant controller who confirms which budget the contract comes from, how finance books it and who approves it, is Verified.

A signed multi-year term is weaker evidence than it looks, because a good price or attractive financing can win the signature on its own. For the OEM, this has a direct consequence for how the business case is built. Early adoption numbers will look good if the price is attractive. The real test comes at renewal, often three or five years later, when the customer compares the service with buying the machine outright or with a competitor's offer.

Waiting five years to learn the renewal rate is too slow. Three ways to get evidence earlier:

Ask existing support customers. Many OEMs already sell service agreements alongside equipment. Their attach and renewal rates, and the reasons customers give for renewing or not, are the closest evidence available.

Ask customers near the end of a contract, away from the sales relationship. The question is what they would do if the support were unbundled and quoted from scratch. The answer is more useful when someone other than the account manager asks it.

Test the price without the financing. If the offer only appeals because it moves the purchase from capital to operating spend, the value is in the financing, not the service. That can still be a business, but it is a different one, and a finance company can offer the same thing.

Public OEM filings show where the service business sits once it is established, and how uneven adoption of the software layer can be. One elevator OEM reports service as 65% of net sales and 91% of segment operating profit [4]. One HVAC OEM reports new equipment at 72% and parts and service at 28% of net sales [5]. One agricultural OEM reports slower than expected customer adoption of some of its precision technology solutions and of its software-as-a-service subscription services [6]. These are Benchmarked figures from other sectors; an OEM's own attach and renewal rates are the evidence that counts.

Moving to EaaS changes the OEM's own revenue and organization before it changes the customer's.

Revenue timing. A machine sale books revenue at delivery. A multi-year contract spreads it over years. During the transition, reported revenue can fall even when the commercial result is good. The board and lenders need to understand this before the launch, not after the first quarter.

Capital. Under many EaaS structures, the OEM or a financing partner owns the equipment. That ties up capital and changes the balance sheet. The case should state who funds the fleet and at what cost.

Field service incentives. Field technicians are often paid on service calls, parts and time-and-materials work. Under a multi-year contract with an availability promise, every call is a cost to the OEM, not a source of revenue. This is a stakeholder inside the OEM that a launch plan often misses. The practical step is to redesign service incentives before launch and to ask field operations leaders directly whether the new model reduces what their teams earn.

Sales incentives. Sales teams paid on equipment revenue at signing have little reason to sell a contract that pays over years. Commission plans need to change, or the sales team will keep selling machines.

Dealers and distributors. Where the OEM sells through dealers, the dealer's margin on equipment and parts can disappear under a direct service model, and dealers may resist. The route has to be chosen and tested: the dealer sells and bills the subscription, the distributor stocks it, the OEM sells direct and pays the dealer for delivery, or an OEM marketplace carries it. Each changes who owns the renewal. How to test a partner route is covered in Through the integrator.

The service paradox research suggests that the internal transition, not the customer's willingness, is where many servitization efforts lose money [2]. It deserves the same testing as the market.

EaaS is a family of offers. Each moves a different amount of risk to the OEM.

Offer	What the customer pays for	Risk the OEM carries
Service agreement with the equipment sale	Maintenance and support, alongside a purchased machine	Cost to serve, within the agreement's scope
Subscription or leasing with service	Use of the machine for a term, plus service	Asset value and cost to serve
Uptime guarantee (service-level agreement, SLA)	Machine availability above an agreed level	Cost to serve and penalties for downtime
Pay-per-use	Hours, cycles or units produced	Usage and cost to serve
Outcome-based	A business result, such as output or quality	Usage, service, measurement and part of the customer's process

Moving down the table increases both the potential margin and the number of assumptions that must hold. One path is to start one row up from the target, gather evidence on usage, serviceability and disputes, and move down once that evidence exists.

The right starting row depends on what customers will pay for, which is a question for customers, not for the OEM's product team.

Confirm the outcome and the price first, then the budget pathway and the renewal, then the channel, and launch after that. The tests below add detail within that order.

With customers:

• Which outcome they would pay for, measured how, and by whose instrument.
• What they would pay, with and without the financing benefit.
• Who approves an operating-cost contract of this size, and whether that process exists.
• What they would do at renewal if the support were quoted on its own.

With the installed base data:

• Cost to serve by site condition, from existing records.
• Which sites are connected, and which could be.

With the OEM's own teams:

• Field operations leaders on incentives and on which sites are expensive.
• Sales leaders on commission plans.
• Finance on revenue timing, capital and the reporting effect.

With a rehearsed dispute:

• A month of real data, calculated independently by both sides.

Each claim in the business case then receives a grade: Verified where the customers or internal teams who make the decision have confirmed it in scored interviews, Benchmarked where it rests on benchmarks or internal estimates, Assumption where it is still assumed. The evidence standard is described on the method page, and what we test for OEMs covers the equipment manufacturer's side.

Customers

1. Which result would you pay for: availability, output, quality, or something else?
2. How would you measure it, and whose system would be the record?
3. What downtime would you expect to be excluded from the calculation?
4. Would this come from your capital (capex) or your operating (opex) budget, who approves it, and would your finance team book it as a lease?
5. If the support were quoted separately from the machine, would you still buy it?
6. What would make you not renew at the end of the term?

Installed base

7. Which sites are connected, and which allow remote access?
8. What is the cost to serve per site today, grouped by distance, duty cycle and customer maintenance capability?
9. Which sites do field teams already know to be expensive?

Inside the OEM

10. How are field technicians paid today, and what happens to their pay under the new model?
11. How are sales teams paid, and does the plan reward a multi-year contract?
12. What happens to reported revenue in the first two years of the transition, and have the board and lenders seen it?
13. Who funds the fleet of machines under contract?
14. What margin does the dealer keep on the subscription, and who owns the renewal?
15. Which sites does the dealer service today that the OEM cannot reach?

Dispute rehearsal

16. Given one month of real data, do both sides calculate the same payment?
17. If not, which definition caused the difference?

1. Deloitte, "2026 Manufacturing Industry Outlook." https://www.deloitte.com/us/en/insights/industry/manufacturing-industrial-products/manufacturing-industry-outlook.html
2. Heiko Gebauer, Elgar Fleisch and Thomas Friedli, "Overcoming the Service Paradox in Manufacturing Companies," European Management Journal 23, no. 1 (2005): 14 to 26. https://doi.org/10.1016/j.emj.2004.12.006
3. International Organization for Standardization, "ISO 22400-2:2014 Automation systems and integration: Key performance indicators (KPIs) for manufacturing operations management, Part 2: Definitions and descriptions." https://www.iso.org/standard/54497.html
4. Otis Worldwide, Form 10-K for the fiscal year ended 31 December 2025. https://www.sec.gov/Archives/edgar/data/1781335/000178133526000011/otis-20251231.htm
5. Carrier Global, Form 10-K for the fiscal year ended 31 December 2025. https://www.sec.gov/Archives/edgar/data/1783180/000178318026000008/carr-20251231.htm
6. Deere & Company, Form 10-K for the fiscal year ended 2 November 2025. https://www.sec.gov/Archives/edgar/data/315189/000110465925122321/de-20251102x10k.htm
7. IoT Analytics, "OEM servitization strategies: Why Equipment as a Service hasn't taken off yet," and the method note for its "Equipment as a Service Market Report 2024–2028." https://iot-analytics.com/oem-servitization-strategies-equipment-as-a-service/ https://iot-analytics.com/product/equipment-as-a-service-market-report-2024-2028/
8. IFRS Foundation, "IFRS 16 Leases." https://www.ifrs.org/issued-standards/list-of-standards/ifrs-16-leases/. PwC Viewpoint, "Scope of the leasing guidance and embedded leases." https://viewpoint.pwc.com/dt/us/en/pwc/accounting_guides/ifrs_and_us_gaap_sim/ifrs_and_us_gaap_sim_US/Chapter_14Leases_1/14_2_scope.html
9. CAPA – Centre for Aviation, "Aircraft leasing in equilibrium at just over half the world fleet," 9 February 2024. https://centreforaviation.com/analysis/reports/aircraft-leasing-in-equilibrium-at-just-over-half-the-world-fleet-675212
10. Rolls-Royce, "TotalCare." https://www.rolls-royce.com/media/our-stories/discover/2017/totalcare.aspx
11. Alstom, "Alstom has been awarded a maintenance contract for 70 locomotives by Alpha Trains Group," 28 April 2022. https://www.alstom.com/press-releases-news/2022/4/alstom-has-been-awarded-maintenance-contract-70-locomotives-alpha-trains-group
12. Wabtec, "Wabtec and La Compagnie du TransGuinéen sign a $700 million plus rail services agreement," 21 September 2026. https://www.wabteccorp.com/newsroom/press-releases/wabtec-and-la-compagnie-du-transguin-en-sign-a-700-million-plus-rail-services-agreement
13. Power Engineering, "Long-Term Service Agreements: Weighing the Risks and Rewards," 1 August 2001. https://www.power-eng.com/gas/turbines/long-term-service-agreementsweighing-the-risks-and-rewards/
14. US Internal Revenue Service, "Treasury, IRS issue guidance on the additional first year depreciation deduction amended as part of the One, Big, Beautiful Bill," Notice 2026-11, 14 January 2026. https://www.irs.gov/newsroom/treasury-irs-issue-guidance-on-the-additional-first-year-depreciation-deduction-amended-as-part-of-the-one-big-beautiful-bill. BDO, "One Big Beautiful Bill Act Expands 100% Depreciation Expensing Opportunities," for the qualified production property rules. https://www.bdo.com/insights/tax/one-big-beautiful-bill-act-expands-100-depreciation-expensing-opportunities
15. KPMG, "Lease accounting: IFRS Accounting Standards vs US GAAP," 5 December 2025. https://kpmg.com/us/en/articles/2025/lease-accounting-ifrs-standards-us-gaap.html
16. Kaeser Compressors, "Compressed Air Utility Service." https://us.kaeser.com/services/compressed-air-as-utility-service/
17. International Federation of Robotics, "World Robotics 2025: Service Robots," executive summary and press release, 7 October 2025. https://ifr.org/img/worldrobotics/Executive_Summary_WR_2025_Service_Robots.pdf https://ifr.org/ifr-press-releases/news/service-robots-see-global-growth-boom

What is not yet sourced

• Renewal rates for EaaS and outcome contracts. No published figure was found. The paper recommends gathering the evidence from existing service customers instead of assuming a rate.
• How often outcome contracts lead to disputes, and what they cost. No published figure was found. The dispute rehearsal is a way to find the answer for one OEM's offer.
• The variation in service cost across the installed base is described from Thing Company's experience. Each OEM's own service records are the better source.
• ISO 22400-2 (reference 3) was confirmed through standards-catalog and peer-reviewed descriptions of its OEE formula; the standard itself is paywalled and was not read.
• The turbine service agreement terms (reference 13) come from a 2001 article, so they show the structure of such contracts and not current terms.
• The offer table and the suggestion to start one row up are Thing Company's framing, not an industry standard.

Thing Company is an independent market validation practice for industrial technology. We test whether a buyer exists at a price that works. For OEMs, that means testing a service or outcome model with customers, the installed base and the OEM's own teams before the legacy revenue winds down.

Planning a service or outcome model? Start with a Sprint Assessment.

Not ready to talk? Run the free service offer dispute rehearsal self-check from the toolkit.