Embodiment and Service Margin Discount
When you need this method
A business presents itself as software, but the value is delivered on devices, at the edge, or with humans in the loop, such as teleoperation or experts-as-a-service. Anyone valuing or planning with software margin expectations structurally overestimates scalability. This affects investors in valuation as much as leadership teams in their own planning.
Approach
- 1Check whether delivery is embodied: software on hardware, physical integration, or humans permanently in the loop.
- 2If so, mark the margin expectation down to services or systems-integration levels instead of applying software multiples.
- 3Net the discount against the positioning gain: the company often buys its differentiated position precisely with this margin disadvantage.
- 4Flag the tension explicitly, for example as "position strong, economics burdened".
- 5Check whether the embodied share declines over time (automation) or is structural.
Typical application
A typical case: a robotics software vendor markets an end-to-end solution behind which humans intervene remotely when things go wrong. Its market position is strong precisely because of that completeness promise. Valuing it as a pure software business would still be wrong: the human intervention and hardware proximity push margin permanently toward services levels. The honest reading: the company pays for its differentiated position with margin, and exactly that trade-off belongs in every scaling and valuation decision.
Limits and counter-indications
The discount is a qualitative diagnostic heuristic, not a precise markdown formula. It was confirmed as a diagnostic building block in an internal blind test of 30 companies, but it is not a validated valuation model. Temporary service shares during rollout do not justify the discount; what matters is whether the embodiment is structural.
How to measure impact
Gross margin versus the software benchmark (roughly 70 to 85 percent), plus the share of embodied delivery in total cost.
Related methods
Sources
- 1.William J. Baumol: Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis, The American Economic Review 57(3), 1967, S. 415-426 (opens in a new tab) · 1967 · academic and scholarly literature · supports the underlying mechanismBaumol divides economic activities by the role of labour - an incidental input to the final product (technologically progressive) versus labour that is itself the end product (teaching, live performance, government, restaurants) - and shows in a two-sector model that real unit costs in the latter rise cumulatively and unavoidably relative to the former.
- 2.The New Business of AI (and How It's Different From Traditional Software), Martin Casado & Matt Bornstein, Andreessen Horowitz (opens in a new tab) · Andreessen Horowitz (a16z) · 2020 · investment, consulting and analyst firms, industry bodies and public agencies · supports the underlying mechanismPuts gross margins for AI business models at 50 to 60 percent, against 60 to 80 percent for classic software and 30 to 50 percent for service firms, and traces them back to compute cost and permanently involved people. The source does not address software shipped on its own hardware or physically integrated products; it covers the human-in-the-loop part of the mechanism, not the hardware part.
- 3.The State of AI 2025, Bessemer Venture Partners (BVP Atlas) (opens in a new tab) · Bessemer Venture Partners · 2025 · investment, consulting and analyst firms, industry bodies and public agencies · provides benchmark figuresQuantifies across recent cohorts that vendors with a high model and service share sit well below customary software margins: around 25 percent gross margin on average across the ten fastest-growing AI vendors (the "supernovas"), against roughly 60 percent in the second cohort.
- 4.William J. Baumol, William G. Bowen: On the Performing Arts: The Anatomy of Their Economic Problems, The American Economic Review 55 (Papers and Proceedings), 1965, S. 495-502 (opens in a new tab) · 1965 · academic and scholarly literature · supports the underlying mechanismThe two-years-earlier first statement of the same hypothesis, confined to the performing arts: where the performance itself is the product, labour input cannot be reduced, so costs structurally outrun the rest of the economy. Baumol cites this work in the 1967 footnote as the origin of the hypothesis.
- 5.Richard B. Chase: The Customer Contact Approach to Services: Theoretical Bases and Practical Extensions, Operations Research 29(4), 1981, S. 698-706 (opens in a new tab) · 1981 · academic and scholarly literature · supports the underlying mechanismThe potential efficiency of a service system is a function of the degree of customer contact entailed in creating the service - the operations-side version of the same mechanism, the peer-reviewed successor to the 1978 HBR piece.
Origin: Convios