Guest article by Tor W. Andreassen.
In a recent commentary in AI Innovator and the Financial Times, Stanford professor Erik Brynjolfsson argues that we are finally seeing AI showing up in the productivity statistics. Revised US figures indicate slower growth in the number of jobs while GDP growth remains strong – a classic sign of higher labour productivity. His own analysis suggests that US productivity growth in 2025 may be around 2.7 per cent, almost twice the average of the past decade.
For many executives and economists, this is exactly what we have been waiting for: confirmation that AI investments are “finally paying off”. But are we really seeing the whole picture?
HSP: productivity as co‑created value
In a recently published article in the Journal of Service Research, “Reconceptualizing Service Productivity: A Holistic Measurement Framework”, I introduce Holistic Service Productivity (HSP). I argue that traditional productivity measures – priced output per unit of recorded labour and capital – were developed for industrial production and capture service value creation poorly.
The HSP framework starts from the premise that services are co‑created by provider and customer, and therefore combines four elements in a single structure:
- customer value
: benefits, mastery, safety, and risk reduction realised in use
- customer effort
: time, cognitive, and emotional effort required from the customer
- provider value
: revenues, cost reductions, and quality‑ and risk effects for the firm
- provider effort
: labour, capital, technology, and organisational resources
Holistic service productivity is expressed as the relationship between the customer’s and the provider’s net gains, and
. The key question is no longer just “how much more output do we get per hour worked?”, but “how are effort and gains distributed between the actors in the service system?”.
Why HSP offers a deeper understanding than labour productivity
1. Making hidden customer effort visible
Brynjolfsson’s numbers show that firms achieve more with fewer employees, but they say nothing about how much extra work is shifted to customers. In a digital, AI‑driven service economy, users fill in forms, upload documentation, interpret regulations and compensate for lack of guidance – all of this is real resource use that never appears in today’s productivity statistics. HSP makes this customer effort explicit through and therefore helps distinguish genuine efficiency gains from pure cost shifting from provider to user.
2. Bringing perceived value into the productivity concept
Labour productivity rewards solutions that cut time and cost, even if perceived quality and safety deteriorate for customers. HSP incorporates customer value – including time savings, risk reduction, and emotional value – into the productivity concept itself. An AI solution that reduces internal costs while increasing users’ uncertainty and cognitive load will therefore produce a weak or negative HSP, even if labour productivity on the provider side improves.
3. Evaluating AI projects on “joint value creation” – not just the bottom line
Brynjolfsson notes that many use generative AI as a “glorified dictionary”, while a smaller group of “power users” automate entire workflows. HSP provides a concrete criterion to distinguish cosmetic from transformative projects: high HSP requires AI solutions that simultaneously reduce both and
while maintaining or increasing
. This gives leaders a framework for directing AI investments towards solutions that create shared gains – not just nicer numbers in internal productivity reports.
From AI optimism to responsible management
Brynjolfsson’s message from the Financial Times is that we are moving out of the AI productivity paradox: after years of investment, we finally see a macro‑level productivity uptick. The HSP framework broadens this picture by asking how this uplift is generated and whocarries the effort.
For business leaders, this implies that AI‑driven productivity must be understood as more than extra euros or dollars per labour hour. For policymakers, it suggests that future productivity policy must consider both value creation and effort across the entire service system – including customers. Only then can we move from counting labour‑productivity growth to managing for holistic service productivity.
Tor W. Andreassen, PhD
Professor emeritus at the Norwegian School of Economics
Leader of the Professional Council, Open Innovation Lab of Norway
Image credits: kris
