Technology uptake in service settings depends on two-sided incentives: when frontline staff must mediate customer use, salesperson adoption largely determines customer adoption, and rollout failures often reflect misaligned costs or benefits across the two user groups rather than problems on only one side.

Main Finding

The paper introduces the Dual Adoption Model (DAM), which shows that technology adoption in service settings often depends on two users (e.g., frontline employees and customers) and that understanding adoption requires jointly accounting for the benefits and costs to both. In a field study of an e‑commerce platform, the authors demonstrate that frontline salespeople’s adoption mediates customers’ adoption, and that failures in rollout often reflect misaligned incentives, costs, or perceived benefits across the two user groups rather than problems on either side alone.

Key Points

• Standard technology acceptance models focus on single-user decisions and miss important dynamics when adoption is dyadic (first user → second user).
• Dual adoption is common in frontline service contexts: a customer’s uptake of a firm technology is frequently mediated by a frontline employee (or vice versa).
• The DAM formalizes that the net payoff to adoption must be evaluated for both the mediator (first user) and the end user (second user) simultaneously.
• Adoption failures can result from:
  • Low perceived benefit or high cost for the first user to facilitate second-user uptake;
  • Low perceived benefit or high cost for the second user even when the first user adopts;
  • Misaligned incentives or complementarities that make joint adoption unattractive even if individual adoption would be.
• The model yields actionable managerial levers: reduce costs or increase benefits for one or both users, reallocate incentives, redesign workflows or interfaces to lower mediation burden, or change rollout sequencing.

Data & Methods

• Empirical setting: a field study of an e‑commerce platform rollout where frontline salespeople were positioned to mediate customers’ adoption of the platform.
• Design: real-world rollout data were used to test DAM predictions about mediation effects and joint payoff considerations (paper reports a field test rather than a purely theoretical or lab study).
• Analysis approach (as reported): the study examines adoption patterns of salespeople and customers and evaluates how salesperson behavior affects customer uptake; findings are used to validate the DAM as a diagnostic and prescriptive framework for rollout problems.
• Note: the summary reflects the paper’s reported field-study test of the DAM; detailed sample sizes, statistical techniques, and robustness checks are reported in the full paper.

Implications for AI Economics

• Complementarities and network effects: AI or digital services deployed in customer-facing roles often produce externalities across user types; evaluation of economic returns must account for two-sided adoption complementarities.
• Investment appraisal and ROI: Firms should estimate joint adoption probabilities and joint payoffs (not just individual-user benefits) to avoid overestimating returns on technology investments.
• Incentive design and principal–agent concerns: When frontline workers mediate customer uptake of AI tools, incentives and compensation should be explicitly aligned to encourage facilitation rather than resistance.
• Diffusion and rollout strategy: Sequencing, targeted subsidies, training, and interface design that reduce friction for mediating users can accelerate broader adoption and increase welfare from AI deployment.
• Policy and market design: Regulators and platform designers should recognize dual-adoption frictions when assessing the social value of new technologies and when designing marketplaces or standards that depend on mediated adoption.
• Measurement and empirical work: Future AI-economics research should model and empirically estimate joint adoption decisions (e.g., using dyadic data, mediation analysis, and models for complementarities) to accurately capture adoption dynamics and welfare implications.