Spence (1973) showed that costly observable signals communicate unobservable qualities. Technology choices do this constantly. What you run says something beyond what it does.
A startup looking for Series A funding puts "Powered by AWS" and "Built on Postgres" in its technical documentation. Another puts "Built on a proprietary cloud infrastructure" with no recognizable vendor names. All else equal, which one does the investor find more credible?
The answer most investors would give is the first one, and the reason is not only technical. AWS is a recognizable signal. It implies a certain level of infrastructure competence, a certain kind of cost discipline, and a kind of decision-making that is aligned with how the rest of the technology industry thinks. The proprietary infrastructure might be better engineered. But without the signal, there is no shortcut to credibility.
Spence (1973) introduced signaling theory in the context of job markets, though my study-hub files do not contain his original paper, so I am drawing on my understanding of the broader economics literature here. The core idea is that in markets with asymmetric information, where the buyer cannot directly observe the quality of what they are buying, sellers use costly observable signals to communicate unobservable quality. The signal must be costly to be credible. If it were cheap, everyone would use it regardless of actual quality, and it would convey no information. A college degree works as a signal in Spence's original formulation because it is expensive, in time and money and effort, and the assumption is that the cost is differentially lower for high-ability individuals.
In IS contexts, technology choices are signals in exactly this sense. They communicate unobservable qualities of the organization, its resources, its competence, its sophistication, and its strategic direction, to audiences who cannot directly verify those qualities. A healthcare system that adopts Epic is signaling something. Epic implementations are expensive, time-consuming, and organizationally demanding. The willingness to undertake one signals that the organization has the resources and the organizational maturity to operate at enterprise scale. Smaller regional hospitals that run on Epic get reflected legitimacy from the signal even if their actual Epic use is more limited.
The same dynamic shows up in vendor partnerships and cloud provider relationships. A startup that announces a partnership with Microsoft or Google gets a credibility halo from the association. The halo is a signaling effect, not a direct quality measure. Microsoft or Google has certified or endorsed this organization in some way, and that certification functions as a signal to customers, partners, and investors who do not have the time or technical capacity to evaluate the startup's actual capabilities.
One place this shows up clearly is in what the Li et al. (2023) work in IS Research discusses in the context of IT investments as signals in healthcare contexts, where visible IT adoption communicates organizational commitment in ways that affect stakeholder responses. The signaling mechanism, the idea that observable costly actions communicate unobservable properties, appears across multiple IS papers I have in my study-hub files, even where the primary theoretical framing is something other than Spence's original formulation.
The dark side of signaling is what happens when the signal drives adoption rather than the underlying fit. Organizations buy technology that is right for their audience but wrong for their operational context. An organization adopts Salesforce because its investors expect CRM sophistication, not because its sales team actually needs the features. An organization moves to Kubernetes not because its engineering scale requires it, but because "we're cloud-native" is the current signal of technical credibility. The implementation cost is real. The integration pain is real. The mismatch between the tool and the actual problem is real. But the signal paid off, at least for the period when credibility with the audience mattered more than operational efficiency.
This is related to but distinct from the legitimacy dynamics I wrote about in an earlier post on legitimacy theory and technology adoption. Legitimacy theory focuses on organizational conformity to field expectations and institutional norms. Signaling theory focuses on how individual observable choices communicate unobservable quality to specific audiences under information asymmetry. Both mechanisms can produce the same observable behavior, adopting a technology for reasons beyond functional fit, but through different causal paths.
Gartner's Magic Quadrant reports are themselves a fascinating example of signaling infrastructure. The concept of the Magic Quadrant as a methodology for ranking vendor capability is publicly documented (see Gartner's research methodologies). Being placed in the leaders quadrant is a signal that organizations use to evaluate vendors, and vendors invest substantial resources in activities aimed at maintaining or improving their placement. The signal, a vendor's position on a chart, communicates something about capability and vision that prospective customers use as a proxy when they lack the time or expertise to evaluate vendors directly. And because it works as a signal, vendor strategy is partly shaped by the requirements for the signal rather than purely by the underlying product logic.
This is not a critique of Gartner's methodology. It is an observation about how signals work once they become established. A credible signal creates incentives to invest in signal-relevant activities, which is why college rankings shape what universities optimize for, why Michelin stars shape restaurant menus, and why Magic Quadrant positioning shapes vendor roadmaps. The signal is supposed to track the underlying quality. Once it becomes a recognized signal, though, organizations also invest in the signal directly, which can diverge from investing in the quality it was meant to represent.
For IS researchers, this raises a question about how to interpret technology adoption data. If organizations are partly adopting technologies as signals, then adoption rates and market share data conflate functional fit with signaling behavior. A technology might have high adoption not because it is the best technical solution but because it became the credible signal in its category. Understanding which mechanism is operating matters for predicting adoption patterns, for explaining why technically superior alternatives sometimes fail to displace established ones, and for advising organizations on technology decisions that are actually aligned with their context rather than their audience.
About the author
Share
More notes
Related notes