Same Tool, Different Department, Total Disaster

I was reading Giddens (1984) for the third time when a line I had skipped over finally landed. He writes that social structures are both the medium and the outcome of human action. Not one or the other. Both at the same time. That sentence rewired how I think about every enterprise software failure I have ever seen. The same tool, rolled out across an entire organization, producing radically different results in different departments. Not because the software changed. Because the appropriation changed.

Giddens built structuration theory around what he called the duality of structure. Structures are not external cages that constrain us from above, and they are not mental constructs we invent on the spot. They are both. People draw on structures to act, and their patterned action reproduces or transforms those very structures. The loop is recursive. He specified three modalities that link structure to agency. Signification uses interpretive schemes to produce shared meaning. Domination allocates resources to produce power relations. Legitimation deploys norms and sanctions to produce moral order. When you act, you draw on all three. When you act repeatedly, you reinforce or reshape all three. That is the engine underneath every technology rollout that goes sideways.

Orlikowski (1992) took Giddens' duality and pointed it straight at technology. She argued that technology is both a product of human action, designed and built through organizational and social processes, and a medium of human action, shaping what users can do and how they do it. This is the duality of technology. It is not that technology determines behavior, and it is not that people freely choose how to use a tool. It is both. Technology was made by people, and it then structures what people can make with it. Orlikowski introduced the concept of technology-in-practice to emphasize that the structural properties of technology are not fixed. They are enacted through recurrent use. Two workgroups using the same system can develop entirely different technologies-in-practice, because their patterns of use are different. The technology does not come with a single predetermined outcome. It becomes what people make of it, and what they make of it then becomes part of the structure that shapes future use.

This is where DeSanctis and Poole (1994) come in. They extended structuration to group technology use with Adaptive Structuration Theory, or AST. Their key move was to show that technologies come with structural features, meaning the rules, resources, and capabilities embedded in the system, and a spirit, meaning the general intent or official line for how the technology should be used. The spirit is the designer's vision. The structural features are what the system can do. But groups appropriate those structures, and appropriation can be faithful or unfaithful. Faithful appropriation means the group uses the technology in a way consistent with its spirit and structural design. Unfaithful appropriation means the group departs from that spirit, either by underusing, repurposing, or subverting the features.

Think about what that distinction explains in practice. A university rolls out a learning management system with the spirit of collaborative learning. Faculty in one department appropriate it faithfully, using the discussion boards, the peer review tools, the shared document spaces. Faculty in another department appropriate it unfaithfully, using it only as a file repository, stripping out every collaborative feature and treating the platform as a glorified FTP server. Same software. Same license. Same training. Different spirit in practice. Different technology-in-practice. The LMS did not change. The appropriation changed, and with it, the structure of use.

I keep seeing this pattern everywhere. SAP implementations that transform procurement in one division and stall out in another. Salesforce deployments where the sales team builds an elaborate pipeline tracking system and the marketing team enters the minimum data required and ignores every dashboard. Slack, which becomes a lifeline in engineering, where developers thread deeply about code reviews and incident response, and a ghost town in marketing, where it gets used as a glorified email forwarder with notification fatigue piled on top. The 2020 remote work shift was a natural experiment in this. Organizations bought Zoom, Teams, and Notion for everyone at once. Whether those tools thrived or withered inside a specific team depended almost entirely on how people appropriated the structural features, how the spirit of the tool aligned or misaligned with existing work norms, and whether the existing interpretive schemes, resource distributions, and legitimacy structures supported or undermined the new practice.

Giddens' three modalities give this pattern its theoretical backbone. When a new tool arrives, the signification modality determines whether people share an interpretive scheme for what the tool is for. If engineering and marketing hold different meanings for "Slack," they will use it differently. The domination modality determines who controls the resources tied to the tool. If a department head mandates Slack standups but never attends, the power structure undermines the practice. The legitimation modality determines whether the new use pattern is seen as appropriate or deviant. If senior colleagues mock people who post in Slack channels, the normative structure punishes adoption. The tool is the same. The modalities differ. The outcome diverges.

This is exactly why the emergent causal view gets it right. Markus and Robey (1988) showed that almost every real IS phenomenon is emergent. Technology does not determine outcomes, and organizations do not simply choose them. Outcomes arise from the interaction of technology, people, and context over time. Structuration is the mechanism underneath that emergence. The same tool in two contexts produces two different outcomes because the duality of structure plays out differently in each. Orlikowski made this concrete. DeSanctis and Poole made it testable. The trio together says: stop blaming the tool, stop crediting the tool, and start looking at how the structures of meaning, power, and legitimacy in a specific context shape what the tool becomes.

I think the biggest practical mistake organizations make is treating technology adoption as a procurement decision rather than a structuration problem. They buy the tool, train people on the features, and declare adoption complete. They measure login rates and call it success. But as Burton-Jones and Grange would argue, logging in is not using, and using is not using effectively. The duality of structure says that adoption is just the starting point. What matters is what happens next: whether the appropriation is faithful, whether the interpretive schemes align, whether the power structures support or suppress the new practice, and whether the normative order legitimizes the change. When those conditions are not met, the tool becomes something its designers never intended, and the technology-in-practice that emerges may look nothing like the spirit of the original system.

This also connects to what theory is and is not. Sutton and Staw (1995) warned that references, data, and variables are not theory. The duality of structure is theory. It specifies a mechanism, which is that structures shape action and action reproduces or transforms structures, mediated through three modalities. The theory predicts that the same technology in different structural contexts will produce different outcomes. It explains why. A framework like TOE, which Tornatzky and Fleischer (1990) proposed, identifies contextual factors but provides no mechanism. It tells you where to look, but not what you will find. Structuration tells you what to look for: the interplay of signification, domination, and legitimation as people appropriate the features and spirit of a technology into their existing practices. That is the kind of explanatory mechanism that turns a description into a theory.

The uncomfortable implication is that there is no such thing as rolling out a tool. There is only rolling out a tool into a specific structural context, where people with specific interpretive schemes, specific resource distributions, and specific normative expectations will appropriate it in specific ways. Change the context, change the appropriation, change the outcome. The tool is constant. Everything else is in motion.