In the mid-2010s, a technological fever swept boardrooms around the world. The promise, championed by the software industry, was simple and seductive: pour all of a company's raw data into a massive storage infrastructure — then dubbed a "Data Lake" — and economic value would emerge spontaneously. It was the stubborn belief that accumulating technical resource would mechanically generate wealth. Millions of dollars were poured into enormous servers to centralise information. A few years later, analyst firms were documenting a resounding failure: the majority of these lakes had transformed into what they named "Data Swamps" — toxic, unnavigable zones where information accumulated without governance, generating massive costs with zero operational utility.
At the same time, Airbnb, in the midst of global hypergrowth, was facing a crisis of a different kind. The company had clean, accessible data, but its data science team was on the verge of suffocation. Every department — from marketing to human resources — was demanding reports and analyses to drive its activity. The experts had become a fatal bottleneck. Rather than hiring an additional army of analysts, Airbnb made a counter-intuitive bet. In 2016, the company created the "Data University", a tiered internal training programme. The goal was not to centralise data in the hands of an elite, but to make every employee capable of querying the databases themselves.
This face-off explores the critical boundary between owning a technical tool and mastering an organisational capability.
The data swamp phenomenon describes a situation where the IT department deploys a storage infrastructure and pours petabytes of raw files into it — integrated without metadata (the labels that explain what the data actually is), without an identified owner to guarantee its quality, and without access rules intelligible to end users. The result is a technical pile-up that is unusable by operational teams.
The R6 analysis identifies a major design error along the realisation and coordination axes:
Strategic level (S). The company invested all its capital in the pole of direct production (S3a): acquiring, collecting, storing the raw material. It completely ignored orchestration (S3b) — the capacity to organise flows, define interfaces, and make these assets exploitable by others. It is a strategy of pure storage logistics that ignores the engineering of usage.
Organisational level (O). The swamp is the clinical result of an absence of cooperation structuring (O2b). Without governance — Data Steward roles, shared data dictionaries — the technology was deployed into an organisational vacuum. The IT teams managing the server do not speak to the marketing teams who need the content. The software tool replaced the human architecture, generating an opaque and costly complexity.
Individual level (I). Engineers focused on the direct execution (I3a) of technical tasks — connecting digital pipes — without collaboration (I2b) with operational units being required or valued, leaving no one to understand the actual use cases the data was supposed to serve.
In 2016, only 30% of Airbnb employees were regular users of the company's internal data tools — significantly below comparable peers such as Facebook or Dropbox. Airbnb refused the ivory tower model where a technical elite holds the monopoly on knowledge and access. Faced with its analysts' paralysis, the company built a three-tier training curriculum: introductory courses on data-informed decision-making (accessible to everyone), intermediate classes on SQL and visualisation tools, and advanced courses in machine learning and Python aimed at engineers. Within months, 500 employees had completed at least one course. By the end of 2016, 45% of employees were regular users of the data platform — a 50% increase in under a year. Ad hoc requests to data scientists were drastically reduced.
The R6 analysis reveals a strategic reconfiguration of the realisation axis to unlock the coordination axis:
Individual level (I). Airbnb formally changed the mandate of its data experts. Instead of asking them to solve business problems themselves (I3a — direct execution), the company required of them a contribution through process mastery and tooling (I3b). Experts stopped producing reports and focused on creating simplified tools and delivering training. They moved from the status of exclusive producers to that of methodological facilitators.
Organisational level (O). This competence transfer activated genuine distributed autonomy (O2a). By equipping peripheral functions — marketing, product, finance — with analytical competence, the organisation eliminated the paralysing back-and-forth with the central department. Data became an operational fluid immediately available for local decision-making.
Strategic level (S). This approach validates an orchestration strategy (S3b). The company does not create value through the hyper-centralisation of its engineers, but through the systemic elevation of competence across its entire internal ecosystem.
The comparison between the Data Swamp and the Airbnb model illustrates two antagonistic conceptions of technological power: the logic of stock versus the logic of flow.
In the Data Swamp case, power is understood as material accumulation. The company considers itself rich because it owns a massive volume of data. Yet without human and procedural architecture to process it (O2b) and orchestrate it (O3b), that stock becomes a toxic liability — infrastructure costs, security risks, widespread internal frustration. The tool, deprived of a master to structure it, turns against the organisation.
At Airbnb, by contrast, power is conceived as an enabling flow. The value of data is zero when it sits on a hard drive; it only actualises at the moment it is consumed to inform a human decision. By shifting the interrogation competence from the technical centre to the operational periphery, Airbnb eliminated the friction. Where the Data Swamp creates sterile dependency on the machine and on experts, the Data University creates independence through the elevation of collective competence.
The technological transformation of an organisation never consists of buying more powerful software to compensate for human dysfunctions. It consists of using that software to make teams more autonomous and more effective.
The directive is categorical: stop building technological cathedrals if you are training no one to enter them. Investment in the tool must systematically be accompanied by a shift of competence from the centre to the field. Do not recruit experts to do the work in place of your teams — mandate those experts to build the methods and tools that will allow the entire organisation to rise. A truly high-performing company is not the one with the largest data centre. It is the one where every employee has both the technical capability and the structural permission to find the answer they need in order to act.