![Cleveland skyline](/sites/g/files/omnuum10826/files/styles/hwp_21_9__1920x825/public/2026-08/AdobeStock_1047363299%20Cleveland.jpg?itok=3BlubDBP) # Cleveland Data Governance Relies on Hub-and-Spoke Model Chief Innovation and Technology Officer Elizabeth Crowe explains how Cleveland’s data system, in which the IT department owns the technology but has strong ties to other agencies, both maintains standards and empowers staff. August 24, 2026 [ Stephen Goldsmith ](/stephen-goldsmith) *This article originally appeared in Government Technology.* Cities building a modern data enterprise face a question of centralization: Expensive enterprise software, data warehouses, BI tools and GIS platforms pay off when standardized, but the people who actually use that software sit in dozens of agencies, each with its own operational reality. Try to centralize too hard, and IT becomes the bottleneck; agencies route around it with shadow systems. Decentralize too much and create confusion and cross-agency obstacles due to inconsistent systems, definitions and data structures. Cleveland’s Chief Innovation and Technology Officer (CITO) [Elizabeth Crowe](https://www.govtech.com/workforce/cito-elizabeth-crowe-gets-cleveland-braced-for-the-future) resolves this tension by providing strong top-down leadership with the mayor, viewing the centralization balance as a rubber band that flexes in and out, rather than a policy decision with a right-or-wrong answer. Cleveland runs a hub-and-spoke model. A small central team — the hub — owns the enterprise choices: Power BI as the standard analytics layer, Esri for GIS, a single data platform, and the engineering standards, metadata and integration protocols that make those choices coherent. The spokes sit inside agencies. “We govern the tools they’re using,” Crowe explained. “We require a central analytics infrastructure, but the spokes sit with the \[agency\] directors.” A monthly forum where agency “data leads” meet to exchange insights and share what they’re building maintains the shape. These leads report to their department heads, with a dotted line to central innovation. Without that, decentralization would drift into fragmentation. With it, central standards land as peer practice rather than IT mandate and the hub gets early signals on where the next investment needs to go. The hub holds what only the hub can — standards, infrastructure, integration. Everything else stays close to the work. This model has proven especially effective in terms of property data. In Cleveland, 15-plus city and county systems reference parcel data — tax, code enforcement, planning, water and the land bank. Pre-standardization, a police officer serving a warrant or a building inspector responding to a complaint would each pull from different systems, with separate definitions of place and info gaps. To address this, Cleveland developed Property Insights, a geospatial AI tool designed with a single-pane-of-glass view to share information across the enterprise. Property Insights consolidates all sources into a single platform with standardized definitions and relationships, using Esri for the user experience. Departments use the same underlying view for entirely different operational decisions — economic development looking at vacant lots, housing flagging nuisance properties, public works planning right-of-way work — and the CITO’s team does not dictate workflows. Different users get different access tiers and agency-facing user interfaces without forcing them to go through central decision-makers. Crowe credits the hub and spoke with avoiding the complexity typical in centralized systems, which deters users, while maintaining standardization. “Sometimes we bump into data problems that are process problems,” Crowe said. For example, GeoAI tools will let non-coders ask spatial questions in natural language only if the IT leadership has made the underlying data searchable. An IT-housed analytics group tends to diagnose everything as a data problem and fix the data. But a slow permitting workflow with messy underlying records is a process problem first. Cleveland pairs data scientists with process analysts so that agencies can determine whether a problem lies in the data, the workflow or both. Distributing diagnostic capacity is itself a form of decentralization. The problem definition does not reside in one place, but the underlying data layer stays centrally governed. The Cleveland process also started carefully, mapping existing tools, costs and usage. “Imposing a governance framework before mapping the landscape almost guarantees it will be wrong,” Crowe said. “We just started by striking off wins, and now we’re going through the maturation of it.” Each dashboard, each integration, each tool rollout was scoped as a product with a timeline. Governance documents trailed the wins and codified what worked, rather than trying to specify it in advance. Cleveland’s success comes from building the infrastructure to manage centralization and decentralization at once: centralized enough to guarantee standards, consistency and strong data engineering, and decentralized enough to keep domain expertise, operational agility and the actual decisions close to the work. ## About the Author ### Stephen Goldsmith ![Headshot of Stephen Goldsmith](/sites/g/files/omnuum10826/files/styles/hwp_1_1__100x100_scale/public/datasmart/files/goldsmith_headshot_2018.jpg?itok=_stVEJro) Stephen Goldsmith is the Derek Bok Professor of the Practice of Urban Policy at the Harvard Kennedy School and the director of Data-Smart City Solutions at the Bloomberg Center for Cities at Harvard University. He previously served as the mayor of Indianapolis and deputy major of New York City. [Read Professor Goldsmith's full bio here](/stephen-goldsmith). See also:- [ Artificial Intelligence ](/topics/artificial-intelligence) - [ Civic Data ](/topics/civic-data) Share on:- [ Facebook ](#) - [ Twitter ](#) - [ Linkedin ](#)