Building a data strategy for multi-site health systems

Why multi-site data strategies stall

Most multi-site health systems do not lack data. They lack a shared way to use it. Clinical, financial, and operational teams each run their own platforms, their own data structures, and their own definitions of success. When leadership asks for a system-wide answer, the honest reply is often that the numbers do not reconcile across sites.

The result is predictable. A strategy document gets written, a target architecture gets drawn, and twelve months later little has changed. According to Hutchins Data Strategy, most failures are structural rather than technical: operational leaders who would actually execute the strategy were excluded from shaping it, and the roadmap ignored competing budget and political realities.

A workable strategy starts somewhere else. It starts with the decisions being made poorly today because the right data is unavailable, low quality, or simply unused.

Start from decisions, not architecture

Before you buy anything, list the decisions that currently go wrong across sites. Examples: which service line to staff next quarter, which patients are drifting out of a care gap, which contracts are losing money. For each decision, name the data required, the owner, and the current failure mode.

This reframes the strategy as a plan to fix specific problems rather than a document about an ideal future state. It also gives you a natural prioritization order: fix the decisions with the highest cost of being wrong first.

Build the three systems that make data-first work

Health Catalyst describes three systems that underpin a data-first transformation strategy:

• Best practices — what should we be doing? Identify root causes and validate evidence-based standard work.
• Analytics — how can data accelerate improvement? Capture correct data, integrate it, and grant broad access so you can see gaps between best practice and actual results.
• Adoption — how do we transform? People learn the required skills, knowledge, and attitudes, then execute the change.

A strategy that only addresses analytics will produce dashboards nobody uses. A strategy that only addresses adoption will produce enthusiasm without evidence.

Solve interoperability and governance together

Interoperability is the ability of different systems to access, exchange, integrate, and cooperatively use data across organizational boundaries. HIMSS breaks it into four levels: foundational connectivity, structural format, semantic meaning, and organizational governance. Most multi-site systems get stuck at level three or four.

Governance is where the levels get enforced. data.world notes that governance supports consistency across systems by applying standards such as Fast Healthcare Interoperability Resources (FHIR). Without that discipline, each site keeps its own definitions and the semantic layer never converges.

A practical checklist for this stage:

• Name a single accountable owner for enterprise data definitions.
• Publish a shared data dictionary for your top ten metrics.
• Adopt FHIR-based exchange where your vendors support it.
• Document who can access what, and why, at each site.
• Review definitions quarterly, not annually.

Align the strategy with business drivers

Info-Tech Research Group recommends a sequence that keeps the strategy tied to outcomes: identify business drivers and value opportunities, map business outcomes to processes, analytics, and data, define metrics that measure the impact of the strategy, map technical capabilities to those processes, establish an organizational structure and governance framework, and then communicate the strategy.

That order matters. If you map technology first, you end up with tools looking for problems.

A short implementation sequence

1. Interview operational leaders at each site and record the decisions they get wrong today.
2. Rank those decisions by cost of error and frequency.
3. For the top five, document required data, current source, and quality gaps.
4. Assign an executive sponsor and a working owner for each.
5. Stand up governance for shared definitions before building dashboards.
6. Deliver one working analytics product per decision, then expand.
7. Review adoption monthly and retire anything nobody uses.

Common failure modes to avoid

• Writing a strategy that describes a target state without naming the operational problems it solves.
• Excluding site-level operators from the design, then expecting them to adopt it.
• Letting vendor agendas drive technology recommendations.
• Treating governance as a compliance exercise rather than the mechanism that makes shared metrics possible.
• Building dashboards before definitions are agreed.

How the Featured Expert Can Help

ArcadientIQ LLC is a healthcare data analytics consulting firm specializing in Tableau, Alteryx, SQL, and business intelligence solutions. They provide project-based consulting to help organizations improve reporting, automate workflows, and gain operational visibility without building an internal analytics team. Their work covers business intelligence, healthcare analytics, value-based care reporting, executive dashboard design, KPI reporting, data integration, and analytics strategy. You can learn more at arcadientiq.com.

Quiz: test your understanding

A multi-site health system has connected its EHRs and can exchange messages between sites, but each site still defines "length of stay" differently, so enterprise reports do not reconcile. Which level of interoperability is missing?

• Foundational
• Semantic
• Organizational