Why Multi-Site Health Organizations Can't Trust Their Own Numbers
September 24, 2026 Tony Terranova Reimagined Health
It's Monday morning and the leadership team is reviewing last week's performance. The COO has one census number. The regional director has a slightly different one. Finance has a third. Nobody is wrong, exactly. Each number came from a real system, pulled by a capable person, using a reasonable method.
The meeting spends twenty minutes reconciling figures and five minutes making decisions.
If this sounds familiar, the problem isn't your people. It's the way data moves, or fails to move, across a multi-site organization. It rarely gets fixed on its own.
How it happens
Most multi-site health organizations didn't design their data environment. It accumulated.
Growth brings inherited systems. Each acquisition or new location arrives with its own EHR, or its own configuration of a shared one, plus its own intake workflows and its own habits. On paper everyone is on the same platform. In practice, "admission date" might be captured three different ways across twelve sites.
Definitions drift. What counts as an active patient? When does a length of stay start? Does a transfer between sites count as a discharge? Each site answers these questions sensibly for its own needs. Those answers stop matching once you try to roll them up.
Spreadsheets fill the gaps. When a system can't produce the report leadership needs, someone builds it by hand. That spreadsheet becomes critical infrastructure, even though it lives on one person's laptop and has no documentation. The person who built it becomes the only one who can explain it.
Reporting runs on exports, not pipelines. Data gets pulled, pasted, cleaned, and re-pulled. Every manual step adds a chance for a filter to be missed, a date range to shift, or a formula to break. It also means the numbers are already stale by the time anyone sees them.
None of these is a failure on its own. Together, they produce an organization where every number comes with an asterisk.
Why it costs more than it looks
The obvious cost is time. Skilled analysts spend their weeks reconciling instead of analyzing, and leaders spend meetings debating whose number is right.
The less obvious costs are bigger.
Decisions slow down. When leaders don't trust the data, they wait for another report, another check, another month of trend. Staffing adjustments, site-level interventions, and expansion decisions all get pushed back.
Trust erodes quietly. After enough conflicting reports, people stop looking at the data and fall back on instinct and anecdote. The dashboards still exist; they just stop influencing anything.
Comparing sites becomes impossible. One of the biggest advantages of operating many sites is learning which ones run best and why. That only works if you're measuring the same thing everywhere. Without consistent definitions, a site that looks like an outlier might just be counting differently.
External reporting gets riskier. Board packets, accreditation, and regulatory submissions all depend on figures someone has to stand behind. "We think this is right" is an uncomfortable position.
Why another dashboard won't fix it
The instinctive response is to buy or build a better reporting tool. But a dashboard is only a view on top of the data underneath. If the underlying numbers disagree, a new dashboard presents the disagreement more attractively.
The real fix sits a layer below the visuals, in how data is collected, defined, and governed before anyone builds a chart.
What fixing it actually looks like
Organizations that get this right tend to follow a similar path.
They agree on definitions first, and the business owns them. Before any technology work, operations and clinical leaders decide what "census," "admission," and "active patient" mean across the whole organization, and they write it down. This is a leadership conversation, not an IT task, and it's the step most often skipped.
They replace exports with pipelines. Data flows directly from source systems into one central platform on a schedule, with the same logic applied every time. No copy-paste, no personal spreadsheets in the critical path.
They put everything on one governed platform. Instead of each site or department keeping its own version of the truth, all the data lives in one place with consistent rules, access controls, and definitions built in. Modern lakehouse platforms, such as Databricks, are built for exactly this. They can bring clinical, operational, and financial data together at scale while keeping governance and security intact. That matters when you're handling protected health information across many locations.
They make every number traceable. When a leader asks where a figure came from, anyone should be able to follow it back to the source system and the rule that produced it. Traceability is what turns a number into something people will act on.
They start with the numbers people argue about most. You don't need to fix everything at once. Pick the three or four metrics that cause the most debate in leadership meetings, get those consistent and trusted, and expand from there. Early wins build the credibility the larger effort needs.
A quick self-check
If you lead operations at a multi-site organization, ask yourself:
If I asked three people for last week's census, would I get the same number?
Could someone trace our most important KPI back to the system it came from?
How many of our leadership reports depend on a spreadsheet that one person maintains?
Can we compare performance across sites with confidence, or do we mostly explain away the differences?
How long does it take to answer a new question from the data: hours, days, or weeks?
If the answers make you uneasy, you're not alone. Most growing health organizations hit this wall. The ones that break through don't do it with a better chart. They do it by building a foundation their numbers can stand on.
Reimagined Health helps multi-site health organizations build unified, governed data platforms so leaders can make decisions with confidence. If your team spends more time reconciling numbers than acting on them, let's talk.