Healthcare CFO Dashboard in Power BI: Metrics, Data Sources, and What Actually Gets Used

Most CFO dashboards fail before they’re ever used. Not because of the tool, not because of the data — because whoever built it tried to show everything. Every account, every variance, every department, all on one screen.

In a healthcare organization the failure has a second, more specific mode: the dashboard shows “revenue” from the practice management system, the general ledger shows a different number, nobody can explain the gap, and the CFO stops trusting the dashboard entirely. That’s a data-layer problem, not a Power BI problem — but it kills Power BI projects all the same.

A functional CFO dashboard is a decision support tool. It answers the 6–8 questions leadership checks weekly and nothing else. Build around that constraint and you’ll build something people actually open.

What a Healthcare CFO Dashboard Should Actually Do

Three questions it should answer every morning:

1. Are we collecting what we earn — and how fast are denials and aging AR eroding it?
2. What is our cash position, and how long does it last at current burn?
3. Which locations, payers, and providers are performing — and which are off plan?

If the dashboard can’t answer those in a single view, it’s not doing its job. Everything else — claim-level detail, provider schedules, invoice-level expenses — belongs on drill-down pages, not the executive summary.

The 6–8 Core Metrics

These are the metrics that earn a spot on the primary view of a healthcare CFO or controller dashboard:

Denial-adjusted net revenue. Gross charges less contractual adjustments, less denials and write-offs, versus budget and prior year. Gross charges are a vanity number; adjusted net revenue is what pays payroll. Show the variance indicator, not just the dollar figure.

Days in AR — total and by payer. A blended 45 days can hide 35 from commercial payers and 65 from Medicaid. Break it out by major payer and trend it monthly; direction matters more than the snapshot.

Net collection rate. Collections divided by net billable charges (after contractual adjustments). This is the single best answer to “are we keeping what we earn?” Healthy practices generally target the high-90s percentage range industry-wide; anything persistently below that deserves a denials-and-AR investigation, not a shrug. Pair it with contract variance — actual reimbursement against your negotiated fee schedule by payer and procedure code. A 96% collection rate can still conceal systematic underpayment on specific codes, and contract variance is what turns that from suspicion into a recoverable dollar list.

AR aging with movement. Buckets (0–30, 31–60, 61–90, 91–120, 120+) in dollars and as a share of total AR. The 90+ bucket is where revenue goes to die — and an aging table without a month-over-month variance column misses the only question that matters: improving or deteriorating?

Denial rate and top denial reasons. Sourced from clearinghouse remittances (835 files), broken out by payer and reason code. Front-end denials (eligibility, missing authorization) and back-end denials (coding, medical necessity) point at different fixes and different owners.

Cash vs. accrual view. Your PMS lives in an accrual-ish world; the bank lives in cash. Show both: collections posted this month, actual cash in, cash out, and implied runway. For any group that isn’t cash-positive every month, this is non-negotiable.

Labor cost per visit (or per RVU). Payroll divided by encounters or productivity. Neither your EHR nor QuickBooks can produce this alone — it requires joining payroll data to visit data, which is exactly why it’s one of the most valuable metrics on the dashboard.

Per-location P&L. For a 3–20 location group: production, collections, labor, occupancy, and allocated overhead by site. This is usually a page-2 view the leadership team reviews weekly or monthly — not daily-page material. Read every site’s margin against its case mix — acuity and payer mix by location. Without that, a site treating a heavier or more Medicaid-weighted population looks mismanaged rather than different, and the ranking produces the wrong management action. If any of your contracts are value-based, add the attributed-lives and quality-threshold view here too; shared-savings exposure belongs on the CFO dashboard, not in an annual reconciliation letter.

What actually gets used: metrics one through six live on the executive summary as cards and trends. Labor per visit and location P&L live one drill-down away. If everything is on page one, nothing is.

The Data Sources — and Why the Joins Are the Project

EHR / practice management (athenahealth, eClinicalWorks, NextGen, Epic, DrChrono): charges, encounters, provider productivity, schedules. Access is via API, a reporting database, or scheduled exports depending on the vendor.

Clearinghouse remittances: 835 files are the payer’s version of the truth — adjudicated amounts, denials, and adjustment reason codes (CARC/RARC). Without parsed remittance data, your denial rate is an estimate and your net collection rate is a guess. Turning 835s into tables Power BI can query is the step most internal builds skip, and it shows.

QuickBooks / accounting: the general ledger, cash accounts, operating expenses, and the write-off entries that reconcile PMS production to booked revenue. QuickBooks Online connects via API; Desktop typically means scheduled exports or an intermediary.

Payroll (ADP, Gusto, Paychex): labor cost by location, provider, and FTE. For most groups, a scheduled export to a location Power BI can read is the practical approach.

None of these systems join to each other. Labor cost per visit is payroll divided by EHR encounters. Denial-adjusted revenue is PMS charges reconciled against 835 adjustments and GL write-offs. That join work belongs in a data layer between the sources and Power BI — typically a warehouse (PostgreSQL, or ClickHouse at heavy volume) with transformation logic in dbt that enforces one definition of “net collection rate” across every location. Skip that layer and you’re back to two revenue numbers arguing in every leadership meeting.

For the close process that feeds the accrual side of this dashboard, see our guide to healthcare financial reporting automation.

Power BI Connection Realities

Refresh: scheduled refresh (daily overnight, hourly for cash and AR if needed) covers CFO use. Real-time streaming is rarely justified for financial dashboards.

Gateway: any source behind your firewall — a practice management server, a local database — requires the Power BI On-premises Data Gateway.

Licensing: Power BI Pro is $10/user/month; Premium Per User is $20/user/month. Manageable for a finance team of three; it scales up meaningfully when you share with practice managers and location administrators across a multi-site group.

HIPAA: claim-level and remittance data is PHI. Microsoft will sign a BAA covering Power BI Service, but row-level security, tenant export settings, and sharing rules need to be configured like you mean it. The simpler and more defensible pattern: aggregate to location/payer/provider level before data reaches Power BI, so the semantic model carries minimal PHI in the first place.

Layout: Structure Determines Whether It Gets Used

Page 1 — executive summary, single screen, no scrolling.

• Top row: four KPI cards — net revenue MTD vs. budget, net collection rate, days in AR, cash position. Each with a variance indicator (green/yellow/red) measured against budget or target, not against last month.
• Middle: 12-month trend of collections and denial-adjusted net revenue on the same axis. Seasonality (South Florida groups know this one well), payer shifts, and collection deterioration all show up here.
• Bottom: AR aging table with month-over-month movement columns, and denial rate by top five payers.

Pages 2+ — drill-downs. Location P&L, payer-level days in AR, denial reason detail, labor cost per visit/RVU by provider. Reach these by drilling through from a summary tile, not by navigating a tab bar.

Keep a paginated path for fixed-format output. Interactive dashboards are how leadership explores; board packs, lender reporting, CMS quality submissions, and payer remittance reconciliation need pixel-perfect fixed-format documents. In Power BI that’s Report Builder, not Desktop — a separate skill set worth confirming your partner actually has, since most dashboard developers never touch it.

Avoid dual axes. If two metrics must appear together, use a combo chart with clear labels or separate panels.

Common Mistakes

Two revenue numbers, no reconciliation. PMS production will never equal QuickBooks revenue — timing, write-offs, and cash vs. accrual guarantee it. Reconcile the bridge before anyone sees the dashboard, or the first question in every meeting will be “which number is right?”

Too many pages. A 12-tab report means nobody made hard decisions about what matters. CFOs don’t click through eight tabs.

Manual refresh. If someone has to run an export or paste a spreadsheet to keep the dashboard current, it will be stale within weeks. Tie financial pages to the close calendar and let operational pages refresh nightly — automatically.

Numbers without context. “$1.2M collections” means nothing next to no benchmark. Every KPI needs budget, prior period, or target built in.

Denial data from the wrong source. PMS adjustment summaries are approximations. The 835 remittance is the actual adjudication. Build denial reporting on remits or don’t build it.

When Power BI Is the Right Choice (and When It Isn’t)

Power BI is a strong fit when the organization is already Microsoft-native — Azure, Microsoft 365, Teams — and the finance team will maintain it. Our Power BI development practice covers exactly this: the data layer, the semantic model, and the executive dashboard.

It’s a weaker fit when per-viewer licensing across many locations gets expensive, or when the organization’s HIPAA posture says financial data with PHI adjacency should never leave its own infrastructure. In those cases Metabase — self-hosted, no per-viewer fees — or Apache Superset deliver the same dashboard from a warehouse you own.

And when the real bottleneck is upstream — 835 parsing, EHR exports, payroll joins — no BI tool fixes that. The pipeline is the project.

Getting Started

Define the 6–8 metrics first. Before opening Power BI, write down the questions the CFO checks weekly. Those become the executive summary. Everything else waits.

Connect one source at a time. Reconcile each against its source system before connecting the next. Data quality issues are much harder to debug across four connected sources than across one.

Schedule automatic refresh from day one. A stale dashboard is worse than no dashboard.

Start narrow and earn trust. A dashboard that answers three questions accurately gets used. One that answers everything but has two metrics that are occasionally wrong gets abandoned.

If you’re building a CFO dashboard on healthcare data — remittances, multiple EHRs, payroll, and a general ledger that has to reconcile to all of it — reach out to discuss what you’re building. We scope and build custom financial dashboards for multi-site healthcare organizations that have outgrown spreadsheets and out-of-the-box connectors.