Measure your analytics program in dollars, not dashboards
Ask most health system data teams how they’re doing and you’ll hear about output: dashboards built and tickets closed.
Those numbers feel like progress. They aren’t the point.
Your analytics program exists to help the health system save money and run better. If it does that, it’s worth what you spend on it. If it doesn’t, the number of dashboards doesn’t matter.
This year makes that test harder to dodge. Kaufman Hall reported in August that hospital margins are running below 2025. Bad debt and charity care keep growing, and supply and drug costs are rising well above inflation. When margins are that thin, every budget line gets asked what it returns.
My advice is simple. Measure the program in dollars.
Why output is the wrong scoreboard
Output is easy to count, so it becomes the goal. A team judged on dashboards will build more dashboards. A team judged on tickets will close tickets fast. Neither one has to ask whether anything changed in the business.
I’ve seen this pattern in most health systems I’ve worked with. The data team is busy and proud of its work. The CFO, meanwhile, can’t name one thing it did that showed up in the budget. Both sides are frustrated, and both are partly right.
The fix is to change what you count.
What a dollar scoreboard looks like
Start with a short list of the work that matters most. For each item, write down the money it is supposed to move and who owns that money.
A few examples:
- Supply chain: savings from moving spend to contracted items, or from catching price errors on invoices.
- Revenue cycle: cash recovered from denials that were appealed or prevented.
- Operations: overtime or agency labor avoided because staffing matched demand.
- Pharmacy: less waste on high-cost drugs.
Each line has a number, an owner outside the data team, and a way to check it. The owner matters most. If the VP of supply chain agrees the savings are real, they’re real. If only the data team says so, they’re a guess.
Put AI on the same list. AI agents are coming into back-office work fast. At its August user conference, Epic showed its agents for clinicians, patients and the revenue cycle working as one system. An agent that drafts appeals or sorts invoices is worth exactly what it moves in dollars. Give each one a number, an owner and a check, like any other project. The agent can do the work. A person outside the data team still decides whether the savings are real.
Then report it the way finance reports anything else. Monthly, in the same format each time, with booked savings kept separate from projected savings.
Some of your numbers will be smaller than you hoped. A few projects will show nothing at all. That’s useful. It tells you where to stop spending time.
You’ll also find that credit is messy. When denials drop, was it the new report, the new hire in patient access, or a payer changing its rules? You won’t always know. Be conservative. Claim a fair share, show your math, and let finance check it. A modest number everyone trusts is worth far more than a big one nobody believes.
Not everything fits on this scoreboard. Regulatory reporting and basic data plumbing still have to happen. Call that work what it is: the cost of running the business. Keep it lean and keep it separate. Don’t let it hide inside the program you’re asking leaders to judge on value.
Once you start counting this way, the conversations change. Instead of “can you build me a report,” requests start to sound like “we think we’re losing money here, can you help us find out how much.” That’s a better request. It comes with a reason and an owner.
Choosing between projects gets easier too. When two efforts compete for the same analysts, you pick the one with the clearer dollar case. Nobody has to win an argument about whose dashboard matters more.
The data team’s standing changes as well. A team that can say “we helped recover this much last quarter, and finance signed off on it” gets budget. A team that says “we shipped forty dashboards” gets asked why it costs so much.
Where to start
You don’t need a new framework or a new tool. Pick your three biggest analytics efforts. For each one, answer these in writing:
- What dollar outcome is this supposed to move?
- Who outside the data team owns that outcome?
- How will we know, in money, whether it worked?
If you can’t answer them for a project, that’s your first finding. Either sharpen the goal or pause the work.
Then put the answers in front of your CFO next month and ask them to hold you to it.
Discussion