Start with the CFO: why your first analytics wins should be financial
When a health system starts a new analytics program, there is a long list of good ideas. Quality, patient flow, readmissions, staffing, revenue. All of them matter.
My advice is to start with the ones your CFO cares about most. Not because money matters more than patients. It doesn’t. But the first year of an analytics program is about earning the right to do the second year, and nothing earns that right faster than a result someone can count.
This year the case is even stronger. Kaufman Hall reported that the median hospital operating margin closed 2025 at 1.3%, with bad debt, uninsured patients and charity care all rising. By March it had improved only slightly, to 1.7%. At those margins, your CFO is already hunting for money. Help them find it.
Why money first
Financial wins are easy to measure. If you recover underpayments from a payer, the cash shows up. If you cut contract labor on a unit, the invoices get smaller. There is a before and an after, and finance can check your math.
Clinical wins are just as real, but they are harder to prove quickly. A change in readmission rates takes months to show up and can have many causes. You will get to that work. You’ll do it better, with more support, once the program has already paid for itself.
Financial data is also easier to start with than most people expect. Claims, remittances, the general ledger, invoices, and payroll are structured and already reconciled by someone every month. That doesn’t make them clean. It means there is a person who knows what the numbers should be, and that person can tell you fast when you’ve gotten something wrong.
There is a quieter benefit too. When the CFO sees analytics recover real money, the program stops being an IT expense. It becomes something the finance team defends in budget season. That shift is worth more than any single project.
Where to look
Most health systems have the same handful of places where money leaks. You don’t need to find a clever new one. You need to pick one and finish it.
Underpayments are a good start. Payers don’t always pay what the contract says. Comparing what you were paid against what you should have been paid, claim by claim, often turns up money that is owed and easy to recover.
Denials are another. Most systems already track denial rates. Fewer can say which denials are worth fighting, which ones keep coming back for the same root cause, and which department can fix that cause upstream.
Supply spend is often overlooked. Prices drift away from contract terms, and items get bought off contract. Matching invoices to contracts is unglamorous work that tends to pay well.
Labor is the largest cost in most health systems, so even small changes add up. Overtime and contract labor are good places to look, because they can often be shifted with better scheduling and earlier warning.
Pick one. Not all four. The goal of the first project is a clean, believable win, not a broad program.
You’ll hear that AI agents can do this work now. Some of it, they can. An agent can compare each payment to the contract terms, sort denials by root cause, and draft an appeal for a person to review. That is real help on work that repeats and follows rules.
It doesn’t change the order of things. You still need a baseline finance trusts. A person still decides which appeals go out and which payer conversations to have. An agent can make a good first project faster. It can’t pick the project for you.
How to run the first project
Agree on the target with finance before you start. Write down the baseline: what this costs or loses today, using numbers finance already trusts. If you can’t agree on the baseline, you won’t agree on the result.
Keep the scope small enough to finish in about a quarter. Long projects lose sponsors. Short ones build a track record.
Put a finance person on the team as a working member who checks the logic each week, rather than a reviewer who sees it at the end. They will catch problems early, and when the result comes in, they will be the ones telling the CFO it’s real.
Then make the result routine. A one-time recovery is nice. A monthly report that keeps finding the same kind of money, owned by the revenue cycle or supply chain team, is much better. That is when analytics becomes part of how the organization runs.
Once you have a few financial wins, the conversation changes. You can go to the CFO and say: we found this money, and here is what we’d like to do next. That is a very different meeting from asking for budget on a promise.
That is when you take on the harder work, like throughput and quality. By then you’ll have the credibility and the funding to do it right.
This week, sit down with your CFO for thirty minutes and ask one question: where do you believe we are losing money that we can’t see clearly today? Write down the answer. That’s your first project.
Discussion