An ownership group was evaluating a proposed healthcare services venture with a potential operating partner. Working with Claude, I built a financial model in Excel that let the owners test proposed terms during negotiations and see how those terms affected the investment case. The project did not proceed beyond negotiations. The modeling helped explain why: the available terms did not produce returns that justified the investment for the ownership group.
Identifying details have been omitted or generalized to preserve confidentiality.
How I work: I find the problem or opportunity, determine what capability is missing, build it, put it to work and measure what happens.
A promising partnership with unresolved economics.
An ownership group was exploring a healthcare services venture with a potential partner. The proposed arrangement brought together complementary resources and capabilities.
The strategic fit looked promising. The parties, however, held different assumptions about patient volumes, ancillary-service activity, service revenue, and operating costs.
The proposal required a substantial upfront investment and funding through the opening ramp. The owners needed to know whether the full package of terms justified that commitment.
Individual terms obscured the whole investment.
Each proposed concession affected more than one part of the business. Cost-sharing arrangements changed contribution margins. Clinician start dates affected both revenue and staffing costs. Fixed operating costs weighed more heavily during the opening ramp than at full capacity.
Without an integrated model, it was hard to evaluate those terms together. Revised proposals required offline analysis, slowing negotiations and making it difficult to tell whether a workable agreement was emerging.
The missing tool was a shared view of how the proposed terms flowed through to net operating income.
Make the economics usable in the room.
I set out to build a calculator detailed enough to hold up to scrutiny and fast enough to use during a negotiation.
Assumptions would be grounded in published sources, real operating data, or stated counterparty positions. Missing inputs and provisional assumptions would be identified explicitly.
The model would also show the revenue the venture could generate for its partner, giving the discussions a clearer view of the economics for both sides.
Connect clinical activity to investment returns.
I built the model with Claude. A first working version, a revenue model by practitioner with monthly, quarterly and annual rollups, took about two to three days. The full negotiation model grew over two to three months alongside the negotiations, through more than a dozen versions. Claude generated and revised the Excel workbook using Python. The workbook used live Excel formulas linked to editable inputs, allowing the owners to change assumptions and see the financial effects without running AI.
Excel was deliberate. The owners already worked in it, and it ran on a call with no setup.
One calculator that carried the key financial terms from clinical activity through revenue, costs, and the opening ramp to net operating income and the investment case, with the inputs a negotiation would change kept adjustable.
A change made in the negotiation flowed through revenue, costs, and the ramp to net operating income and the investment case, so its effect could be seen before the conversation moved on.
Use the model to test the deal.
As new positions came in, I used Claude to research the underlying assumptions and revise the affected parts of the model, often the same day. Where the parties' assumptions differed, the model carried alternative scenarios so their financial effects could be compared. I consolidated the key negotiation inputs onto one tab, which the owner used live in the discussion that informed the decision not to proceed.
A running input register tracked information still outstanding from the counterparty and the assumptions used in its absence. Scenario testing showed how sensitive the results were to those missing inputs.
When assumptions were changed to test the owners' target returns, I documented those changes and what they implied.
The calculator served as a negotiation tool, and its limitations were stated explicitly.
A clear decision not to proceed.
The owners came away with more than a decision. They have a working model of how a venture like this earns, or fails to earn, its return, and a disciplined way to test the next partnership before committing capital to it.
A venture development engagement can create value before a business opens. Here, the work gave the owners a clear understanding of the proposed investment and a sound basis for declining it.
I directed the work and made the judgment calls; Claude did the research, built the formulas and ran the checks. I verified the work myself by reviewing original sources, checking selected calculations, reconciling totals and testing scenarios. Working with Claude, I developed the first working revenue model in two to three days and continued expanding it as negotiations evolved, often turning revisions around the same day. The judgment still came from selecting credible inputs, making uncertainty visible, and explaining what the results meant for the owners' decision.
It's worth twenty minutes. No pitch, just a straight read on whether I can help and how.
Start a conversation