Finance × Analytics

A pipeline tracker and a cash calendar, two views of the same film

Revenue arrived as quotes and renewals negotiated over WhatsApp. Nobody could say what cash would land, or when.

FPS logs every quote, renewal, draft and potential deal: issue date, due date, days late, notified, replied, next action. An Apps Script pulls every open deal into one list. FDC is a day-by-day cash calendar: confirmed movements in black, forecasts in yellow, moved to actuals within 24 hours of payment.

The routine around it: sheet updates every working day, a one-hour call with the CEO on Monday, Wednesday and Friday, a written recap on Tuesday and Thursday, and a monthly outlook at the start of each month.

Result. Payment delays, inbound and outbound, down by about 30%. CV

What each client tier really cost to serve

I priced every service line by the hour: rent and running costs plus team pay, spread over the hours actually available. Then I multiplied by the hours each tier consumes per month across media buying, social strategy, sound direction, consulting and project management.

Cost to serve ate between 64% and 71% of list price, depending on the tier. Mid-tier clients were the most profitable; top-tier clients, the ones everyone wanted, were not. Model

Finding. The top tier had the thinnest margin of all in the month a deal closed: 10.9% once the closer's commission was paid. A repricing scenario of +20–25% lifted closing-month margin to 25–30% on every tier.

The hour budget: where the team was full and where it was idle

Each service line had an owner and a monthly hour cap. I enumerated every possible roster mix, written like 22-5-2 for entry-mid-top, computed idle capacity per line and intersected the feasible sets across lines, so the answer to “can we take this artist?” became a lookup.

Finding. In the November 2025 plan the team was booked for about 297 of 560 available hours, yet sound direction sat at 100% while influencer marketing and magazine placement sat under 5%. The constraint was one bottleneck, not headcount.

Same team, different mix: +22%

With hours per tier as constraints and monthly fee as the objective, I solved the roster as a small linear program. Moving from 10 entry, 10 mid and 5 top clients to 30, 5 and 5 raised monthly contribution by 22% under the same caps. Media buying, social strategy and visual direction all ended exactly at their limits.

The model used list price, not margin. The October cost-to-serve work above is what corrected it, and why entry-tier volume only holds up with repricing.

A business plan that became my thesis, and a hiring machine

The business plan: Italian recorded-music market from FIMI data, Porter's five forces, a market of about 30,000 emerging artists narrowing to 5,000–8,000 who buy professional services and 250 reachable in three years; a quarterly cohort model for churn, retention and acquisition; a 15-month P&L, a five-year projection, and a microcredit with a seven-year amortisation schedule to fund growth without dilution. Thesis

On the people side: job specs for four roles, about 100 applications screened across three pipelines, a 72-hour work-sample test (“organise the CEO's Monday”) scored on a weighted rubric, and the operating manual for my own successor in accounting.