At Giraux, I work with the owners on operating decisions and build the software behind them. These models connect that work to the cost of running it.
Giraux cost decision
When does keeping our own platform make sense?
I built Giraux's media platform when our Sirv plan was about to increase from $59 to $89 a month. The production system is used daily. This separate analysis uses the bills we have to compare keeping it with a managed service over the next 12 months.
93% lower
Recent Cloudflare invoice average versus the planned Sirv monthly price.
- Cloudflare, four recent invoices
- $6.20
- Sirv, planned monthly price
- $89.00
The SQL ledger reconciles all nine supplied invoice totals, including two issued in June. The Sheet separates the observed bills from assumptions about future charges, maintenance hours and switching costs. Change an assumption to see when the lower-cost option changes, then check whether that option meets the service requirements.
Service capacity matters too. The planned $89 tier lists 40 GB of storage, while the migration inventory was 46 GB. The model keeps that mismatch visible before treating a vendor plan as an equivalent option.
Evidence and open questions
The invoice screenshot shows paid totals and issue dates. It does not show service periods, product line items or which charges belong solely to this platform. The ledger groups charges by invoice issue month.
The Sirv prices come from a supplied plan screenshot and my account of the planned upgrade. The $89/month and $979/year payment options stay separate. Sirv invoices are unavailable.
Prepared with AI assistance; inputs and formulas are available in the downloads. Maintenance hours, future usage and switching costs are assumptions. Historical build time is separate from the forward decision. A like-for-like service comparison still needs confirmed storage and delivery requirements.
Uber R&D forecast
How does the spending outlook change with the assumptions?
Six quarters of Uber's public financial filings support a forecast for the second half of 2026. The workbook separates stock-based compensation from other R&D expense, compares three spending scenarios and reconciles the historical inputs to their sources.
R&D is a company-level measure. It gives context for a planning exercise; Uber's internal Tech Finance budget and hiring plans are outside the data. R&D excluding stock-based compensation still includes noncash expense.
Independent portfolio work. No affiliation with Uber or access to its internal financial data.
Walk through the models
Upload either workbook to Google Drive and open it with Google Sheets. Keep a copy of the starting assumptions before changing them.
- Start with one invoice. Follow its issue month into the SQL ledger and reconcile the total. Check why both June invoices belong in the ledger.
- Change the assumed maintenance hours. Explain what happens to the cash comparison and modeled economic cost, then find the point where the lower-cost option changes. Check the separate eligibility result before making a recommendation.
- Check whether the selected vendor plan meets the storage requirement. Explain why a lower price alone is insufficient.
- In the Uber model, change the spending scenario and trace the full-year result back to historical actuals and the two forecast quarters.
The source download includes the Giraux rebuild instructions and checks. The workbooks preserve their source notes and assumptions.