# Case: build our own video classroom or keep paying per minute? > A make or buy decision in cash: two options, their incremental cash flows, the NPV and the assumptions that decide it. - Course: [Finance for Product People: From Unit Economics to Business Cases](https://productovi.com/courses/finance-for-product.md) - Module: Is this bet worth making? - Lesson 24/32, ~39 min - Source: https://productovi.com/courses/finance-for-product/build-or-buy-case ## The engineers' pitch **Alex Tutor:** Back at Klasna's product team: the engineers want to build their own video classroom. It's a make or buy decision. **Leo:** We pay the video vendor per minute: 1.5 million a year and growing. The engineers say they can build our own classroom for 2 million. It saves 4.6 million over five years and pays back in under three. Let's do it. **Hanna:** The vendor bills in dollars: USD 0.45 per lesson, 18.90 UAH at 42 hryvnias per dollar. Our own servers would cost USD 0.10 a lesson, 4.20 UAH, plus 600,000 a year to keep the platform running. **Sofia:** So the yearly saving is the vendor bill we no longer pay, minus hosting, minus upkeep. **Dataset (CSV):** [Video classroom: build vs keep the vendor, five years](https://productovi.com/datasets/finance-for-product/build-or-buy-case/klasna-video-build-vs-buy.csv) Year 0 is the six-month build; both options pay the vendor during it, so only the build cost is shown. Lessons grow by 12,000 a year. UAH at 42 per USD. - `year`: 0 = the build, now; 1–5 = years of running - `lessons`: 1:1 lessons that year - `buy_vendor_uah`: Vendor bill if we keep paying per minute - `build_capex_uah`: Building our own classroom, paid now - `build_hosting_uah`: Own servers and bandwidth (billed in USD) - `build_maintenance_uah`: Engineers' time, monitoring, on-call **Alex Tutor:** Year 1: 1,512,000 − 336,000 − 600,000 = 576,000 UAH saved. Do it for each year: the five savings add up to the 4.6 million on Leo's slide. > **Exercise:** Use the dataset. How much does building save in **year 3** (vendor bill minus own hosting minus maintenance)? **Alex Tutor:** Before we discount anything: how should the 4.6 million be weighed against the build? > **Exercise:** Leo's slide: "Building costs 2,000,000 UAH now and saves 4.6 million over five years." How should the savings be weighed against the build? ## Incremental cash flows: what belongs in the case **Hanna:** The engineering spreadsheet also subtracts depreciation, 400,000 a year, and the 300,000 prototype we paid for last year. **Sofia:** Should those come off too? **Alex Tutor:** NPV works on cash. The 2,000,000 leaves now, in year 0. Depreciation is the same 2,000,000 spread over five years in the P&L. Subtract both and you pay for the build twice: the NPV would fall to −785,185.79. **Alex Tutor:** Only **incremental cash flows** belong in the case: cash that changes because of the decision. The prototype is spent whatever we decide now: a sunk cost. It stays out. **Leo:** And the four developers? They could build the mock-exam mode instead. **Alex Tutor:** That one is real. If the build delays the mock exams, what that delay costs belongs in the decision, at least as a named risk. **Alex Tutor:** Leo's question about the developers deserves a clear answer. > **Exercise:** **True or false:** the four developers who would build the video classroom are already on Klasna's payroll, so their time costs the build nothing and can be left out of the decision entirely. > **Exercise:** Which items belong in the **cash flows** of "build our own classroom" compared with "keep the vendor"? ## NPV at the right rate **Alex Tutor:** Now discount the five savings at 25% and subtract the 2,000,000 in year 0. We stop at five years: after that the platform would need a rebuild anyway. **Sofia:** The factors are 0.8, 0.64, 0.512, 0.4096 and 0.32768. **Hanna:** And which rate is right? Cutting a bill sounds like efficiency, 25%. But building and running video ourselves is a technical risk we have never taken. **Alex Tutor:** Start with 25%, then we'll see what 30% does. > **Exercise:** Use the dataset. What is the **NPV** of building at 25%: the five yearly savings discounted, minus the 2,000,000 build cost? **Alex Tutor:** And how long is the 2,000,000 really out? > **Exercise:** The build costs 2,000,000 UAH now and saves 576,000; 752,400; 928,800; 1,105,200 and 1,281,600 UAH at the end of years 1–5. What is its **discounted payback period** at 25%, in years? Assume the saving comes in evenly within a year; round to two decimals. ## NPV sensitivity analysis: how sure is that number? **Sofia:** At 30% the NPV is only +43,175.71, and the IRR is 30.97%. Payback is 2.72 years, but discounted it's 4.31. **Alex Tutor:** So in today's money the 2,000,000 is out for over four of the five years. Now the assumptions. The 290,526 rests on lessons growing by 12,000 a year: the saving grows with every lesson. **Hanna:** The build cost matters too. Above 2,290,526, a 14.53% overrun, the NPV at 25% is gone. A 30% overrun gives −309,473.79. **Leo:** And the dollar? Everything here is in USD. **Alex Tutor:** The vendor is more dollar-heavy than our servers. If the hryvnia weakens 10%, the build looks better: +680,935.63. The currency helps; growth and the build cost decide. **Hanna:** One more assumption I'd test: the upkeep. > **Exercise:** In the base case the build has an NPV of +290,526.21 UAH at 25%, with maintenance of 600,000 UAH a year. What would the NPV be if maintenance turned out to be **800,000** a year in each of years 1–5? Round to two decimals. > **Exercise:** What is the build's NPV at 25% if lessons **stay flat** at 80,000 a year instead of growing? ## Which hurdle? **Leo:** So is it a 25% project or a 30% one? It changes the story. **Hanna:** The savings are fairly certain: we know the vendor price and our volumes. The risk sits in the build and in running video ourselves. **Sofia:** Then the rate is a judgment, and the memo should make it out loud. **Alex Tutor:** Exactly. Pick one, say why, and show what the other gives. A business case hides nothing the decision depends on. > **Exercise:** The build costs 2,000,000 UAH now and saves 576,000; 752,400; 928,800; 1,105,200 and 1,281,600 UAH in years 1–5: NPV +290,526.21 at 25% and +43,175.71 at 30%. Klasna has never built or run video in-house. In 2–4 sentences: should the video build be judged at 25% (efficiency, known costs) or 30% (a riskier bet)? Say why and what the NPV is at your rate. ## The memo for Leo **Leo:** I need a decision by Friday: build, or keep paying per minute? **Hanna:** From the cash flows, not the slide: what we save, what it's worth today, and what would make it wrong. **Alex Tutor:** And a verdict he can act on: yes, no, or yes if. **Sofia:** Numbers first, verdict at the end. > **Exercise:** Write the business case for Leo: build our own video classroom or keep paying per minute? > **Key ideas** > > - Compare the options by the **cash that differs**: vendor fees saved minus own hosting and upkeep (576,000 → 1,281,600 a year) against 2,000,000 now. > - Leave out **depreciation** (the build cost again) and **sunk** costs (the prototype); name **opportunity costs** (the delayed mock exams). > - **NPV**: +290,526.21 at 25%, +43,175.71 at 30%; IRR 30.97%. Choose the rate by the bet's risk and say so. > - "4.6 million saved" is undiscounted (2.29 million today); payback 2.72 years, discounted 4.31. > - **Sensitivity**: flat lessons −450,974.72; a 30% overrun −309,473.79; break-even build cost 2,290,526.21; a 10% weaker hryvnia +680,935.63. > - A business case ends with a decision and the conditions that would change it. **Alex Tutor:** That's the module in one decision: money now against money later, at a rate that matches the risk, with the assumptions on the table. - Previous lesson: [Hurdle rate and cost of capital](https://productovi.com/courses/finance-for-product/hurdle-rate.md) - Next lesson: [Driver-based forecasting](https://productovi.com/courses/finance-for-product/driver-based-forecast.md) --- Productovi Learning. Free, no sign-up needed to read.