Should Costing - Why and How to Perform

Doesn't sugarcoat the shortcut mindset; it frames should costing as work you choose because the math matters, not because it's comfy. From a TeamLead lens, the value was aligning PMs and engs around cost drivers so PR debates don't drift, and keeping infra choices honest before they hit prod or k8s scale. The BOM sanity check in Section 2.4, where he walks a fastener cost up from raw material to margin, stuck with me; we used a similar table in automotive sourcing, and the spreadsheet layout was immediately reusable. I wasn't sold on the early history tangent, and I wished there was more on tying should-cost models into CI gates without overfitting RPS assumptions. Applied it the next week to a vendor quote review, and the team caught two padding items without torching the relationship. The win wasn't new facts so much as reconciling what I knew with what I assumed, which changed how we ask for numbers.

The course starts grounded in real constraints, not academic math, and ties should costing to decisions engineers actually make under budget and schedule pressure. The teardown/BOM walkthrough in Chapter 3, especially the labor rate sensitivity table around 18:30, stuck because it showed how a 5% assumption swing ripples through unit cost. I wasn't sold on the early definitions—they lingered—but the parametric model section clicked, and I've since used that framing to simplify a gnarly pricing path in prod and a PR touching our arch.

The angle on testability went further than expected, especially how cost assumptions get treated like code paths you can actually exercise. The clean-sheet BOM walk-through in the “Material Drivers vs Process Drivers” section stuck with me; mapping cycle time to cost felt like reviewing a PR where the arch finally lines up with infra reality. Coming from legacy spreadsheets and tribal knowledge, the way it framed sensitivity checks as CI gates made sense, even if the beginner pacing lingered a bit on definitions. It's mostly practical, though I wasn't sold on the quick detour into vendor quotes without more on variance handling. Between meetings, I kept thinking how this would’ve helped earlier automotive programs where prod RPS targets were set before obs existed. I've been bridging old cost models to modern workflows for a while, and this gives a shared reference—one I’ve been missing the last couple years.