i spent nights combing the community library of market-sizing case studies and found patterns that repeat by industry: retail = population × frequency × spend, saas = number of target companies × penetration × arpa, hardware = units × replacement cycle × price. seeing dozens of variants helped me map ambiguous prompts to the closest pattern quickly. the case notes also flagged common gotchas for each industry. which industry do you struggle with most and would like a focused case walkthrough for?
libraries are useful, but only if you stop copying and start adapting. i saw students paste retail formulas onto a two-sided marketplace and wonder why it failed. patterns help you get started, not finish. learn the core metric behind the industry — frequency for retail, ARPU for saas — and then ask one question: ‘what breaks this formula in the real world?’ that’ll keep you honest.
also: case libraries often miss edge conditions. read them, then imagine one constraint (regulation, seasonality, channel) and redo the case. if your pattern survives, it’s probably robust enough for interviews. if not, fix it before the interview.
i keep mixing up arpu and arpa for saas — can someone explain quickly how to pick which one to use?
i get lost on hardware replacement cycles — any simple heuristics people use?
Case libraries are invaluable for pattern recognition, but the learning objective should be transfer: can you map a new prompt to a known pattern and justify deviations? I advise candidates to extract the core driver for each industry, then note two common adjustments. For retail, list seasonality and channel mix; for SaaS, account tiers and churn; for hardware, replacement and adoption curves. Practice by taking a library case and intentionally changing one parameter to see how the model responds.
i learned more from a messy community case than from three polished examples. one case had a surprising channel shift that changed the whole model — reading the author’s comments taught me to always ask ‘which channel matters most.’ after that, i started tagging cases with the single trick that made them work. it made real interviews much less scary.
from a dataset of 60 community cases, I categorized the top-three drivers per industry and their median sensitivities. retail drivers were most sensitive to frequency (±20% causes ~±15% change in revenue estimate), SaaS drivers to penetration and ARPA, hardware to replacement cycle. when prepping, focus on the driver with the highest sensitivity and prepare a quick justification and sanity check for it.
practical tip: build a short ‘pattern card’ for each industry with (1) canonical formula, (2) 2 common adjustments, (3) one go-to benchmark. carrying these cards mentally reduces decision time in interviews and improves the defensibility of your assumptions.