Choose the Platform by the Hardest Recurring Part

Pick the platform that best handles the hardest, most-recurring part of the workflow — not the part that's cheapest to prototype. The demo is a one-time cost; the hard recurring step is paid forever.

Explanation

The source (Working Heuristics — Data, Matching, Product, Org, Tooling) warns against a seductive failure: choosing a platform because it makes the first version easy. Ease-of-prototype optimizes a one-time cost. The thing you'll actually live with is whatever is hard and happens repeatedly — the recurring bottleneck, the step that's painful every single cycle. Choose the platform that makes that part tractable, and accept some prototype friction in exchange.

Why it matters / where it shows up

• You amortize the wrong cost otherwise. A platform that wins on demo-day but fumbles the recurring hard step taxes you on every future iteration — the opposite of where leverage lives.
• It's a "weight the decision by frequency × difficulty" rule. Identify the step that is both hard and frequent; let it dominate the choice. (Compare Match Process Weight to Stage and Stakes — both size a decision to where the real, recurring cost is, not where the cost is most visible up front.)
• Data gravity is often the deciding factor. The hardest recurring part is frequently "operate on the big data," which is exactly where Data Gravity says the platform should sit.

Open questions

• How do you identify the hardest-recurring part before you've run the workflow enough times to feel it? Choosing well requires foresight the prototype can't give.

Connections

• Data Gravity — the hardest recurring part is often the data-heavy step, and gravity says put the platform there.
• Match Process Weight to Stage and Stakes — see also: both weight a decision by real recurring cost rather than the most salient up-front cost.
• Build It as an Extension — a related "don't optimize for the easy first build" stance: reuse the proven mechanism rather than the quickest-to-demo new one.

Sources

• Working Heuristics — Data, Matching, Product, Org, Tooling — "choose the platform by the hardest, most-recurring part of the workflow — not the part that's cheapest to prototype."