Page 256 of 506 · Back to latest writing
A customer can look profitable while consuming the roadmap through bespoke workflows, integrations, controls, and exceptions. Treating product and engineering attention as an account cost helps distinguish reusable segment signals from one-off concessions that fragment the product.
Implementation is part of the product’s true cost, spanning integration, data, training, permissions, workflow design, and trust. Its burden is defensible when each deployment creates reusable capability; when every customer starts from zero, software economics drift toward custom consulting.
Support volume is less revealing than what generates it. Recurring questions may expose a repeatable adoption journey, an underbuilt product, or a mismatched customer; tracking whether support declines with maturity turns service work into evidence about margin and fit.
AI usage carries direct and indirect costs that blended gross margin can conceal. Account-level instrumentation must connect workflows, retries, model choices, human review, evaluation, and support to customer behavior, revealing whether heavy adoption creates profitable scale or margin compression.
Unprofitable customers can represent valuable learning or permanent structural drag, and growth stage changes the acceptable tradeoff. The key is whether costly accounts produce repeatable product insight, strategic value, or future margin—or merely consume the company’s future.
In AI businesses, revenue is a claim on compute, support, implementation, product attention, and reliability—not proof of healthy scale. Separating booked revenue from what remains after serving customers reveals whether usage builds a repeatable product or conceals bespoke burden.
Product quality increasingly depends on closing the gap between design intent and shipped behavior. Across ten connected arguments, the sequence moves from role definition and prototypes to systems, AI, management, and a practical audit of where quality leaks.
A meaningful design engineer audit tests behavior, prototypes, build fluency, craft authority, systems, AI use, and role clarity against the shipped product. Its purpose is to expose where quality still disappears between intent, implementation, and release.