fermi-estimation

To estimate token cost and compute effort without precise metrics, decompose the target quantity into independent estimable factors like rate × time or count × average, then compute low, central, and high bounded ranges to surface key driver assumptions.

Fermi estimation is a quantitative reasoning method that produces order-of-magnitude answers for scenario planning when precise data is missing, allowing you to sanity-check plans and make fast feasibility calls for scaling and resource questions.

Perform a cost feasibility sanity check by defining your target quantity and units, decomposing it into estimable factors, computing bounded ranges, and checking reference points to identify which key driver factor would need validation.

Yes, you can use order-of-magnitude estimation for token budgeting and resource scaling by breaking down unknown quantities into independent factors like number of turns multiplied by average tokens per turn to generate a practical low to high range.

The best way to identify the key driver factor in a quantitative estimate is to decompose the target into independent factors, compute bounded low, central, and high ranges, and perform sanity checks against reference points to determine which assumption most influences the result.

Order-of-magnitude estimates are limited to sanity checking and feasibility calls rather than precise measurement, meaning the bounded ranges identify which key driver factor requires validation but do not replace actual runtime token usage data.