Estimate a representative sample size for a attribution model proportion. Enter the eligible population, expected proportion, margin of error, confidence level, and expected usable-response rate. The calculator applies finite-population correction and reports both the minimum usable sample and the raw collection target, helping teams plan realistic data acquisition.
The calculation estimates a sample for measuring an attributed proportion. It does not determine the amount of data needed to fit every attribution model or identify causal effects.
Randomness and representativeness are assumptions. More records do not repair systematic exclusion, correlated observations, or poor measurement.
For 200,000 eligible conversion paths, a 30% expected attributed share, ±5% margin of error, 95% confidence, and 85% usable-response rate, the minimum usable sample is 323 and the collection target is 380.
Use the eligible conversion paths, users, or journeys that match the analysis scope and measurement period.
Use the proportion tied to the estimate whose precision matters, such as the share of conversions attributed to a channel.
Only through the usable-response input; estimate the share of collected paths that remain valid after those losses.
Not by itself. It sizes a proportion estimate; complex models may need more observations per feature and outcome.
You cannot collect more distinct eligible population units than exist in the defined finite population.
| Variable | Meaning |
|---|---|
| N | Eligible population size |
| p | Expected population proportion |
| e | Desired margin of error |
| z | Critical value for confidence level |
| Usable rate | Expected valid share of collected records |