Resources

    Free browser tool

    Plan a fixed-horizon binary test.

    Estimate per-arm units for two independent binary groups with equal allocation, a fixed horizon and the stated normal approximation. Use the manual worksheet for another design.

    No email gate. Inputs stay in this page, with no saving or submission.

    Your inputs

    Use assumptions or nonpersonal evidence labels. Keep contact records, credentials and confidential customer details out of the tool.

    Your result

    Enter your own assumptions, or try the clearly labelled fictional example.

    How to use this result

    For baseline p1 and relative lift l, comparison p2 = p1 × (1 + l). The fixed-horizon normal planning equation uses two-sided alpha, target power and equal allocation. No continuity correction or opposite-tail power is included. It requires at least ten expected events and non-events in each group at the estimated count.

    This is not a significance checker, winner declaration or sample-size guarantee. Outcome delay, clustered observations, multiple variants and sequential monitoring need a different design. Traffic days exclude outcome waiting time and unstable traffic.

    Method context: R two-proportion power documentation. Alpha is not the probability a hypothesis is true.

    Keep planning

    Explore paid ads management