{% extends "flags/base.html" %} {% load flag_extras %} {% block title %}Results - {{ flag.key }}{% endblock %} {% block content %}

Experiment results

{{ flag.key }} metric {{ results.metric_key }} in {{ env.name }} Back to flag
{% if total == 0 %}
No exposures recorded yet. Evaluate the flag from an SDK to start collecting data.
{% else %}
{% for v in results.variations %} {% if v.is_control %}{% else %} {% endif %} {% endfor %}
VariationExposuresConversionsRate Uplift vs control95% CI (abs. diff)p-valueMore samples needed
{{ v.value|pretty }} {% if v.is_control %}control{% endif %} {{ v.exposures }}{{ v.conversions }}{{ v.rate|pct }}baseline{{ v.uplift|signed_pct }} {% if v.ci_low is None %}-{% else %}{{ v.ci_low|signed_pct:2 }} to {{ v.ci_high|signed_pct:2 }}{% endif %} {{ v.p_value|pvalue }}{% if v.significant %} ✓{% endif %} {% if v.needs_more is None %}-{% elif v.needs_more == 0 %}enough{% else %}{{ v.needs_more }} per arm{% endif %}

Exposures are unique contexts that evaluated the flag; a conversion is an exposed context that sent the metric at least once. p-value: two-sided pooled two-proportion z-test. Sample estimate: 80% power at alpha 0.05 to detect the observed difference (or a 10% relative lift when the observed one is smaller).

{% endif %} {% endblock %}