{% extends "base.html" %} {% block title %}Glossary{% endblock %} {% block page_title %}Terminology Glossary{% endblock %} {% block page_subtitle %}Standardized definitions for all dashboard metrics and terms{% endblock %} {% block content %}
{% for cat_key, cat in categories.items() %}

{{ cat.icon }} {{ cat.label }}

{% for term_key in cat.keys %} {% set card = glossary.get(term_key) %} {% if card %}
{{ term_labels.get(term_key, card.term) }} {% if card.aliases %} {{ card.aliases[:2] | join(', ') }} {% endif %}

{{ card.what }}

{% if card.how %} {{ card.how }} {% endif %}
{% if card.why %}
Why {{ card.why }}
{% endif %} {% if card.not_this %}
β‰  Not {{ card.not_this }}
{% endif %}
{% if unit_rules.get(term_key) %}
Format: {{ unit_rules[term_key].example }} {{ unit_rules[term_key].pattern }}
{% endif %}
{% endif %} {% endfor %}
{% endfor %}

πŸ“Š Aggregation Types

Daily

Aggregated by request date (UTC). Midnight-to-midnight boundaries.

Hourly

Grouped by hour boundary (UTC). Partial hours are included.

Per-request

Individual request data. No aggregation β€” each row is one API call.

πŸ“ˆ Percentile Latency Explained

Percentile latency tells you how fast a given percentage of requests completed.

p50 (median) Half of requests completed faster than this. The "typical" experience.
p95 95% of requests completed faster than this. Only 1-in-20 were slower.
p99 99% of requests completed faster than this. Only 1-in-100 were slower β€” worst-case experience.
πŸ’‘ High p99 with low p50 means most requests are fast, but a few outliers are very slow. Investigate spikes in the Engineering dashboard.

πŸ“ Unit Format Rules

{% for unit_key, rule in all_unit_rules.items() %}
{{ unit_key | replace('_', ' ') | title }}

Example: {{ rule.example }}

{{ rule.pattern }}
{% endfor %}
{% endblock %}