Performance 2025 Queries (Part 2)

sql/2025/performance/early_hints_usage_rank.sql

@@ -0,0 +1,31 @@
+#standardSQL
+# Early Hints (HTTP 103) usage by site rank
+# Measures the percentage of sites using early hints at different popularity ranks
+
+SELECT
+  IF(ranking < 100000000, CAST(ranking AS STRING), 'all') AS ranking,
+  r.client,
+  COUNT(DISTINCT r.page) AS total_sites,
+  COUNTIF(JSON_QUERY_ARRAY(r.payload._early_hint_headers) IS NOT NULL) AS sites_with_early_hints,
+  COUNTIF(JSON_QUERY_ARRAY(r.payload._early_hint_headers) IS NOT NULL) / COUNT(DISTINCT r.page) AS pct_early_hints
+FROM
+  `httparchive.crawl.requests` r
+JOIN
+  `httparchive.crawl.pages` p
+ON
+  r.page = p.page AND
+  r.client = p.client AND
+  r.date = p.date,
+  UNNEST([1000, 10000, 100000, 1000000, 10000000, 100000000]) AS ranking
+WHERE
+  r.date = '2025-07-01' AND
+  r.is_main_document AND
+  r.is_root_page AND
+  p.is_root_page AND
+  p.rank <= ranking
+GROUP BY
+  ranking,
+  r.client
+ORDER BY
+  ranking,
+  r.client

sql/2025/performance/early_hints_usage_trends.sql

@@ -0,0 +1,35 @@
+#standardSQL
+# Early Hints (HTTP 103) adoption trend over time
+# Shows usage by client (mobile, desktop) across years
+
+SELECT
+  r.client AS client,
+  r.date AS date,
+  COUNTIF(
+    JSON_QUERY_ARRAY(r.payload._early_hint_headers) IS NOT NULL AND
+    ARRAY_LENGTH(JSON_QUERY_ARRAY(r.payload._early_hint_headers)) > 0
+  ) AS sites,
+  COUNT(0) AS total,
+  COUNTIF(
+    JSON_QUERY_ARRAY(r.payload._early_hint_headers) IS NOT NULL AND
+    ARRAY_LENGTH(JSON_QUERY_ARRAY(r.payload._early_hint_headers)) > 0
+  ) / COUNT(0) AS pct
+FROM
+  `httparchive.crawl.requests` r
+JOIN
+  `httparchive.crawl.pages` p
+ON
+  r.page = p.page AND
+  r.client = p.client AND
+  r.date = p.date
+WHERE
+  r.date IN ('2023-07-01', '2024-07-01', '2025-07-01') AND
+  r.is_main_document AND
+  r.is_root_page AND
+  p.is_root_page
+GROUP BY
+  r.client,
+  r.date
+ORDER BY
+  r.client,
+  r.date

sql/2025/performance/speculation_rules_rank.sql

@@ -0,0 +1,29 @@
+SELECT
+  IF(ranking < 100000000, CAST(ranking AS STRING), 'all') AS ranking,
+  client,
+  COUNT(DISTINCT page) AS total_sites,
+  COUNTIF(
+    custom_metrics.performance.speculation_rules IS NOT NULL AND (
+      ARRAY_LENGTH(JSON_QUERY_ARRAY(custom_metrics.performance.speculation_rules.htmlRules)) > 0 OR
+      ARRAY_LENGTH(JSON_QUERY_ARRAY(custom_metrics.performance.speculation_rules.httpHeaderRules)) > 0
+    )
+  ) AS sites_with_speculation_rules,
+  COUNTIF(
+    custom_metrics.performance.speculation_rules IS NOT NULL AND (
+      ARRAY_LENGTH(JSON_QUERY_ARRAY(custom_metrics.performance.speculation_rules.htmlRules)) > 0 OR
+      ARRAY_LENGTH(JSON_QUERY_ARRAY(custom_metrics.performance.speculation_rules.httpHeaderRules)) > 0
+    )
+  ) / COUNT(DISTINCT page) AS pct_speculation_rules
+FROM
+  `httparchive.crawl.pages`,
+  UNNEST([1000, 10000, 100000, 1000000, 10000000, 100000000]) AS ranking
+WHERE
+  date = '2025-07-01' AND
+  is_root_page AND
+  rank <= ranking
+GROUP BY
+  ranking,
+  client
+ORDER BY
+  ranking,
+  client