fix: Fix analytics page

pages/analytics/index.md

@@ -20,43 +20,80 @@ published; no visitor identifiers or raw API responses are included.
   repeated views of the same page. The Web platform filter excludes app screens.
 - **Active users** (`activeUsers`): distinct users GA4 classifies as active,
   based on engagement and applicable first-visit or engagement signals. This is
-  queried once over the entire 30-day period, not added from daily totals. It is
-  not a count of all visitors or a census of individual people.
+  queried over each entire 30-day period, not added from daily totals. It is not
+  a count of all visitors or a census of individual people.
 - **Sessions** (`sessions`): sessions that began during the reporting period, as
   measured by GA4. A user can have more than one session.
 
 All dates are inclusive calendar dates in the **GA4 property's reporting
 timezone**, shown above. The summary covers 30 completed days ending yesterday
-at the time of the last successful refresh. Monthly history covers up to 12
-completed calendar months; the current month is excluded. Only months returned
-by GA4 are shown. Missing months are not invented as zero: they may predate
-collection or have no available rows. Returned zero counts mean no measured
-traffic for that query, not necessarily no actual visits.
+at the time of the last successful refresh. Comparisons use the immediately
+preceding 30 completed days in the same timezone, not the previous calendar
+month. Percentage changes have no meaningful baseline when the previous count is
+zero. The daily chart covers the summary period; monthly history covers up to 12
+completed calendar months, excluding the current month. These are different
+reporting windows, not directly comparable totals.
+
+Missing dates and months are marked **Not reported**, with gaps in the chart;
+they are not invented as zeros and may predate collection. Explicit zero counts
+mean no measured traffic for that query, not necessarily no actual visits.
 
 The hostname scope above is an exact allowlist, combined with a Web-only filter.
 Other hostnames, including previews and localhost, and app traffic are excluded.
 This is website traffic, not OSL's combined audience across social media,
 community platforms, or other websites.
 
+### Audience breakdowns
+
+- **Countries** show pageviews by GA4's `country` dimension. This is approximate
+  activity location, not nationality or residence. VPNs, proxies and location
+  determination can affect it.
+- **Devices** show pageviews by `deviceCategory` (such as desktop or mobile),
+  not a count of distinct devices or people.
+- **Channels** show sessions by `sessionDefaultChannelGroup`, GA4's rule-based
+  classification of how sessions started. This is session acquisition, not a
+  user's first-ever acquisition channel or a campaign attribution audit.
+
+Each panel uses the same 30-day window and hostname scope as the headline. Up to
+ten named categories are ranked by their additive metric. A category needs at
+least ten `activeUsers` over the full period to be named; smaller, remaining and
+unclassified categories are combined as **Other / unknown**. Those per-category
+user counts are used only for grouping and are never published or summed into
+headline users. Grouping reduces detail; it is not a formal anonymity guarantee.
+No city, visitor, page-path or cross-category breakdowns are published.
+
+Shares use each panel's returned total, including Other / unknown, not the
+headline user count. Rounded shares may not add to exactly 100%. Dimensioned and
+headline reports may differ during processing or because of reporting semantics;
+we do not adjust them to force agreement. A retained older snapshot may lack
+comparisons or audience panels; that absence is not zero traffic.
+
 ### Limitations and freshness
 
 Measured traffic can exclude visits affected by **consent choices, blockers**,
 disabled JavaScript, or collection errors. Identity settings, estimation, and
 cross-device behavior can affect user counts. Recent figures may change during
-GA4 processing. Every refresh re-queries both the rolling window and the full
-historical window, so earlier figures can be revised too.
+GA4 processing. Every refresh re-queries the current and comparison windows,
+daily and monthly histories, and all breakdowns, so earlier figures can be
+revised too. A zero report warrants checking collection and scope before
+concluding that nobody visited the site.
 
 Refreshes are scheduled daily at **06:23 UTC**, but GitHub Actions scheduling
 may be delayed. If a refresh fails, we retain the previous successful report and
 its original timestamp. A report more than three days old is marked stale. If
-GA4 signals thresholding, sampling, truncation, or unavailable data, we do not
-replace the previous report with incomplete results or fabricated zeros.
+GA4 explicitly signals a privacy threshold or metric access restriction for an
+audience breakdown, that panel is marked **Not published**, not zero. A
+restriction on core statistics, sampling, truncation, unavailable data, or a
+failed request instead retains the entire previous report. We never publish
+partially fetched rankings or label a failed attempt as a successful refresh.
 
 Metric definitions follow Google's
 [GA4 Data API schema](https://developers.google.com/analytics/devguides/reporting/data/v1/api-schema).
 See also Google's
 [data freshness guidance](https://support.google.com/analytics/answer/12233314).
-The endpoint follows our [version 1 JSON schema](schema.json).
+The endpoint follows our [versioned JSON schema](schema.json). New exports use
+version 2; older version 1 snapshots remain readable without changing their
+original refresh timestamp.
 
 ## Support open science