Matomo Connector Improvements#1624

1. Page URLs Not Available as a Dimension
   Priority: High
   The Matomo connector currently exposes page data exclusively from the Actions.getPageTitles API endpoint (table: action_page_titles). There is no access to Actions.getPageUrls, which means page URLs are not available as a dimension.
   Impact: Page URLs are essential for joining Matomo web analytics data to other platforms (HubSpot form submissions, Google Ads landing pages, Search Console performance data). Without URLs, cross-platform attribution requires manual lookup tables or supplementary direct Matomo API calls, undermining Windsor’s value as a unified data layer.
   Evidence from field schema:

Every page_ field maps to upstream_api_response_name: action_page_titles__
No action_page_urls table exists in the connector schema
The page_segment field encodes the page title, not the URL

Requested addition: A parallel set of page fields sourced from Actions.getPageUrls, or a page_url dimension added to the existing page fields. The Matomo Reporting API supports this as a standard endpoint with identical metrics (nb_visits, bounce_rate, avg_time_on_page, exit_rate, etc.).

1. Visit Metrics Return Null When Combined with Dimensional Fields
   Priority: High
   When querying dimensional fields (referrer_type, referrer_name, visitor_type, campaign_name, device_type) alongside visit metrics (nb_visits, nb_actions, bounce_count, sum_visit_length), the metrics collapse to a single null-dimension row. The dimension values appear but with zero values for all metrics.
   Example query: fields = [“referrer_type”, “nb_visits”, “nb_actions”]
   Actual result:
   referrer_type: null → nb_visits: 16893, nb_actions: 24603
   referrer_type: direct → nb_visits: 0, nb_actions: 0
   referrer_type: search → nb_visits: 0, nb_actions: 0
   referrer_type: campaign → nb_visits: 0, nb_actions: 0
   Expected result: Visit metrics distributed across dimension values.
   Workaround discovered: goal_conversions is the only metric that correctly cross-tabulates with dimensions. This suggests the connector handles the goal_conversions aggregation differently from standard visit metrics.
   Affected combinations tested:

referrer_type + nb_visits → null collapse
visitor_type + nb_visits → null collapse
device_type + nb_visits → null collapse
referrer_type + goal_conversions → works correctly
visitor_type + goal_conversions → works correctly
device_type + goal_conversions → works correctly

Impact: Users cannot build channel-level, device-level, or visitor-type traffic reports through Windsor without using goal_conversions as a proxy metric. This is the single largest limitation of the Matomo connector for marketing analytics use cases.

1. Multi-Month Date Ranges Collapse Dimensional Data
   Priority: Medium
   When querying dimensional breakdowns across multi-month date ranges using year_month or month as a time dimension alongside referrer_type or referrer_name, only the most recent month resolves correctly. All prior months collapse into a null row.
   Example: Querying [“month”, “referrer_type”, “goal_conversions”] for Sept 2025–Feb 2026 returns proper data only for February. September through January collapse to a null month with aggregated totals.
   Workaround: Single-month date range queries appended in the destination (Power BI). This works but requires parameterized queries and adds complexity.
   Requested fix: Multi-month dimensional queries should return data for each month, consistent with how the date field works (daily granularity queries work correctly with dimensions).

2. Filters Do Not Apply to Dimensional Queries
   Priority: Medium
   Windsor’s filter parameter does not reduce dimensional query results. Filtering by referrer_type = “campaign” still returns all dimension rows with zero metrics, identical to unfiltered behavior.
   Example: fields = [“date”, “nb_visits”], filters = [[“referrer_type”, “eq”, “campaign”]] returns daily nb_visits values of 0 for every date.
   Expected: Only visits with referrer_type = campaign should be counted.
   Note: This may be a symptom of issue #2 (the metric-dimension collapse), not a separate bug.

3. Entry/Exit Page Metrics Lack URL Context
   Priority: Low
   Entry page metrics (page_entry_nb_visits, page_entry_bounce_count, page_entry_nb_uniq_visitors) and exit page metrics (page_exit_nb_visits, page_exit_rate) are available but only associated with page titles (via page_label), not page URLs. Without URLs, entry/exit page analysis cannot be joined to campaign landing page data or Google Ads final URLs.
   This is a downstream consequence of issue #1 and would be resolved by adding Actions.getPageUrls support.