Changelog

Trino connections can use HTTPS with a custom certificate, and dashboard titles can be edited. Cohorts now work on Snowflake.

Added​

• Trino over HTTPS. Connect to Trino over HTTPS, including with a custom certificate.
• Dashboard titles. Rename dashboards and manage their titles.

Fixed​

• Cohorts work on Snowflake.
• The create cohort dialog resets when it is closed.
• Fixed several issues with Trino connections.

Cohorts become usable in analysis: filter any insight by cohort membership and see a cohort's size before you create it. The explore page gets a refreshed layout, property aggregations gain median and percentiles, and warehouse errors are translated into readable messages.

Added​

• Cohort filters. Filter insights by whether users belong to a cohort, just like any other property.
• Cohort size preview. The create cohort dialog shows how many users the cohort will contain.
• Median and percentile aggregations. Aggregate an event property by its median or a chosen percentile.
• Readable warehouse errors. Common warehouse errors are shown as clear messages that explain what went wrong.

Changed​

• The explore page has a redesigned, cleaner layout.
• The property aggregation dropdown uses the property catalog, with display names.
• Browser tabs show the title of the current page.
• Chart tooltips work on multi-event charts without breakdowns.
• Read-only users no longer see create cohort, save and update actions.
• The run button stays disabled until the metric is complete.
• Schema and table lists can be refreshed when the warehouse changes.

Fixed​

• Drill-down works on charts with breakdowns and on segmentation charts with an overall time group.
• Creating a cohort from a chart that already uses a cohort filter works.
• Removing a cohort filter removes the right one.
• Hovering over breakdown values highlights the right series.

The results table and generated SQL now sit in an accordion under the chart, and event discovery reports warnings instead of failing silently.

Changed​

• The results table and generated SQL are grouped in an accordion below the chart.
• Event discovery shows warnings for tables or fields it could not read.
• Scrollbars have a custom, more compact style.

Mitzu shows a loading screen while it starts, and event discovery and filtering handle events with missing names.

Changed​

• A splash screen is shown while the app loads.

Fixed​

• Event discovery skips rows where the event name is missing.
• Filtering on event names works on warehouses that need an explicit type cast.

Saved metrics load faster and the property picker is easier to scan, with properties sorted by display name and shown with their full path.

Changed​

• Saved metrics load faster.
• Properties in the property dropdown are sorted by display name.
• Property pickers show the full property path instead of only the last field name.

Fixed​

• The manage properties link works after an event's display name is changed.
• Charts no longer show a broken tooltip when the data and the chart type don't match.
• Properties with unexpected empty values load correctly in value lists.
• Legend badges on dashboards resize with the card.

A small fix release for charts that fill missing periods with zeros.

Fixed​

• Missing periods on charts are filled with zeros correctly.

A hotfix for Databricks charts.

Fixed​

• Charts on Databricks work again after missing periods started being filled with zeros.

The event property catalog arrives, so you can browse and describe every property in your workspace. Charts fill missing periods with zeros, and retention can be measured on a property other than the user ID.

Added​

• Event property catalog. Browse, search and describe event properties, on desktop and on mobile.
• Custom holding constants for retention. Measure retention on a property other than the user ID, such as an account ID.

Changed​

• Periods with no data are shown as zero instead of being skipped.
• Drill-down charts are shown as bar charts.

Fixed​

• Drill-down from chart tooltips opens the right users and events.
• Pagination in the catalog works and lines up correctly.

Metric trends can use a rolling window, and retention can count users who come back at any point after a period. The explore page also gets a round of usability improvements.

Added​

• Rolling windows. Smooth metric trends with a rolling time window.
• Unbounded retention. Count users who return in a period or any time after it.

Changed​

• The explore page is easier to use, with clearer controls.
• The navigation bar shows when you are in a sample project.

Fixed​

• Retention charts with no data no longer fail.

Explore now uses the property catalog, so properties show their display names and default values from discovery. Several fixes land for nested properties and retention on Trino.

Changed​

• Property pickers on the explore page use the property catalog, with display names and discovered values.
• Only the button you clicked shows a loading spinner.

Fixed​

• Nested properties whose names collide are kept apart.
• Overall retention works on Trino.
• Fixed small layout issues in the property dropdown and elsewhere.

A round of explore usability updates, with fixes for retention charts and saved metrics.

Changed​

• The explore page gets several usability improvements.

Fixed​

• Switching the time group on retention charts no longer blocks the page.
• Refreshing a saved metric works again.

Segmentation and funnels can be measured on a property other than the user ID, and warehouse connections accept extra URL parameters. Dashboards load faster, and BigQuery handles case-sensitive names.

Added​

• Custom holding constants. Count users by a property other than the user ID, such as an account or company ID.
• Connection URL parameters. Add extra URL parameters to warehouse connections, including during onboarding.

Changed​

• Dashboards load much faster.
• The sample SaaS project has richer, more realistic data.
• Event discovery is disabled until at least one event table is configured.

Fixed​

• BigQuery works with case-sensitive table and column names.
• Dashboards no longer flicker when they refresh.
• Fixed several dashboard bugs and a drill-down issue.
• New organizations get a correct copy of the event catalog.
• Fixed Athena and event table settings.

Funnels now run on ClickHouse, and date filters no longer wrap partition columns in functions, so warehouses can prune partitions.

Added​

• ClickHouse funnels. Build funnel insights on ClickHouse.

Changed​

• Date partition columns are filtered directly, so warehouses can skip partitions and queries run faster.
• BigQuery credentials are stored the same secure way as other passwords.
• The unused connection name field is removed.
• Zooming is disabled on heatmaps.

Fixed​

• The formula placeholder is cleared when there are no segments.

Segmentation insights now run on ClickHouse. The navigation bar is reworked for desktop and mobile, and saved metrics load faster.

Added​

• ClickHouse segmentation. Build segmentation insights on ClickHouse.

Changed​

• The navigation bar is redesigned for desktop and mobile, with the organization switcher back in place.
• Saved metrics load faster.
• The sample SaaS project has more demo data.

Fixed​

• Fixed several small usability issues on the explore page.

You can restrict an insight to users' first event in the selected period, filter individual segments, and give insights a title and description. ClickHouse warehouses can be connected and discovered.

Added​

• First-event filter. Count only each user's first event in the selected time range.
• Segment filters. Add filters to individual segments.
• Insight titles and descriptions. Name and describe insights on the explore page.
• ClickHouse discovery. Connect a ClickHouse warehouse and discover its events.

Changed​

• The explore page gets several usability improvements, including on mobile.
• Onboarding uses radio buttons to pick a warehouse.
• Save buttons show a loading state while saving.

Fixed​

• Custom time filters are saved completely.
• Dashboard cards have the right width.

A small fix release for formulas, onboarding and dashboards.

Fixed​

• Division in segmentation formulas works correctly.
• Onboarding errors are shown instead of staying hidden.
• Fixed layout issues on dashboards.

Segmentation insights can now combine segments with a formula, such as a ratio of two events.

Added​

• Segmentation formulas. Enter a formula that combines segments, such as A / B, to chart a calculated metric.

A hotfix release for dashboards.

Fixed​

• Fixed several dashboard issues.

New users are guided through connecting a warehouse and discovering events. The dashboards page is redesigned and refreshes cards in parallel, and segments can each have their own aggregation.

Added​

• Guided onboarding. A step-by-step flow walks you through connecting your warehouse, picking event tables and discovering events.
• Redesigned dashboards. The dashboards page has a new design, and cards refresh in parallel.
• Per-segment aggregation. Choose a different aggregation for each segment in a segmentation insight.

Changed​

• Testing a connection runs in the background, and connection details are kept when the warehouse type does not change.
• Schema and table lists are cached, so connection settings load faster.
• The results table sits under the chart.
• Metrics are validated before a query runs.
• Retention insights accept up to two segments.
• The sample data and logos are updated.

Fixed​

• Chart tooltips work with per-segment aggregation.
• Fixed previous and next navigation in onboarding and several small UI issues.

A small fix release for retention heatmaps.

Fixed​

• Heatmap tooltips show the right values.
• Heatmaps handle missing values.