RUM Explorer Analytics

The RUM Explorer gives you two view modes: List and Analytics. The Analytics view aggregates the events of the current event type by the aggregation and group you pick, which is what answers questions like “which page loads slowest” or “which browsers do the errors come from”. Switch back to the list when you need individual events.

​

List and analytics

The top of the Explorer has two tabs: List shows events one by one, Analytics aggregates them into charts. Both share the same event type, query conditions and time range, so switching tabs never drops your current filters. The mode is recorded in the mode URL parameter: mode=analytics opens analytics, and omitting mode opens the list, so existing Explorer links keep opening the list. Copy the URL of an analytics view to share the same chart with others.

​

Choosing the aggregation and fields

The Measure row at the top of the analytics view decides what is computed: pick an Aggregation, then a Measured field where one is needed, then a Group-by field.

​

Aggregations

Every aggregation except Count needs a Measured field. With Average, Sum, Min, Max or any percentile, the field picker lists numeric fields only; timestamp fields (unit unix_millisecond) are excluded, since averaging them is meaningless. With Unique count, the field picker lists the scalar facets that can be grouped by. After you switch the event type, a field that does not exist on the new event type is reported as missing and you are asked to pick another one, rather than the question silently becoming a count. The legend, the table columns and the CSV header all use the same naming: the measure of Count and of a Distribution reads “Events”; every other aggregation reads “Aggregation (field name)”, for example “Average (view_loading_time)”.

​

Group-by field

The Group-by field decides what the measure is split by. Once one is chosen, the chart draws each group value separately and a top selector appears next to it, offering 5, 10, 20 and 50 groups, 10 by default. The group-by field can be cleared (the placeholder reads “No grouping”), and the result then collapses into a single number. Only scalar facets can be grouped by; an array field holds several values per event and therefore has no meaningful grouping. A Distribution does not use a group-by field, and that selector is not shown.

​

Chart types

The analytics view offers five chart types: A Timeseries also offers a Draw style selector that switches between Line (line) and Bars (bar), bars by default. When a group-by field is set, the draw style is bars and the aggregation is Count or Sum, the bars are stacked by group and the bar height is the total; the other aggregations (averages, percentiles, unique counts) cannot stack, because their per-group values do not add up. Without a group-by field, Top list and Table print the measure itself instead of drawing a single bar; a Timeseries still plots one line over time.

​

Drilling down

Clicking an element on a chart opens a drill-down menu that takes you from the aggregate back to individual events: a point in time on a timeseries, a row in the top list or table, a slice of a pie, or a range of a distribution. A drill-down writes the same filters the facet list on the left writes, so they show up in the search box and can be undone the same way. Drilling down from a timeseries point narrows the list to that time bucket, and drilling down from a distribution range filters the list by that numeric range. After View in list, a Back to analytics bar appears above the list with the note “The list is filtered by the drill-down; going back restores the previous query”. Clicking it returns to the analytics view and restores the query you drilled from. Switching tabs by hand ends that drill-down, and the bar disappears with it.

​

Zooming a timeseries

Drag across a Timeseries to zoom into that window. The zoom applies to this chart only; the Explorer’s own time range stays as it is. While zoomed, the Zoomed badge in the top-right corner of the chart shows the current window, and the close button on it (Reset zoom) restores the full time range. Changing the event type, the query conditions or the Explorer’s time range resets the zoom.

​

Downloading an image and exporting CSV

Two export buttons sit in the top-right corner of the chart: Both buttons are unavailable while the result is empty.

​

Distribution

A Distribution buckets one numeric field by value range and shows the number of events in each range. It ignores the aggregation and the group-by.

• Buckets span from the field’s minimum up to P99: RUM timings have a long tail (one 46s page load among thousands of 1s ones), and splitting up to the maximum would crush the whole body of values into the first bar.
• The bucket width is rounded to 1, 2 or 5 times a power of ten, so bucket edges read as round numbers; the observed range is split into about 40 buckets.
• Everything above P99 goes into one open-ended last bucket, shown on the chart and in the CSV as ”≥ edge”.
• Ranges are labelled in the field’s own unit, for example 1s–1.5s for a duration field.

​

When a Pie applies

A Pie requires both of the following; otherwise the chart area shows “Pie charts need a group-by and only apply to count and sum”:

• a Group-by field is set;
• the aggregation is Count or Sum — only these two add up to a whole, which is what a pie claims. Unique counts, averages and percentiles do not.

​

Sampled estimates

When the data volume is large enough that the result is estimated from a sample, the Sampled badge appears in the top-right corner of the chart, with the tooltip “Large data volume: results are estimated from a sample”.

​

URL parameters

The analytics view keeps its state in the URL, so a copied link reproduces the same analysis. Values that are not allowed fall back to the default.

​

Next steps