Highlights in Explore

Highlights help you see how spans are distributed across fields, so you can quickly spot commonalities, outliers, and root causes.

To access Highlights, select:

• Explore, then Tracing. Open Highlights.
• Explore, then Visual Explorer. Within your Widget more actions menu, open Highlights (coming soon).

How it works

Highlights analyzes the data currently in view—respecting your selected Explore time range and filters—and renders per-field distributions.

Spans are grouped by fields such as application, user_id, browser, geolocation, or http_route. You can view all fields or filter by field frequency (for example, show only fields above or below a threshold).

Fields are presented side by side for quick insights. Each field is presented as a chart with contextual data:

• Distribution of values for the field
• A pie chart that shows how many spans include the field, as a percentage of all spans in the selected period
• Change in values, as a percentage (displayed as Diff )
• Exact count of spans for each field value, shown when you hover over the chart

Sharpen your analysis

Pinpoint the types of spans you’d like to view

Select one of the following tabs: All spans, Errors, or Duration.

Search

Search for a specific field using free text in the Search bar.

Filter

Filter fields. Highlights applies all Explore-level filters and any Lucene or DataPrime query filters, even if they do not appear in the Highlights filter controls. Add additional filters that will apply solely to your Highlights investigation.

Sort your results

• When investigating errors, use Sort to surface distribution patterns.
• When investigating duration, display results by any of the following: Avg, Min, Max, P99, P95, P50.

Focus your query

From the more actions menu on each field value, you can:

• Filter in / Filter out: Add or remove a filter to the query that appears in Highlights only.
• Include in / Exclude from query: Quickly refine your query in Explore to include or exclude a field value without leaving Highlights.

For instance, including the eu-north-1 value for the cloud.region field, will add the following to your DataPrime query: filter $d.process.tags['cloud.region'] == 'eu-north-1'.

• Copy to clipboard: Copy the field and value to your clipboard for reuse elsewhere.

These actions let you pivot immediately from pattern recognition to focused analysis—for example, filtering directly into a problematic endpoint or service.

Investigating errors

Highlights excels at surfacing which fields dominate during error spikes, helping you find the fastest path to root cause.

Typical workflow

1. In Explore Tracing, open Highlights.
2. Select the Errors tab.
3. From within All fields, select View all or Field frequency to limit fields above or below a specific threshold.
4. The results are sorted by Uneven spread by default. If the pattern isn’t obvious, sort by Even spread to see fields whose values are balanced.
5. Look for dominant field values—for example, applicationName, subsystem, region.
6. Use Filter In from the more actions menu to focus on a value and inspect traces.

Once you isolate a spike window or error cluster, use Compare mode to see how this period differs from a previous period.

Sort options

Use sort options to surface distribution patterns.

Option	Description	Value
Uneven spread (default)	Fields dominated by a single value	Surfaces likely culprits or concentrations when triaging spikes or errors
Even spread	Fields whose values appear about equally	Helps rule out non-differentiators (for example, balanced regions)
Field name A–Z / Z–A	Alphabetical order	Speeds scanning or lookup when you know the field name

Investigating duration

Switch to Duration view to analyze span durations using Max (default), Min, Avg, P99, P95, or P50.

Compare mode

Compare mode helps you understand what changed between two time periods—whether a sudden spike, a regression, or a recurring pattern.

What it is

Compare Mode compares two time-based data sets:

• Selection — The time window you’re analyzing
• Comparison — Data from the rest of the timeline or another comparison period

When you change your selection (B), Period A—the comparison range—updates automatically.

Comparison patterns

Compare your selection to a previous period of up to seven days. Select from:

• Previous period (default): If your current time window is 12 minutes, the previous period is the 12 minutes immediately before the current time window.
• Previous day
• Previous 7 days

How to use Compare mode

1. Open Highlights to view Compare mode.
2. Choose your comparison type: Previous period (default), Previous day, or Previous 7 days.
3. Review side-by-side counts and percentages for each field to spot shifts.
4. Use the more actions menu on either period’s value to Filter In/Out or Include/Exclude, and continue your investigation.

Use case: Spike triage and regression analysis

1. In Explore Tracing, open Highlights.
2. Select the Errors view, with the sort set to Uneven spread. Here, you may find that endpoint=/checkout dominates errors, for examples.
3. Filter In that endpoint and view the results in Compare mode.
4. Sort by Diff to see over-represented fields, such as region=us-east-1 or version=v2.3.
5. Switch to Duration view to check whether these fields also show elevated latency (for example, P99).
6. If the same fields reappear, compare to Previous 7 Days to confirm if it’s recurring after a release.
7. Use Filter In to focus on the affected service or region and inspect traces.

Best practices

• Filter before opening Highlights. Narrow your dataset using queries.
• Use smaller time windows to reduce noise and expose clear patterns. To change the time range, use the time range picker in Explore.
• Avoid high-cardinality fields (for example, UUIDs or event IDs); focus on service, version, endpoint, or region.
• Act directly from Highlights. Use Filter In/Out or Include/Exclude to update queries instantly.
• Use Compare mode to validate whether an issue is isolated, new, or recurring.
• For quick results, the period of comparison should ideally be in minutes or hours.

Sampling and limits

• When sorting, Highlights analyzes a sample of up to 5,000 spans from your selected time range. To ensure a complete view, browse all pages.
• The system displays a maximum of 10 fields per page.
• Sort results show the top 20 values for each field.
• Highlights applies all Explore-level filters and any Lucene or DataPrime query filters, even if they do not appear in the Highlights filter controls.
• If your query results extend beyond 2 weeks (due to the original query or the query and the time of comparison), we fetch the data from All Traces TCO.