KPI Relationships

Feature

A KPI Relationship is a directional link from one KPI to another KPI, configured directly on the "source" KPI. It tells the platform how the two KPIs relate to each other — for example, that Mouse Click errors contribute to User Click failures, that Salesforce API errors show up as contextual markers on Order Sync issues, or that a slower Checkout Page Load Time is inversely correlated with Shopify Conversion Rate.

KPI Relationships are the data behind the Analysis dashboard's root-cause-analysis features: when you open a KPI's Analysis view, the platform looks up its configured relationships, scores each related KPI for relevance (based on deviation, direction, strength, and type), and surfaces the most relevant ones as "Related KPIs" alongside the primary KPI's trend. This is where the real power shows up during an investigation: instead of manually jumping between dashboards to check whether other KPIs moved at the same time, you get an instant, pre-scored view of which KPIs are leading, lagging, or moving in lockstep with the one you're investigating — collapsing what used to be a slow, manual correlation hunt into a single glance.

Configuration

The easiest way to add a relationship is the KPI Relationship wizard, available from a KPI's configuration page:

1. Open the KPI you want to add a relationship from (the source KPI).

2. Launch the Add KPI Relationship wizard.
   

   

3. Select the related KPI and fill in the fields described below.
   

   

4. Save — the new entry appears under the source KPI's KPI Relationships list.

Each relationship entry has the following fields:

• Name — a unique name for this relationship entry.

• KPI — the name of the related KPI.

• Type — how this KPI relates to the current KPI:

  • Contributor — this KPI's problems feed into the current KPI's problems.

  • Correlated — this KPI and the current KPI move together, without one causing the other.

  • Contextual — useful to see alongside, with no direct relationship.

• Direction — whether the two KPIs move in the same direction (Same) or in opposite directions (Inverse).

• Strength — how strongly the relationship is expected to manifest: Strong, Moderate, or Weak.

• Constraint (optional) — a filter expression that must hold for the relationship to apply, e.g. userError == true.

• Field (optional) — the field used to correlate or join this KPI with the current KPI, e.g. sessionId.

• Measure (optional) — the measure expression evaluated for this relationship, e.g. avg(duration). If left blank, it defaults to the average of the related KPI's own property (or a plain count, if it doesn't have one).

• Show As — how this relationship is displayed on the Analysis dashboard: Trend (default) or Marker — see UI Results below.

• Child — if enabled, marks that this KPI executes within the current KPI's context (for example, Mouse Click runs inside User Click). This drives the nested execution hierarchy shown in the RCA Trace Timeline.

Example

Suppose "Mouse Click" errors are a leading contributor to "User Click" failures within the same user session. On the User Click KPI, you would add a relationship:

• Name: MouseClickContributor

• KPI: Mouse Click

• Type: Contributor

• Direction: Same

• Strength: Strong

• Show As: Trend

• Child: enabled

Analysis dashboard - Relationships Portlet

Configured relationships surface on the KPI's Analysis dashboard, in the trend chart and the Related KPIs section beneath it.

Show As: Trend vs. Marker

• Trend (default) — the related KPI gets its own card in the Related KPIs grid, with a mini trend sparkline, its current value vs. baseline, and a co-movement indicator showing how often it moved in step with the primary KPI. Clicking the sparkline promotes that KPI's full trend onto the main chart for direct comparison.
  

  

• Marker — the related KPI does not get a card. Instead, every individual occurrence is plotted directly on the main chart's timeline as a small shape (diamond, circle, square, or triangle, color-coded per KPI). Hovering a marker shows details for that occurrence; clicking it opens the Analysis view for that specific instance.
  

  

Switching a related KPI to pivot (table) view

Any Trend-type related KPI card can be broken down by a dimension instead of showing a single sparkline:

1. On the related KPI's card, click the Pivot By icon.

2. Choose a dimension from the dropdown (e.g. Country, Browser, Server).
   

   

3. The card switches to a table, broken down by the values of that dimension, showing each value's health (color distribution), a mini trend, the measure, and a count.

   

   
   

4. Click a row to pin that segment onto the main chart for comparison.

Filtering markers

When a KPI's Analysis view has one or more Marker-type relationships, a Filter Events icon appears in the portlet's toolbar:

1. Click the Filter Events icon.
   

   

2. In the popover, use the checkboxes to show or hide each marker KPI entirely, or expand a KPI to filter by the values of its individual fields.

3. Use the search box to quickly find a KPI or a specific field value.

4. Use Select all / Clear all to reset the filter.

Filter selections apply immediately to the markers on the chart and are specific to that portlet instance (they reset if you change the KPI, scope, or time range being analyzed).