Metrics, Dimensions & Filters

Metrics, Dimensions & Filters

This page explains the core concepts that power everything you do in Watching That — across Reports, Analysis, Inspect, and Monitoring. These concepts apply regardless of which data source you're connected to.


On This Page


Events, Metrics, and Dimensions

Watching That ingests data as a continuous stream of events. Whether the source is ad server log files, bid-level logs, pixel tracking, or API feeds, the data arrives as individual event records — an impression served, a request made, an error fired, a bid received, and so on.

Two things determine what you can measure and how you can slice it:

  • Event types → Metrics. Each event type you enable unlocks a set of metrics — what you can count and calculate. More event types = more metrics, especially derived metrics that combine data across event types.

  • Fields → Dimensions. Each field you enable becomes a dimension — what you can break metrics down by in reports, analysis, and monitoring.

Your specific configuration determines which event types and fields are active for your organisation. Not every metric or dimension will appear in your account — contact your account team if you need something that isn't currently enabled.


Three Types of Metrics

Base Metrics — direct counts

One event in, one count out.

Example: Every impression event increments Gross Impressions. Every request event increments Ad Requests. Every error event increments Errors.

Base metrics require only their parent event type to be enabled.

Derived Metrics — calculated from base metrics

Rates, ratios, averages, and aggregations — often combining data from multiple event types.

Example: Fill Rate (%) divides filled ad slots by total ad slots, requiring both availability and impression event types. Error Rate (%) requires both error and request event types.

Derived metrics unlock progressively as you enable more event types. You can also create your own derived metrics using formulas in Reports and Analysis — the standard ones are just the starting point.

Conditional Metrics — filtered by a dimension value

A subset of a base or derived metric, isolated by a condition.

Example: Gross Invalid Impressions = Gross Impressions where the invalid traffic flag is set. Impressions With Errors = impressions that had associated error events.

Conditional metrics require their parent event types plus the relevant dimension fields to be active.


Dimensions

Dimensions are the fields you break your metrics down by — the rows and columns in your reports and analyses.

Every field your data source provides and your configuration enables is a potential dimension. Common categories:

Category

Examples

Category

Examples

Content & Inventory

What content was playing, which site/section/placement

Ad Details

Creative IDs, ad unit types, position (pre/mid/postroll), CPM

Ad Break & Slot

Break structure, slot position, fill capacity, unfilled time

User & Device

Country, region, device type, platform/player, user agent

Traffic Quality

Invalid traffic flags and reason codes

Partners & Value Chain

Distribution partners, selling partners, content providers

Marketplace & Programmatic

Deal IDs, buyer seats, sales channel type

Audience & Targeting

Audience segments, matched targeting items, privacy choices

Which specific dimensions are available depends on your data source and configuration. See your integration's field reference for the complete list.


Filters

Filters narrow the data that metrics and dimensions operate on. You can apply them at multiple levels:

Level

Scope

Example

Level

Scope

Example

Report-level

Every visualisation in a report

country = US across all charts

Block-level

A single chart or table within a report

deviceType = CTV on one timeseries only

Analysis

Applied interactively during ad-hoc exploration

Add and remove filters as you investigate

Monitor

Defines the scope of an anomaly or validation monitor

Monitor error rate only for distributionPartnerId = X

Filters use dimensions as their criteria. They can be combined (AND logic) and many dimensions support multi-value selection (e.g., country IN [US, GB, AU]).


Putting It Together

Think of it as two axes:

  • Vertical (metrics): What do you want to measure? Impressions? Fill rates? Revenue? Error rates? Each requires specific event types.

  • Horizontal (dimensions): How do you want to slice it? By device? By partner? By geography? Each requires specific fields.

The intersection — metrics broken down by dimensions, narrowed by filters — is what powers Reports, Analysis, Inspect, and Monitoring.


Configuration and Availability

Most organisations start with a core set of event types and dimensions, then extend over time. What's available to you depends on:

Factor

What it determines

Factor

What it determines

Your data source(s)

Different integrations provide different event types and fields

Your configuration

Which event types and fields are enabled for your organisation

Cross-event dependencies

Some derived metrics require multiple event types to be active

Your account team can walk you through what's currently active and what's available to add.


Data Source References

For the specific metrics and dimensions available per integration:

Integration

What's documented

Integration

What's documented

FreeWheel: Metrics & Dimensions

Event types, metrics catalogue, field reference for FreeWheel v4 log file customers

FreeWheel Metrics Reference

Complete standard metrics catalogue organised by event type

Getting Started: FreeWheel Setup Guide

Step-by-step setup walkthrough with day-one examples