Resources

Metrics Encyclopedia

Your guide to understanding video, advertising, and performance metrics.

Metric Categories

Audience & Engagement

Filters: app_session_id, content_session_id, device_type, city, country, os_name, os_version, ASN, network_type, asset_id, title

Quality of Experience (QoE)

Filters: app_session_id, content_session_id, device_type, city, country, os_name, os_version, ASN, network_type, asset_id, title

Advertising

Filters: advertiser_name, ad_partner, ad_break_id, app_session_id, content_session_id, device_type, city, country, os_name, os_version, ASN, network_type

Revenue

Filters: advertiser_name, ad_partner, ad_system, ad_creative_ID, ad_break_id, ad_position, app_session_id, content_session_id, device_type, country, asset_id, title, user_id

Delivery

Filters: TBD

Metric Details

Ad Break Duration

Category: Advertising
Datazoom Definition: The duration of an ad break. Can be down to a single ad break or used as an average.
Required Data Points: ad_break_end, ad_break_start
Potential Insight: Tracking ad break duration provides a balance between monetization and user experience. For example, measuring length of ad break to completion rates for understanding user behavior.

(ad_break_end - ad_break_start)

Ad Click Through Rate

Category: Revenue
Datazoom Definition: The rate at which ads clicked. When a cohort or content can be identified with a higher click through rate, that ad space can be sold for a premium.
Required Data Points: ad_click, ad_impression
Potential Insight: Provides an indication of how a particular ad is performing to engage the user. High click-through rates indicate high interest from users.

count(ad_click)/count(ad_impression)*100

Ad Completion Rate

Category: Advertising, Revenue
Datazoom Definition: The rate at which an ad is played to completion. When a cohort or content can be identified with a higher completion rate, that ad space can be sold for a premium.
Required Data Points: playback_complete_ad, ad_impression, media_type
Potential Insight: An indication of how a particular ad is performing to engage the user. High completion rates indicate high interest from users.

count(playback_complete where media_type = ad)/count(ad_impression)*100

Ad Error Rate

Category: Advertising
Datazoom Definition: The percentage of ads that failed to play.
Required Data Points: error, ad_impression, media_type
Potential Insight: A high ad error rate can lead to lost revenue and a poor user experience.

count(error where media_type = ad)/count(ad_impression)*100

Ad Errors

Category: Advertising
Datazoom Definition: Total ad errors.
Required Data Points: error, media_type
Potential Insight: Tracking ad errors is critical for any ad-based video business. Errors mean lost revenue and a poor user experience.

count(error where media_type = ad)

Ad Fill Rate (CSAI)

Category: Advertising
Datazoom Definition: The ratio of ads served to ads requested, expressed as a percentage.
Required Data Points: playback_start, media_request, media_type
Potential Insight: Helps understand the efficiency of the ad inventory. A low fill rate might indicate issues with ad supply, targeting, or configuration.

count(playback_start where media_type = ad) / count(media_request where media_type = ad) * 100

Ad Impressions

Category: Advertising
Datazoom Definition: Total ad impressions.
Required Data Points: ad_impression
Potential Insight: Ad impressions are a critical KPI for any ad-based video business. They are used to calculate other important metrics, such as ad revenue and ad completion rate.

count(ad_impression)

Ad Playback Reliability

Category: Advertising
Datazoom Definition: Measures the percentage of ads that start and play without errors.
Required Data Points: error, media_request, media_type, playback_start
Potential Insight: Provides a comprehensive view of the ad playback health, combining successful starts with error-free playback.

count(playback_start where media_type = ad) - count(error where media_type = ad)) / count(media_request where media_type = ad) * 100

Ad Requests

Category: Advertising
Datazoom Definition: Total ad requests.
Required Data Points: media_request, media_type
Potential Insight: Useful for tracking ad system delivery performance and load. Also useful for monetization metrics like ad requests versus ad impressions.

count(media_request where media_type = ad)

Ad Revenue

Category: Revenue
Datazoom Definition: Total ad revenue for a specific attribute.
Required Data Points: cpm (requires enrichment)
Potential Insight: Helps identify which attributes are performing best and generating the most revenue.

SUM(cpm) GROUP BY any metadata

Ad Stall Rate

Category: Quality of Experience (QoE), Advertising
Datazoom Definition: The percentage of ad playbacks that experience a stall.
Required Data Points: media_type, playback_start, stall_start
Potential Insight: Ad stalls are a major cause of user frustration and can lead to ad abandonment. A high ad stall rate may indicate issues with ad delivery or the user's network connection.

count(stall_start where media_type = ad)/count(playback_start where media_type = ad)*100

Ad Start Failures

Category: Advertising
Datazoom Definition: The percentage of ad attempts where a fatal error occurred before the first frame of the ad was played.
Required Data Points: ad_break_start, ad_sesssion_id, media_type, playback_start
Potential Insight: Ad start failures are a major cause of lost revenue and a poor user experience. A high ad start failure rate may indicate issues with ad delivery or the user's device.

count(ad_session_id where error occurs after media_request and before playback_start where media_type = 'ad') / count(media_request where media_type = 'ad') * 100

Ad Start Time

Category: Quality of Experience (QoE), Advertising
Datazoom Definition: The time it takes for an ad to start playing after the content has been paused for the ad break.
Required Data Points: media_request, media_type, playback_start
Potential Insight: Ad start time is a critical KPI for measuring user experience. Long ad start times can lead to user frustration and ad abandonment.

timestamp of playback_start where media_type is 'ad' - timestamp of media_request where media_type is 'ad'

Ad Tolerance

Category: Advertising
Datazoom Definition: The average number of ads a user watches per session before dropping off.
Required Data Points: ad_impression, content_session_id
Potential Insight: Helps optimize ad load and frequency to maximize revenue without negatively impacting user retention.

COUNT(ad_impression)/COUNT(content_session_id)

Ad Watched Time

Category: Advertising
Datazoom Definition: The total time, in seconds or minutes, that users have spent watching ads.
Required Data Points: ad_playback_duration
Potential Insight: A key metric for measuring overall ad engagement and is often used in revenue calculations with ad partners.

SUM(ad_playback_duration)

Adaptive Bitrate (ABR) Drift Analysis

Category: Delivery
Datazoom Definition: Monitors the frequency and magnitude of bitrate switches.
Required Data Points: ad_impression, media_type, rendition_change
Potential Insight: Frequent switching or large drops in bitrate can indicate network instability or poor CDN performance, affecting user experience.

count(rendition_change where media_type = ad)/count(ad_impression)

Audience Penetration

Category: Revenue
Datazoom Definition: Percentage of viewers watching a selected title (grouped by title).
Required Data Points: asset_id, playback_start
Potential Insight: Useful measure of what content is most watched at a particular point in time to help with content optimization.

count(playback_start where asset_id = PLACEHOLDER) / count(playback_start) * 100

Average Bitrate

Category: Quality of Experience (QoE)
Datazoom Definition: The average bitrate of the video segments that are being played.
Required Data Points: bitrate
Potential Insight: Average bitrate is a good indicator of video quality. A higher average bitrate generally means better video quality.

avg(bitrate)

Average CPM

Category: Revenue
Datazoom Definition: The average cost per thousand ad impressions for any attribute.
Required Data Points: cpm (requires enrichment)
Potential Insight: Understanding which content commands a higher CPM can guide content strategy and sales efforts.

AVG(cpm) GROUP BY any attribute

Average Frame Rate

Category: Quality of Experience (QoE)
Datazoom Definition: The average number of frames displayed per second during video playback.
Required Data Points: frame_rate
Potential Insight: A low or fluctuating frame rate can result in choppy or jerky video, negatively impacting the viewing experience. This metric helps diagnose rendering performance issues on the client-side.

avg(frame_rate)

Average Minute Audience

Category: Audience & Engagement
Datazoom Definition: The average number of users watching a piece of content in any given minute.
Required Data Points: duration, playback_duration_content, asset_id
Potential Insight: A good measurement of audience size for a given piece of content. Can be used to track audience engagement and content performance.

=SUM(playback_duration_content)/SUM(duration) GROUP BY asset_id

Average Playback Stalls

Category: Quality of Experience (QoE)
Datazoom Definition: The average number of playback stalls per viewing session.
Required Data Points: stall_start, content_session_id
Potential Insight: Helps quantify the frequency of interruptions a typical user experiences, providing a clear indicator of playback performance.

count(stall_start) / count(content_session_id)

Bitrate Selection Efficiency

Category: Delivery
Datazoom Definition: Compares the selected bitrate to the available bandwidth.
Required Data Points: bandwidth, rendition_video_bitrate
Potential Insight: Determines if the player is making optimal bitrate choices. Consistently choosing a bitrate lower than what the network can support may indicate a problem with the ABR algorithm.

sum(rendition_video_bitrate) / sum(bandwidth) * 100

Buffer Ratio

Category: Quality of Experience (QoE)
Datazoom Definition: The ratio of buffer length to the content's duration.
Required Data Points: buffer_duration, playback_duration_content
Potential Insight: Indicates the health of the playback buffer. A consistently low buffer ratio can be a leading indicator of future stalling events.

sum(buffer_duration) / sum(playback_duration_content) * 100

Concurrent App Sessions

Category: Quality of Experience (QoE)
Datazoom Definition: The number of users who have the application open at the same time.
Required Data Points: app_session_id
Potential Insight: Provides insight into the app's peak usage and server load, helping with capacity planning and understanding user engagement patterns.

count(distinct app_session_id) at a given time

Concurrent Plays

Category: Audience & Engagement
Datazoom Definition: The number of simultaneous video playbacks occurring at a specific point in time.
Required Data Points: content_session_id
Potential Insight: A key metric for understanding peak traffic and load on the delivery infrastructure. Essential for capacity planning and managing CDN costs.

count(distinct content_session_id) at a given time

Content Delivery Consistency Score

Category: Delivery
Datazoom Definition: A composite score based on various delivery metrics (e.g., latency, throughput, error rates) across different CDNs or regions.
Required Data Points: time_to_first_byte_ms, measured_throughput, http_status_code, cdn, region, country_code, city
Potential Insight: Provides a single, high-level indicator of overall delivery performance and consistency.

(w1 * (1 / avg(time_to_first_byte_ms))) + (w2 * avg(measured_throughput)) - (w3 * count(http_status >= 400))

Content Drop Off

Category: Audience & Engagement
Datazoom Definition: The percentage of users who stop watching a piece of content before it is finished.
Required Data Points: content_session_id, playhead_position
Potential Insight: A good measurement of audience engagement for a given piece of content. Can be used to identify content that is not performing well.

dist(for each content_session_id(max(playhead) - min(playhead)))

Content Penetration

Category: Revenue
Datazoom Definition: Content as a percentage of total watched time.
Required Data Points: playback_duration_content, asset_id, media_type
Potential Insight: Useful to determine which video titles are popular by users to inform content decisions.

( sum(last(playback_duration_content) where asset_id = "Specific Asset ID") / sum(last(playback_duration_content) where media_type = 'content') ) * 100

Content Session Time

Category: Audience & Engagement
Datazoom Definition: The total amount of time a user spends in a single content viewing session, including any time spent paused or buffering.
Required Data Points: playback_duration_content
Potential Insight: Helps understand user engagement with specific pieces of content. Longer session times generally indicate higher engagement.

sum(playback_duration_content)

Content that drives Subscription

Category: Revenue
Datazoom Definition: Titles most watched by subscribed users.
Required Data Points: custom_subscription_status, media_type, playback_start, asset_id
Potential Insight: Useful to determine which video titles are watched by subscribers to inform content decisions.

count(playback_start) where custom_subscription_status = subscribedand media_type = content grouped by asset_id

Conversion to Subscribe

Category: Revenue
Datazoom Definition: Ratio of subscribed users to identified returning viewers.
Required Data Points: custom_subscription_id, custom_user_id
Potential Insight: Measures how many returning users have subscribed to the service. Useful KPI to track retention and marketing campaigns for converting users to subscription levels.

count(unique custom_subscription_id)/count(unique custom_user_id) * 100

Dynamic Network Congestion Monitoring

Category: Delivery
Datazoom Definition: Real-time tracking of network congestion indicators like packet loss, latency, and jitter.
Required Data Points:
Potential Insight: Allows for proactive traffic shaping and routing decisions to avoid performance degradation.

weighted_avg(avg_packet_loss_rate, avg_jitter, avg_round_trip_time)

End-to-End Session Resilience

Category: Delivery
Datazoom Definition: Measures a session's ability to recover from non-fatal errors without termination.
Required Data Points: app_session_id, num_errors
Potential Insight: Highlights the robustness of the player and delivery chain in handling transient issues.

count(app_session_id where num_errors = 0)/count(app_session_id) * 100

Exit Before Video Start (EBVS)

Category: Quality of Experience (QoE)
Datazoom Definition: The percentage of users who abandon a video before the first frame is played.
Required Data Points: app_session_id, content_session_id
Potential Insight: EBVS is a critical KPI for measuring user experience. A high EBVS rate may indicate that video start times are too long, or there are other issues with the user experience.

count(application_session_id where content_session_id = null/count(application_session_i)*100

Exits After Error

Category: Quality of Experience (QoE)
Datazoom Definition: The number of users who close the session immediately after a playback error.
Required Data Points:
Potential Insight: A high number of exits after an error indicates that the errors are severe and user-frustrating, highlighting critical failure points.

count(sessions) where exit occurs within X seconds of an error event

Exits During Ad Break

Category: Audience & Engagement, Advertising
Datazoom Definition: The percentage of viewers who abandon the content during an ad break.
Required Data Points:
Potential Insight: Helps in analyzing ad fatigue and the impact of ad load on user experience. High rates may suggest that ad breaks are too long or too frequent.

(count(exits during ad break) / count(sessions reaching ad break)) * 100

Exits During Stall

Category: Quality of Experience (QoE)
Datazoom Definition: The number of users who abandon the video playback during a rebuffering event.
Required Data Points:
Potential Insight: A direct measure of user frustration with buffering. A high value indicates that stall durations are intolerably long for viewers.

count(sessions) where exit event follows a stall event

Flag interesting segments

Category: Delivery
Datazoom Definition: Identifies and flags time segments within a video that have high rates of errors, buffering, or quality degradation.
Required Data Points:
Potential Insight: Allows for targeted analysis of problematic content segments, which could be related to encoding issues or specific scenes causing performance drops.

IDENTIFY content_segments WHERE stall_count > threshold_X OR avg_bitrate < threshold_Y

Interval Minutes

Category: Audience & Engagement
Datazoom Definition: The total number of minutes watched in a specific time interval (e.g., per day, per week).
Required Data Points:
Potential Insight: Useful for trend analysis and understanding patterns of consumption over time.

SUM(viewing_minutes) GROUP BY time_interval

Login Success Rate

Category: Audience & Engagement
Datazoom Definition: Ratio of identified/returning users to total viewers.
Required Data Points:
Potential Insight: Provides an indication of how many users are accessing the system through logging in rather than browsing without logging in.

total viewers / viewers with userID * 100

Monthly Viewing Users

Category: Audience & Engagement
Datazoom Definition: The total number of unique users who viewed content in a given month.
Required Data Points:
Potential Insight: A standard metric for tracking audience size and growth over a longer period.

count(distinct user_id) per month

Pageviews with Video Playback

Category: Audience & Engagement
Datazoom Definition: Track the number of pageviews that converts into video playback.
Required Data Points:
Potential Insight: Provides a measurement of what pages have video content users want to watch. Can help optimize which video subjects/topics drive engagement.

count(page_sessions where playback_start_event exists) / count(total_page_sessions) * 100

Per-Region QoE Disparity Index

Category: Delivery
Datazoom Definition: A measure of the variance in QoE scores between different geographical regions.
Required Data Points:
Potential Insight: Helps identify regions with consistently poor performance, pointing to potential CDN or ISP peering issues.

stdev(avg_SExI_score per country)

Percent Complete

Category: Audience & Engagement
Datazoom Definition: The percentage of a video's total duration that a user has watched.
Required Data Points:
Potential Insight: Provides a granular view of user engagement with a piece of content. It can be used to identify drop-off points within the video.

(max(playhead_position) / content_duration) * 100

Play Attempts

Category: Audience & Engagement
Datazoom Definition: The total number of times users initiated a playback attempt (e.g., by clicking a play button).
Required Data Points:
Potential Insight: Measures user intent to watch. Comparing this to Play Starts helps calculate the Video Start Failure rate.

count(play_request)

Play Starts

Category: Audience & Engagement
Datazoom Definition: The number of times a video playback successfully begins and the first frame is rendered.
Required Data Points:
Potential Insight: A fundamental metric for measuring consumption. It indicates that the video successfully loaded and began playing.

count(playback_start)

Playback Failures

Category: Quality of Experience (QoE)
Datazoom Definition: The percentage of playback attempts that result in a fatal error.
Required Data Points:
Potential Insight: Playback failures are a major cause of user frustration and can lead to churn. A high playback failure rate may indicate issues with the user's device, network connection, or the video content itself.

count(sessions with a fatal error)/count(total sessions)*100

Playback Stall Rate

Category: Quality of Experience (QoE)
Datazoom Definition: The frequency of stalls during playback, often measured as stalls per hour of viewing.
Required Data Points:
Potential Insight: Normalizes stall occurrences over time, providing a comparable metric across sessions of different lengths.

(total_stall_count / total_viewing_hours)

Playback Stall Ratio

Category: Quality of Experience (QoE)
Datazoom Definition: The ratio of time spent stalling to the total session time.
Required Data Points:
Potential Insight: A direct measure of how much of a user's time is wasted waiting for the video to buffer, offering a clear view of the severity of stalling issues.

(total_stall_duration / total_session_time) * 100

QoE-to-QoS Correlation Index

Category: Delivery
Datazoom Definition: Correlates subjective QoE metrics (like SExI) with objective QoS metrics (like bitrate, latency).
Required Data Points:
Potential Insight: Helps understand which technical performance metrics have the most significant impact on the actual user experience.

correlation_coefficient(SExI_score, stall_ratio)

Rebuffer Rate

Category: Quality of Experience (QoE)
Datazoom Definition: The percentage of viewing sessions that experience at least one rebuffer event.
Required Data Points:
Potential Insight: Rebuffer rate is a critical KPI for measuring user experience. A high rebuffer rate may indicate issues with the user's network connection or the video delivery infrastructure.

count(sessions with stall)/count(total sessions)*100

Session-Level Playback Latency Breakdown

Category: Delivery
Datazoom Definition: Breaks down the end-to-end latency into its constituent parts (e.g., DNS lookup, TCP connection, time to first byte).
Required Data Points:
Potential Insight: Pinpoints the exact stage in the delivery process where delays are occurring.

BREAKDOWN(time_to_first_frame by [dns_lookup, tcp_connect, ssl_handshake, time_to_first_byte, download_time])

SExI (Streaming Experience Index)

Category: Quality of Experience (QoE)
Datazoom Definition: A composite score that combines multiple QoE metrics (e.g., start time, buffering, bitrate) into a single number representing the overall user experience.
Required Data Points:
Potential Insight: Provides a holistic, at-a-glance view of streaming quality, making it easier to track and compare performance over time or across different user segments.

Composite score based on VST, Stall Ratio, and Average Bitrate

Stall Cause Attribution Metric

Category: Delivery
Datazoom Definition: Attempts to attribute stalls to specific causes (e.g., CDN, ISP, client-side buffer management).
Potential Insight: Moves beyond just identifying that a stall happened to understanding why it happened, enabling more effective troubleshooting.

DISTRIBUTION(stall_events by attributed_cause)

Stall Ratio

Category: Quality of Experience (QoE)
Datazoom Definition: The proportion of total viewing time spent in a rebuffering state.
Required Data Points:
Potential Insight: Provides context to the Rebuffer Rate by showing the severity of the rebuffering events. A high Stall Ratio means users are spending a significant amount of their time waiting.

total stall duration / (total stall duration + total playback duration) * 100

Total Sponsor Impressions

Category: Advertising
Datazoom Definition: The total number of times a sponsor's ad or branding was displayed.
Required Data Points:
Potential Insight: Measures the reach of a sponsorship campaign, a key metric for reporting to sponsors and calculating ROI.

count(sponsor_impressions)

Total Sponsor Viewing Time

Category: Advertising
Datazoom Definition: The aggregate time users spent watching sponsored content or ads.
Required Data Points:
Potential Insight: Provides a measure of engagement with sponsored content, offering more insight than impressions alone.

SUM(sponsor_ad_duration)

Total Time Watched

Category: Audience & Engagement
Datazoom Definition: The aggregate amount of time all users have spent watching content.
Required Data Points:
Potential Insight: A fundamental measure of overall platform engagement and content consumption.

SUM(content_playback_duration)

Total Viewing Hours

Category: Audience & Engagement
Datazoom Definition: The total number of hours of content that have been watched by all users.
Required Data Points:
Potential Insight: A good measurement of overall service delivery scale for a given time period. Can be used to track service delivery scale and performance.

=sum(viewing_minutes)/60

Unique Users

Category: Audience & Engagement
Datazoom Definition: The number of distinct users who have watched content on the service.
Required Data Points:
Potential Insight: A good measurement of audience size. Can be used to track audience growth and reach.

=count(unique(user_id))

User Stickiness

Category: Revenue
Datazoom Definition: Measures how frequently users return to the service (e.g., DAU/MAU ratio).
Required Data Points:
Potential Insight: A key indicator of user retention and loyalty. A high stickiness ratio suggests the service provides lasting value to its users.

Daily Active Users (DAU) / Monthly Active Users (MAU)

Video Buffering Time

Category: Quality of Experience (QoE)
Datazoom Definition: The total time a video spends in a buffering state during a viewing session.
Required Data Points:
Potential Insight: Directly measures the amount of time a user is kept waiting. Minimizing this is crucial for a good user experience.

SUM(stall_duration) per session

Video Engagement Ratio

Category: Audience & Engagement
Datazoom Definition: Ratio of time engaged on a page vs the time spent watching a video on that page.
Required Data Points:
Potential Insight: Provides insight into how to optimize the site for higher video consumption and track which videos drive engagement.

SUM(total_video_playback_duration_on_page) / SUM(total_time_spent_on_page) * 100

Video Error Rate

Category: Quality of Experience (QoE)
Datazoom Definition: The percentage of video playback sessions that encounter a fatal error.
Required Data Points:
Potential Insight: A high-level indicator of playback reliability. Spikes in this metric can signal widespread issues with content, CDNs, or player versions.

(count(sessions with errors) / count(total sessions)) * 100

Video Interruption Count

Category: Quality of Experience (QoE)
Datazoom Definition: The total number of interruptions (stalls, errors) in a viewing session.
Required Data Points:
Potential Insight: Offers a simple, aggregate count of how many times a user's experience was disrupted, regardless of the cause.

count(stalls) + count(errors)

Video Interruption Ratio

Category: Quality of Experience (QoE)
Datazoom Definition: The ratio of interrupted viewing sessions to total viewing sessions.
Potential Insight: Measures the likelihood that a user will encounter any kind of interruption, providing a broad measure of overall session quality.

(count(sessions with interruptions) / count(total sessions)) * 100

Video Start Failures

Category: Quality of Experience (QoE)
Datazoom Definition: The percentage of attempts where a fatal error occurred before the first frame of video was played.
Required Data Points:
Potential Insight: VSF is a critical KPI for measuring user experience. A high VSF rate may indicate that video start times are too long, or there are other issues with the user experience.

count(sessions with a fatal error before playback start)/count(total sessions)*100

Video Start Time

Category: Quality of Experience (QoE)
Datazoom Definition: The time it takes for a video to start playing after the user clicks play.
Required Data Points:
Potential Insight: Video start time is a critical KPI for measuring user experience. Long video start times can lead to user abandonment.

=time_to_first_frame

Viewing Minute Distribution

Category: Audience & Engagement
Datazoom Definition: Similar to the “content drop off” metric but ignores scrubbing. Should be the same result for a live stream.
Required Data Points:
Potential Insight: A good measurement of overall service delivery scale for a given time period. Can be used to track service delivery scale and performance.

dist(for each content_session_id(count(heartbeat)/heartbeat.frequency))