Metrics and definitions – Assembled

Overview

This article defines metrics available in Assembled reports, Team Performance, adherence views, Scorecards, and related dashboards. It covers schedule and time-tracking metrics (such as adherence, utilization, and occupancy) and ticket and message metrics (such as handle time, cases solved, and CSAT). Use this as a reference when building reports, explaining data to stakeholders, or troubleshooting discrepancies.

Important note: Many metrics rely on productive time. Productive time is determined by:

Event types (Settings → Event types): which scheduled events count as productive, non-productive, or time off.
Agent state mappings (Settings → Agent states): how statuses from your contact center or CRM (for example Zendesk, Talkdesk, Five9) map to productive, available, busy, or offline in Assembled.
Integrations and queue mapping (Settings → Queues and exclusions): how tickets and agent activity are assigned to channels and queues.

Time spent working on tickets comes from your support platform. Agent activity may be tracked through status selections (aux codes), automatic browser or app activity, or integration syncs, depending on your setup.

How common metrics differ:

Metric	What it measures
Productive adherence	Right activity at the right time during scheduled productive blocks
Schedule adherence	Following all scheduled events (except time off)
Conformance	Right amount of productive time worked vs scheduled (timing does not matter)
Scheduled utilization	Share of the schedule allocated to productive events
Actual utilization	Share of actual worked time that was productive
Occupancy	Share of logged-in time spent actively on customer work
In-office shrinkage	Share of scheduled time spent in non-productive states

High productive adherence does not guarantee high occupancy. An agent can be fully adherent during a productive block but idle when contact volume is low.

What's in this article:

Overview
Schedule and adherence metrics
Productivity and utilization metrics
Case resolution and response metrics
Quality, case management, and messaging metrics
Configuration that commonly changes metrics
When numbers do not match expectations

Schedule and adherence metrics

Actual hours worked (total)

Number of hours the agent worked during the specified interval, regardless of their schedule. This includes both productive and non-productive actual time.

Calculation: sum(hours worked)

Notes:

Based on agent state data and your agent state mappings.
Used as the denominator for Actual utilization and related metrics.

Actual productive hours worked

Number of productive hours the agent worked during the specified interval, regardless of whether they followed their schedule.

Calculation: sum(productive hours worked)

Notes:

Only includes time in agent states mapped to productive (and, if enabled for your account, productive auxiliary states).
Used as the numerator for Conformance and as the denominator for Solved (per actual hour).

Actual productive hours worked (all channels)

Number of productive hours worked across all channels and platforms, including productive auxiliary time when that configuration is enabled for your account.

Calculation: sum(productive hours worked across all channels)

Notes:

Appears when your account uses expanded productive or auxiliary event definitions.
May differ from Actual productive hours worked when auxiliary productive states are included.

Hours in productive adherence

Total time the agent was in the correct productive state while scheduled for productive work during the specified interval.

Calculation: sum(hours in productive adherence)

Notes:

This is the hour-based companion to Productive adherence (percentage).
Requires agent state data and correct agent state mappings.

Productive adherence

Percentage of time an agent was in the correct productive state while scheduled for productive work.

Calculation: (hours in productive adherence during scheduled productive time ÷ scheduled productive hours) × 100

Notes:

Maximum value is 100%.
Only evaluates periods when the agent is scheduled for events marked productive in Event types.
Requires agent state data synced from your platforms and correct agent state mappings.
Agent state exclusions (for example training, outages, or approved exceptions) remove affected periods from adherence calculations when configured.
If capacity is split across queues or channels for your account, adherence time may be allocated proportionally across those dimensions.

How this differs from schedule adherence: Productive adherence only considers scheduled productive events. Schedule adherence considers all scheduled events except time off (meetings, breaks, training, and so on).

How this differs from conformance: Adherence cares about doing the right thing at the right time. Conformance cares about working the right total amount of productive time, regardless of timing.

Productive adherence (all)

Same as Productive adherence, but calculated using productive and productive-auxiliary definitions across all channels when that configuration is enabled for your account.

Calculation: (hours in productive adherence during scheduled productive time ÷ scheduled productive hours) × 100

Notes:

Appears when your account uses expanded productive auxiliary event types.
Compare to Productive adherence only when you understand whether auxiliary time is included in each metric.

Hours in schedule adherence

Total time the agent followed their full schedule (all scheduled events except time off) during the specified interval.

Calculation: sum(hours in schedule adherence)

Schedule adherence

Percentage of time an agent followed all scheduled events except time off.

Calculation: (hours in schedule adherence during scheduled events ÷ total scheduled hours) × 100

Notes:

Maximum value is 100%.
Time off event types are excluded from both numerator and denominator.
Includes adherence to non-productive scheduled events (breaks, meetings, training) as well as productive events.
Often lower than productive adherence because more scheduled activity is included.

Scheduled hours (total)

Time the agent was scheduled to work on any event during the specified interval.

Calculation: sum(all scheduled hours)

Notes:

Includes productive and non-productive scheduled events.
Excludes time off when calculating metrics that use "scheduled hours (total)" as defined in reporting (see Schedule adherence and In-office shrinkage).

Scheduled hours (productive)

Time the agent was scheduled to work on productive events during the specified interval.

Calculation: sum(scheduled productive hours)

Notes:

Used as the denominator for Productive adherence and Conformance.
Does not indicate whether the agent actually worked those hours — only what was on the schedule.

Conformance

Percentage of actual productive hours worked compared to scheduled productive hours, regardless of whether the agent was in the correct state at each moment.

Calculation: (actual productive hours worked ÷ scheduled productive hours) × 100

Notes:

Can be greater than 100% if the agent worked more productive time than scheduled.
Can be less than 100% if the agent worked fewer productive hours than scheduled.
Timing does not matter: an agent can have high conformance but low adherence (or the reverse).
Useful for flexible teams where total productive hours matter more than minute-by-minute schedule match.
When you filter by channel or queue, conformance is calculated in that filter context (productive actual vs productive scheduled for that scope).

How this differs from productive adherence: Adherence requires matching scheduled productive events with the correct agent states in real time. Conformance only compares total productive hours worked to total productive hours scheduled.

Conformance (strict)

Percentage of all actual hours worked (productive and non-productive) compared to all scheduled hours (productive and non-productive, excluding time off).

Calculation: (actual productive and non-productive hours worked ÷ scheduled productive and non-productive hours) × 100

Notes:

Can exceed 100% for the same reasons as Conformance.
Includes non-productive scheduled and actual time (for example breaks and meetings), not only customer-facing productive time.
Available in reporting when enabled for your account; not all customers see this metric in every view.

Out of adherence duration

How long an agent has been out of adherence for their currently scheduled productive event in real-time adherence views.

Calculation: duration of current out-of-adherence period

Notes:

Typically shown in real-time or agent-level adherence views, not as a historical aggregate in all reports.
Measured in seconds or hours depending on the report.

Productivity and utilization metrics

Email productivity (case throughput)

Average number of email contacts handled per agent per hour.

Calculation: average(email contacts handled per hour)

Notes:

Specific to the email channel and your queue and channel configuration.
Distinct from Messages sent (per scheduled hour) or Messages sent (per actual hour), which count outbound messages rather than completed contacts.

Occupancy

Percentage of logged-in time an agent spent actively working on customer issues. Occupied and available states are defined by admins in agent state mappings.

Calculation: occupied time ÷ (occupied time + available time) × 100

Notes:

Only includes agent states mapped to occupied (busy) or available for the channel or platform in scope.
Does not measure schedule compliance — that is adherence.
Does not measure whether scheduled time was productive — that is utilization.
For channel-specific occupancy, only states mapped to that channel are included.

Industry context (informational only; Assembled does not enforce targets):

Many contact centers aim for roughly 70–85% occupancy on real-time channels.
Sustained occupancy above ~85% may indicate burnout risk.
Occupancy below ~70% may indicate underutilization or low volume.

Example state mappings (your configuration may differ):

Platform	Often mapped as occupied (busy)	Often mapped as available
Zendesk Talk	On call, wrap-up	Online
Five9	On call, after-call work, on hold, on park	Ready
Kustomer, Twilio, UJET, Talkdesk	Depends on your agent state mappings	Depends on your mappings

Read more about configuring occupancy in your Assembled help center (search for "configurable occupancy").

Occupancy (all channels)

Percentage of logged-in time spent actively on customer work across all channels and platforms configured in agent state mappings.

Calculation: occupied time (all channels) ÷ available and occupied time (all channels) × 100

Notes:

Use this when you need a single occupancy number across phone, chat, and other channels.
Compare to channel-specific Occupancy when diagnosing a single channel.

Scheduled utilization

Percentage of scheduled time allocated to productive events, whether or not the agent followed the schedule.

Calculation: (scheduled productive hours ÷ total scheduled hours) × 100

Notes:

Answers: "Are we scheduling the right mix of customer work vs meetings, breaks, and other events?"
Time off is excluded from the denominator (total scheduled hours for this metric uses scheduled time excluding time off, consistent with utilization reporting).
Does not use actual agent states — only the schedule.

Actual utilization

Percentage of actual worked time spent in productive states, regardless of adherence to the schedule.

Calculation: (actual productive hours worked ÷ actual hours worked (total)) × 100

Notes:

Answers: "Of the time agents actually worked, how much was productive?"
Ignores the schedule except indirectly (actual hours come from agent states).
Preferred over legacy utilization definitions that divided by scheduled hours.

How this differs from scheduled utilization: Scheduled utilization uses the plan. Actual utilization uses what happened.

Actual utilization (all channels)

Percentage of actual worked time spent in productive and productive-auxiliary states across all channels when that configuration is enabled.

Calculation: (actual productive and auxiliary productive hours worked ÷ actual hours worked (total)) × 100

Notes:

Appears when your account includes auxiliary productive time in utilization reporting.

In-office shrinkage

Percentage of scheduled time spent in non-productive states (for example breaks, meetings, training, or time out of adherence on productive events).

Calculation: actual hours in non-productive states ÷ total scheduled hours

Notes:

Only includes hours during scheduled periods; time off is excluded.
Non-productive states must be mapped correctly in agent state mappings and Event types.
Not reliable for async channels (email, messaging): agents may appear "available" while not actively working. For async workflows, use Actual utilization or Solved (per actual hour) instead of shrinkage alone.
Your organization may separately track planned vs unplanned shrinkage in BPO or vendor-requirement workflows.

How this differs from conformance: Shrinkage measures non-productive share of scheduled time. Conformance measures whether total productive hours matched the schedule.

Actual non-productive hours

Total time an agent spent in non-productive auxiliary or non-productive states during the interval.

Calculation: sum(actual non-productive hours)

Notes:

Appears when your account uses expanded auxiliary state definitions.
Contributes to In-office shrinkage at the percentage level.

Case resolution and response metrics

Handle time

Total time to resolve contacts in the selected time range (typically using cases or contacts by created date, depending on the report).

General calculation: time contact ended − time contact started (definition varies by platform and channel)

Note: Assembled does not calculate handle time for email.

Platform-specific definitions:

Kustomer

Phone: conversation end time − first response time (often endedAt vs first response timestamp; exact fields depend on conversation attributes).
Chat: conversation end time − first response time. Bot-handled portions may be excluded if your account is configured to ignore bot messages when determining the first response.

Zendesk

Zendesk Talk (phone): call end time − call start time (often created_at + duration; waiting and wrap-up may be excluded depending on configuration).
Zendesk Chat: chat end time − first response time (often from Zendesk Chat API combined with ticket data). Waiting time before first response and wrap-up after the chat may be excluded.
Zendesk Messaging / ticket-based chat: often solved_at relative to first reply time (reply_time_in_minutes or reply_time_in_seconds on the ticket metric set).

Salesforce

Case (phone): ClosedDate − firstRespondedAt
Case (chat): ClosedDate − firstRespondedAt
LiveChatTranscript (chat): EndTime − StartTime

Talkdesk

Phone: end_at − start_at for handle duration; first-response timing may use start_at + wait_time or start_at + total_time − talk_time depending on your account configuration.

UJET

Phone: ends_at − connected_at (or queue timing depending on call flow).
Chat: ends_at − first message sent (handle duration start from first handle segment).

Five9

Phone: (timestamp + call_time) − timestamp (equivalent to call duration from answer/end timestamps).

Amazon Connect

Phone: AgentInteractionDuration + CustomerHoldDuration + AfterContactWorkDuration

Intercom

Chat: last admin/lead/user conversation part time − first admin response part time. Bot-authored parts may be filtered depending on conversation structure.

Troubleshooting handle time discrepancies:

Compare the same channel and date definition (created vs solved date).
Confirm whether hold, wrap-up, and queue wait are included on your platform vs in Assembled.
Open metric calculation details in the product when available for your integration.
Account-specific settings (for example excluding bots on Kustomer, Talkdesk wait-time handling) change results.

Average handle time

Average time to resolve a phone or chat contact in the selected time range.

Calculation: sum(handle time) ÷ number of contacts

Notes:

Excludes email.
With units of work enabled, you may also see Average handle time per case or Average handle time per unit of work — see below.

Average handle time per case

Average time to resolve a phone or chat case in the selected range when your account uses tickets v2 / units-of-work reporting with backward-compatible labeling.

Calculation: sum(handle time) ÷ number of cases

Average handle time per unit of work

Average time to handle one unit of work within a case (from when that unit of work started to when it ended).

Calculation: sum(handle time for units of work) ÷ number of units of work

Notes:

Used when cases contain multiple segments (for example reopens, transfers, or multiple handling periods).

Total handle time (units of work)

Total time spent handling all units of work in the selected time range.

Calculation: sum(handle time across units of work)