Real-Time Adherence: Live adherence
Objective. Drill from the floor view to an individual agent, read their current state against plan across the three layers, and decide whether a signal warrants intervention or has a legitimate explanation.
Live adherence answers 'for this specific person right now, are they on plan, working, and producing?' - the agent-level drill-down a lead uses to turn a floor signal into a targeted, fair intervention.
See it in the demo
What this screen answers
The customer question is: for a named agent at this moment, what is their state versus the schedule, and is that time turning into work at standard? A Sales person uses it to show the 'two agents look identical on a normal board but are completely different problems' moment. A CSM uses it to coach the right person for the right reason - and to model the discipline of looking at context before acting, so the tool protects fairness rather than enabling knee-jerk policing of presence.
The task
Reach this view by drilling from the Control Room or a thin interval in Coverage, or by locating the individual directly. Read the agent across all three layers in order: are they in the planned state, is the scheduled time being spent on work, and is the work meeting the output standard. Then check the timeline and any exception attached to them before forming a view. Exact agent-card fields, state labels and drill controls are SME-validation-pending - drive at the level of 'open the agent and read the three layers', not invented UI.
How to read it
The classic misread is treating an off-plan or below-output agent as a performance verdict. A red schedule state can be a legitimate reassignment to a hotter queue; a low work-adherence reading can mean the agent is working in a system that is off-agent rather than idle; a below-standard output read can reflect harder work items, not lower effort. Read all three layers together and pull the surrounding context before concluding. The screen tells you who to talk to, not what to say to them.
Honest limits
This view proves the agent-level model on seeded demo data. In production the work layer depends on activity capture coverage for that person and the output layer on a trusted standard - if either is thin for a real account, read the lower two layers as directional, not definitive. Agent-level data also carries the surveillance concern directly, so capture policy, access scope and employee communication should already be agreed (see Employee Monitoring). Confirm data freshness and sources with product; do not quote them as fact.
Interpretation check
An individual being off-plan or below output is a prompt to look with context, not a verdict on the person. Reassignment, off-agent work and harder work items all explain a red signal, so read all three layers and the surrounding context before intervening.
In the field
OVO (contact centre): the effort-to-outcome POC is precisely this view's thesis - an agent can be present and active yet not producing the outcome, and the fair response is to look at the work, not to police the login.
Practical task
On the demo, open one agent who looks fine on schedule adherence, and write a 4-line note walking their three layers, the one context check you would make before acting, and the single intervention plus how you would frame it to the agent fairly.
Self-check
An agent is schedule-adherent and active but reads below the output standard. What does that tell you, and what does it not?
It tells you presence and activity are not the issue, so the question is about the work itself - complexity, blockers or a weak standard. It does not prove low effort; confirm the standard is trusted before treating it as underperformance.
A lead wants to write up an agent showing a red schedule state. What do you ask first?
Whether the agent was reassigned to another queue or task. A red state can be the correct call under pressure. Check context and the timeline before any write-up.
Why must capture policy and employee communication already be settled before using this view in production?
Agent-level live adherence is the sharpest edge of the surveillance concern. Scoped access, agreed capture policy and honest employee communication are what keep it a fair floor-management tool rather than monitoring that erodes trust.
Verified vs demo. Live adherence is live in the demo at /#/agents on seeded data. Exact agent-card fields, state labels, drill controls, figures and the per-agent capture/standard dependencies are SME-validation-pending; confirm with product.