Best Practices
The 90-Second Rule: What Changes When Every Agent Gets Instant Feedback
Ninety seconds after a call ends, the agent is still at their desk. The conversation is still active in their memory. Their next call has not started. This is the window where behavior change is actually possible — and most QA programs let it close without delivering a single data point.
A Different Kind of QA Program
For most of contact center history, QA has been a retrospective function. Calls are recorded. Analysts review them — eventually. Scores are tabulated. Coaching sessions are scheduled. By the time an agent hears about a call, they are three days removed from the experience, the customer is long gone, and the behavior in question has been repeated dozens of times without correction.
We have accepted this lag as a structural limitation of the model, not a solvable problem. Quarterly coaching calibrations exist not because they produce results but because they are the fastest cadence manual QA can sustain at scale.
The 90-second rule proposes something different: what if the limitation is not structural, but technological? What if the real constraint is not time — but the absence of a system that can process a call before the agent picks up their next one?
What Happens in the First 90 Seconds After a Call
Between call end and next call start, several things are simultaneously true:
The agent's working memory is active
The call is still being processed. The agent knows what happened — the customer's tone, the moment they felt stuck, the question they could not answer cleanly. This is when feedback is most associable with actual experience.
The next call has not started
There is a 30–90 second wrap-up window on most call center floors. This is not dead time — it is the only time during the shift when an agent can receive specific feedback without it competing with an active conversation.
The behavior is still pliable
Habits are formed through repetition. If an agent missed an empathy statement on call 12 and receives feedback before call 13 begins, call 13 becomes a corrective rehearsal. If feedback arrives on Thursday, calls 13–80 have already reinforced the old pattern.
The data is complete
AI scoring at this stage captures the full call, not a manager's selective memory of a 2-minute excerpt. The agent receives a scored assessment of every rubric criterion, not one or two observations the manager happened to write down.
How Instant Feedback Changes Manager Work
The assumption is that instant feedback creates more work for managers. The reality is the opposite. Here is why.
When coaching sessions are scheduled in advance and retrospective, managers must reconstruct context: pull up the call, re-listen, refresh their notes, and then re-engage an agent who may not remember the call clearly either. The session typically runs 20–30 minutes. It covers 2–3 calls. The agent leaves with corrections to apply to something they barely remember.
When scores arrive in 90 seconds, the coaching interaction changes fundamentally. A supervisor sees a low score on call 14. They walk to the agent's desk. “Call 14 — empathy on the escalation moment was low. What happened there?” The agent knows exactly what the supervisor is referencing. The conversation takes 3–4 minutes. The correction is specific. The agent applies it on call 15.
The manager's total coaching time does not increase. It redistributes from long, scheduled, low-ROI sessions to brief, timely, high-ROI interventions. Most contact center managers who switch to this model report that their total coaching time per agent per week actually decreases — while agent QA scores improve faster than they ever did in the scheduled session model.
How Instant Feedback Changes Agent Behavior
The deepest change is not in the manager workflow. It is in what agents do with the information.
When scores arrive within the same shift, agents begin to self-monitor in real time. They know feedback is coming. They know it covers their actual behavior, not a curated sample. They start to evaluate their own calls before the score arrives — developing the internal quality standard that every QA manager wants but cannot install through retrospective coaching alone.
The compounding effect
Agents who receive consistent instant feedback show a pattern that agents in delayed feedback programs do not: their improvement accelerates over time rather than plateauing. By week 6–8 of an instant feedback program, agents are self-correcting at rates that exceed what the QA system is catching — because they have internalized the standard. The QA system is no longer correcting behavior. It is confirming behavior that the agent has already begun to own.
How Instant Feedback Changes QA Program Design
If every call is scored within 90 seconds, several assumptions that drive traditional QA program design become obsolete:
Old model
Random sampling to maximize review coverage
90-second model
Every call scored; sampling used for audit and calibration only
Old model
Scheduled weekly coaching sessions
90-second model
Continuous micro-feedback with periodic deeper review sessions
Old model
Lagging performance indicators
90-second model
Real-time quality signal visible on every shift
Old model
QA analyst as primary reviewer
90-second model
QA analyst as rubric designer, calibrator, and exception handler
Old model
Coaching as retrospective correction
90-second model
Coaching as reinforcement of behavior already in progress
For the program design fundamentals that underpin a high-functioning QA operation — regardless of scoring speed — see the call center QA best practices guide. For the coaching session structure that maximizes the impact of instant scores, see the agent coaching best practices guide.
Common Questions
Why is 90 seconds the target window for post-call agent feedback?
Memory consolidation research shows that behavioral feedback is most effective when delivered while the experience is still vivid. Beyond roughly 90 seconds post-call, agents begin reinterpreting what happened, making it harder to connect a coaching note to the specific moment in the call that triggered it. Same-day review is better than next-week review — but immediate post-call feedback is the gold standard when the workflow supports it.
Does instant feedback replace the need for a formal coaching session?
No — instant feedback and periodic coaching sessions serve different purposes. Immediate feedback anchors a behavioral note to a specific call while the agent's memory is fresh. Formal coaching sessions give managers the space to review patterns across multiple calls, identify trends, and agree on longer-term development goals. Both are needed; instant feedback accelerates the value of the formal session by keeping agents aware of issues in real time.
How does delayed call feedback affect agent performance over time?
Delays longer than 24–48 hours cause agents to lose the specific context that makes feedback actionable. When a manager references a call from six days ago, the agent often cannot recall the exact moment being discussed — so coaching becomes abstract rather than behavioral. Research consistently shows that agents who receive feedback within 24 hours of a call improve 30–40% faster than those on weekly cycles.
What should a 90-second post-call coaching note include?
The most effective brief coaching notes include a specific call moment (timestamp or quote), the behavior that should change, and a concrete alternative action. Vague notes like "improve empathy" without anchoring them to a real call moment rarely change behavior. If the note is AI-generated, supervisors should personalize it with one sentence before delivery to increase agent receptivity.
The 90-Second Window Is Open Right Now
Upload a call and see a complete AI-scored QA result in under 90 seconds. No account needed. See what your agents could be receiving after every call.
Try It FreeRead: 5 Things That Happen When Agents Wait a Week for Feedback →