Time Tracking in Call Centers — What It Measures and Why
Time tracking in a call center records when agents start and stop working, how they spend their time during a shift, and how actual hours compare to scheduled hours. This data is not just an attendance record — it is the foundation for four critical operational calculations: shrinkage, adherence, labor cost, and billable utilization (for BPOs).
Without accurate time tracking data, these calculations rely on estimates — and estimated shrinkage, estimated adherence, and estimated labor cost produce schedules that are systematically wrong, staffing plans that undercount the agents needed, and cost projections that surprise leadership at quarter-end.
What time tracking measures in a call center
Time tracking captures categories of time that ACD data alone does not. The ACD records handle time (talk + hold + ACW) and available time. Time tracking records everything else — the non-call time that makes up shrinkage.
| Time category | What it includes | Tracked by ACD? | Tracked by time tracking? |
|---|---|---|---|
| Handle time | Talk time + hold time + after-call work | Yes | No (ACD is the source) |
| Available/idle time | Agent logged in, waiting for the next call | Yes | No |
| Paid breaks | Scheduled breaks (15-minute, lunch) | Partially (agent goes to break state) | Yes — records actual break start/end vs. scheduled |
| Training | Scheduled or ad hoc training sessions | No | Yes — tracks time off phones for training |
| Coaching | 1:1 coaching sessions with supervisor | No | Yes |
| Team meetings | Huddles, calibration sessions, team calls | No | Yes |
| System downtime | Time agent cannot work due to system issues | Partially (may show as logged-out) | Yes — if tracked as a distinct category |
| Personal time | Restroom, unscheduled breaks, personal calls | No | Yes — shows as non-productive logged-in time or time away |
| Late arrival / early departure | Difference between scheduled start and actual login | No | Yes — compares actual login time to schedule |
| Paid time off | Vacation, sick leave, personal days | No | Yes — recorded as full-day or partial-day absence |
Why ACD data alone is not enough
The ACD tells you what happened while the agent was on a call. Time tracking tells you what happened during all the other hours — and those other hours are where shrinkage lives.
Example: An agent is scheduled for an 8-hour shift (480 minutes).
| Time category | Minutes | Source |
|---|---|---|
| Handle time (talk + hold + ACW) | 300 | ACD |
| Available/idle time | 60 | ACD |
| Paid breaks (2 × 15 min + 30 min lunch) | 60 | Time tracking |
| Training session | 30 | Time tracking |
| Late arrival (logged in 12 minutes after scheduled start) | 12 | Time tracking |
| Personal time (unscheduled) | 18 | Time tracking |
| Total accounted | 480 | |
| Productive time (handle + available) | 360 | |
| Shrinkage (480 − 360) | 120 (25%) |
Without time tracking, the 120 minutes of shrinkage is a black box. With time tracking, you know exactly where the non-productive time went — and which components are controllable.
How time tracking data feeds workforce decisions
1. Shrinkage calculation
Shrinkage is the percentage of scheduled time that agents are not available to handle calls. It is the single most underestimated input in the staffing calculation — and inaccurate shrinkage is the most common reason operations are chronically understaffed.
How to calculate shrinkage from time tracking data:
| Step | Calculation | Example |
|---|---|---|
| 1. Total scheduled hours (all agents, 4-week period) | Sum of all scheduled shifts | 8,000 hours |
| 2. Total productive hours | Hours in handle time + available time (from ACD) | 5,600 hours |
| 3. Shrinkage | (Scheduled − productive) / scheduled | (8,000 − 5,600) / 8,000 = 30% |
Shrinkage by component (from time tracking data):
| Component | Hours (4 weeks) | % of scheduled | Controllable? |
|---|---|---|---|
| Paid breaks | 800 | 10.0% | No — required by law and policy |
| PTO / vacation | 480 | 6.0% | No — earned benefit |
| Unplanned absences | 320 | 4.0% | Partially — can reduce through attendance management |
| Training | 240 | 3.0% | Partially — can schedule during low-volume periods |
| Coaching / meetings | 160 | 2.0% | Partially — can schedule strategically |
| Late arrivals | 160 | 2.0% | Yes — address through adherence management |
| Personal / unaccounted | 240 | 3.0% | Yes — investigate if excessive |
| Total shrinkage | 2,400 | 30.0% |
The decision this drives: If the staffing plan assumes 25% shrinkage but actual shrinkage is 30%, every interval is understaffed by the equivalent of the 5-point gap. For an operation needing 30 agents on phones: at 25% shrinkage, you schedule 40. At 30% shrinkage, you need 43. Those 3 missing agents show up as service level misses every day.
2. Adherence measurement
Schedule adherence measures whether agents are doing what the schedule says they should be doing, when they should be doing it. Time tracking provides the "actual" side of the adherence calculation.
| Adherence component | What time tracking provides |
|---|---|
| Login adherence | Actual login time vs. scheduled start time. If the agent is scheduled at 8:00 AM and logs in at 8:07, that is 7 minutes of non-adherence |
| Break adherence | Actual break start/end vs. scheduled break times. If the agent takes a 15-minute break at 10:15 but was scheduled for 10:30, they are out of adherence for those 15 minutes |
| Logout adherence | Actual logout vs. scheduled end. Agents who log out 5–10 minutes early are non-adherent |
| State adherence | Whether the agent is in the correct ACD state (available, on call, break, training) at the right time. Requires integration between time tracking and ACD |
Adherence calculation:
Adherence % = (Time in correct state / total scheduled time) × 100
Target: 90%+. Below 90% indicates agents are not following the schedule — which means the schedule's coverage plan is not being executed even if it was designed correctly.
The decision this drives: If adherence is below 90%, check whether the cause is individual (specific agents consistently non-adherent) or systemic (break times are unrealistic, shift starts are impractical). Individual non-adherence is a coaching issue. Systemic non-adherence is a scheduling issue.
3. Labor cost calculation
Time tracking data is the source of truth for labor cost. Without it, labor cost calculations use scheduled hours rather than actual hours — and the difference can be significant.
| Cost component | What time tracking provides | Why it matters |
|---|---|---|
| Regular hours | Actual hours worked per agent per pay period | Confirms hours against schedule. Identifies agents working more or fewer hours than scheduled |
| Overtime hours | Hours exceeding 40/week (or daily thresholds in some states) | Overtime at 1.5x rate is a significant cost driver. Time tracking identifies agents approaching overtime thresholds so supervisors can adjust before it triggers |
| Paid non-productive time | Hours paid but not on calls — training, meetings, downtime | Reveals the true cost of shrinkage. An agent paid for 8 hours but productive for 5.6 hours has a 30% overhead |
| Absence cost | Paid absences (sick, PTO) and the overtime or staffing cost to cover them | Connects absence rates to dollar impact |
The decision this drives: If overtime exceeds 5% of total hours for 3+ consecutive weeks, the data supports a hiring business case. Time tracking shows the exact overtime hours and cost, making the comparison to hiring cost concrete rather than estimated.
4. BPO billable utilization
For BPO operations, time tracking determines how much of each agent's time is billable to the client vs. non-billable internal time.
| Time category | Billable? | Notes |
|---|---|---|
| Handle time on client calls | Yes | Core billable activity |
| Available time (waiting for client calls) | Yes (in dedicated models) | Agent is assigned to the client's account and available. In per-agent pricing, this is billable |
| Training on client processes | Depends on contract | Some contracts bill training at a reduced rate. Others make training non-billable |
| Internal meetings, coaching | No | BPO overhead |
| Bench time (not assigned to any client) | No | Direct cost to the BPO with no revenue offset |
| Cross-trained agent on a different client | Billable to the other client | Time tracking must record which client account the agent is serving in each interval |
Billable utilization = billable hours / total paid hours
Target: 85%+ for dedicated agent models. Below 80% means too much paid time is non-billable — either bench time is high, training is excessive, or internal activities are consuming too much agent time.
The decision this drives: Per-client utilization data reveals which accounts are profitable and which are not. If Client A's agents have 88% utilization and Client B's have 72%, Client B's contract may be underpriced or overstaffed relative to volume.
Operational problems that time tracking reveals
Time tracking data, analyzed over 4–8 weeks, reveals patterns that are invisible in day-to-day management.
| Pattern in the data | What it means | Action |
|---|---|---|
| Actual shrinkage is 5+ points above the planning assumption | The staffing plan is built on optimistic assumptions. The schedule is chronically short | Update the shrinkage assumption in the staffing calculation. Reschedule with correct shrinkage |
| Late arrivals concentrated on specific shifts or days | The shift start time is impractical (e.g., 6:00 AM start has high tardiness) or specific agents are habitually late | If systemic: adjust the shift start time. If individual: coach on adherence or apply attendance policy |
| Break durations consistently exceed scheduled length | Agents are extending breaks by 3–5 minutes. At 60 agents × 3 breaks × 4 extra minutes, that is 12 hours of lost coverage per day | Review whether break length is realistic. If 15 minutes is insufficient for the break room setup (distance, microwave queue, restroom line), extending to 18 minutes may be more honest than enforcing 15 |
| Training time is higher than planned | More time spent on training than the shrinkage assumption accounts for | Either reduce training time or increase the shrinkage assumption to reflect reality. Do not pretend training takes fewer hours than it does |
| Overtime concentrates on the same agents each week | A few agents consistently work overtime while others do not. This may indicate uneven schedule distribution or voluntary overtime going to the same people | Review overtime distribution for fairness. Check whether some agents are being asked because they are the only ones cross-trained or available |
| Unaccounted time (gap between logged-in hours and ACD time + time tracking categories) | Time that is neither handle time, available time, nor any tracked category | Investigate: is the ACD not capturing all states? Are agents in unofficial auxiliary codes? Is there a system integration gap between time tracking and ACD? |
Connecting time tracking to other analytics
Time tracking data is most powerful when combined with other data sources.
| Combination | What it reveals | Example |
|---|---|---|
| Time tracking + ACD data | True occupancy and productive utilization of scheduled hours | Agent is scheduled 8 hours, logged in 7.5 hours (time tracking), productive 5.8 hours (ACD). True productive utilization: 5.8 / 8 = 72.5% |
| Time tracking + forecast data | Whether staffing gaps are caused by volume exceeding forecast or by shrinkage exceeding plan | If the forecast was accurate but service level still missed, the problem is shrinkage or adherence — not volume |
| Time tracking + QA data | Whether agents who work more overtime or longer shifts show declining quality | If QA scores drop in hours 9–10 of a shift, fatigue is affecting quality — relevant for 12-hour shift schedules |
| Time tracking + attrition data | Whether schedule patterns correlate with departures | If agents who quit worked 15%+ more overtime than those who stayed, overtime is an attrition driver |
| Time tracking + cost data | Fully loaded cost per productive hour | Total labor cost / productive hours (from time tracking + ACD). This is the true unit cost of getting an agent on a call |
What to track and what not to track
Track
| Data point | Why | Cadence |
|---|---|---|
| Login and logout times | Adherence, actual hours worked, overtime calculation | Every shift |
| Break start and end times | Break adherence, actual vs. scheduled break duration | Every break |
| Time in training, coaching, meetings | Shrinkage components, non-productive time categorization | Every occurrence |
| Absence type (sick, PTO, FMLA, no-call-no-show) | Absence pattern analysis, shrinkage calculation, attendance management | Every absence |
| Client account assignment (BPO) | Billable utilization per client, accurate client billing | Every shift or when account changes mid-shift |
Do not track
| Data point | Why not |
|---|---|
| Keystroke logging | Invasive, does not correlate with call center productivity (ACD and QA data measure performance), creates mistrust |
| Bathroom break duration | Invasive and potentially illegal in some jurisdictions. If personal time is excessive, address it as a pattern through coaching — do not time individual bathroom visits |
| Screen content outside of work applications | Privacy concern. If agents are on non-work websites, address it through adherence management (they should be in available or call state, not browsing). Do not surveil screen content |
| Location tracking beyond login location | For remote/hybrid agents, confirming they log in from an approved location is reasonable. Continuous GPS tracking is not |
The review cadence for time tracking data
| Review | Frequency | What to check | Who reviews |
|---|---|---|---|
| Daily adherence check | Every shift | Late arrivals, early departures, break overruns, agents not in correct ACD state | Supervisor |
| Weekly shrinkage review | Weekly | Actual shrinkage by component vs. plan. Flag if actual exceeds plan by 3+ points | Supervisor + WFM |
| Biweekly overtime review | Every pay period | Total overtime hours, cost, distribution by agent. Flag if approaching 5% threshold | Ops manager |
| Monthly shrinkage recalculation | Monthly | Recalculate shrinkage from actual data. Update the staffing model if actual differs from plan by 3+ points | WFM + ops manager |
| Monthly utilization review (BPO) | Monthly | Billable utilization by client. Identify accounts below 80% | Ops manager + account manager |
