Few metrics start more arguments on a contact centre floor than adherence. A team leader pulls up a report showing an agent at 78% and treats it as a performance issue. The agent insists they worked their full shift. Both can be right, because they are usually talking about two different numbers: adherence and conformance. Real-time analysts (RTAs) live in this gap every day, and getting the definitions and the math right is what turns adherence from a blame exercise into a genuinely useful control metric.
Adherence, conformance, and compliance are not the same thing
Adherence measures whether an agent was doing the specific activity they were scheduled for, minute by minute: on a call when calls were scheduled, on break when a break was scheduled, in training when training was scheduled. It is a timing metric.
Conformance measures whether an agent worked the total number of minutes they were scheduled for, regardless of when. An agent who takes their lunch an hour later than planned but still logs the full number of scheduled minutes has low adherence for that hour, but high conformance for the day.
Compliance is a term some operations use interchangeably with the two above, but in a lot of contact centres it actually refers to something else entirely, regulatory or policy conformance (PCI DSS handling, script adherence, call recording consent), not schedule timing at all. Worth clarifying in your own operation so a QA compliance score and a WFM adherence score never get quoted as if they were the same thing.
Of the three, adherence is the term used most often in North American contact centre operations, and it is the one most WFM systems report by default.
The schedule adherence formula, worked through
The standard schedule adherence formula is:
Adherence % = (Scheduled Time - Non-Adherent Time) / Scheduled Time x 100
Worked example: an agent has an 8-hour shift, 480 scheduled minutes. During the day they log in 6 minutes late, extend one break by 9 minutes, and take an unplanned 10-minute step-away not on the schedule.
Non-adherent time = 6 + 9 + 10 = 25 minutes.
Adherence = (480 - 25) / 480 x 100 = 94.8%
That one agent's number is straightforward. Team or centre-level adherence is the same calculation aggregated: sum every agent's adherent minutes, divide by the sum of every agent's scheduled minutes, multiply by 100. A team of 20 agents each contributing their own scheduled and non-adherent minutes rolls up into one adherence percentage for the shift, the day, or the week.
What a realistic target actually looks like
Most contact centres set a target between 85% and 95%, not 100%. That is not a low bar being tolerated, it is a deliberate acknowledgment that a schedule is a plan, not a guarantee, and that some deviation is legitimate. Chasing 100% adherence usually backfires: agents start gaming the clock instead of handling the actual work in front of them, and coaching conversations turn into policing rather than problem-solving.
Exceptions matter as much as the formula
Not every deviation is the agent's fault. A system outage, a mandatory all-hands announcement, or a supervisor pulling someone into an unplanned coaching session all show up as non-adherent time if you calculate blindly. Good WFM practice is to flag these as adherence exceptions and exclude them from the score, so the number that reaches a team leader reflects genuine agent behaviour, not operational noise the agent had no control over. A raw adherence report with no exception handling will always understate how well a team is actually performing.
Why adherence feeds straight back into service level
Staffing plans built with Erlang C math assume the agents you scheduled are actually on the floor doing what they were scheduled to do. When adherence drops, the real available headcount in an interval is lower than the number your staffing calculation assumed, and service level slips even though the schedule on paper looked fine. This is exactly why real-time management exists as a discipline: someone has to watch adherence live and react before it erodes service level for the rest of the day. If you want to see how staffing requirements are actually calculated from volume and AHT, the [Erlang C Staffing Calculator](/tools/erlang-c-calculator) on this site walks through the same math RTAs are protecting when they chase down adherence gaps.
Four things to actually do about it this week
- Separate the exceptions before you coach. Pull the raw non-adherent minutes and remove anything caused by outages, mandatory meetings, or supervisor-initiated pulls before deciding who needs a conversation.
- Look at patterns, not single days. One rough shift is noise. The same agent drifting 10 minutes late from break three days running is a pattern worth a real conversation.
- Check whether the schedule itself is realistic. Chronic adherence problems across a whole team, not just one person, are often a sign the [shift and break design](/tools/shift-scheduler) doesn't match how the work actually flows, not a discipline issue.
- Coach on the specific interval, not the daily average. "You were at 82% today" is hard to act on. "You were 12 minutes late back from your 2pm break" is something an agent can actually fix tomorrow.
Adherence will never be a perfect number, and it isn't supposed to be. Used well, with exceptions properly excluded and targets set at a realistic level, it is one of the fastest ways to catch a staffing problem before it becomes a service level problem. Used badly, as a bare percentage with no context, it just becomes something agents learn to argue about instead of something that actually helps run the floor.