Last updated: September 12, 2026
What schedule adherence measures
Schedule adherence is the share of scheduled time an agent spent doing what the schedule said, at the time it said. Minutes count against the number whether the agent was early or late, so a fifteen minute early login scores the same as a fifteen minute late one. Divide adherent minutes by scheduled minutes.
That early login rule is where most people meet this metric and decide it is broken. Somebody showed up a quarter hour before their shift, sat down, took calls, and the dashboard docked them for it.
The reason is that a schedule is not a request for eight hours of work. It is a forecast of how many people need to be reachable in each interval, usually fifteen or thirty minutes wide, built from the volume that interval has carried before. An early sign in puts a person in a seat the forecast never asked for.
For one agent on one Tuesday that is harmless, which is the part that makes the rule feel petty. Across forty people it becomes a staffing number that stops describing the floor.
The formula, and the minutes it counts
Adherent minutes divided by scheduled minutes, times 100. Most tools compute it from the other direction, subtracting every minute where the actual state differed from the scheduled one.
Say an agent is scheduled for 480 minutes. They sign in eleven minutes early, come back from lunch nine minutes late, and spend fourteen minutes in an aux code that nothing in the schedule accounts for. Thirty four minutes fall outside the plan, 446 of 480 stay inside it, and the day scores about 93%.
The eleven early minutes cost what the nine late ones cost, and nothing in the arithmetic knows which of the two was somebody trying.
Adherence and conformance answer different questions
Conformance measures how much of the scheduled time got worked and ignores when it happened. Adherence measures whether the time landed where the schedule put it.
Take an agent scheduled nine to five who works nine thirty to five thirty. Conformance reads 100%, since eight hours were scheduled and eight were worked. Adherence reads 87.5%, because sixty minutes sat outside the plan at the two ends of the day, leaving the morning queue short a person and the evening queue carrying one the forecast never asked for. That gap is what a service level agreement gets written to close.
Some platforms call conformance schedule attainment, and a few use attainment for hours worked against hours scheduled across a week. Check which one your tool computes before comparing two reports, since the labels move between vendors.
The tolerance window decides what the number says
Every adherence calculation carries a tolerance, a grace band around each scheduled event inside which a variance does not count.
Run one day through a five minute tolerance and then through a one minute one. Six state changes on a normal shift covers sign in, two breaks, lunch out and back, and sign out. Three minutes late on each of them is eighteen minutes. The five minute tolerance returns a clean 100%, while one minute returns about 96%. Same behavior, four points, and the only thing that moved was a setting.
A wide window is defensible. Clocks drift, a call runs past the moment a break was supposed to start, and a state change takes a second to register. What it also does is set the number, so publish the tolerance beside the percentage and treat any adherence figure that arrives without one as unreadable, which is the rule call center benchmarks applies to every published figure. Widening the window mid quarter and comparing the halves is the version of this that shows up in real reporting.
What counts as being in adherence
Being in adherence means the agent's actual state matches the scheduled one at that moment. A scheduled fifteen minute break is in adherence. The same fifteen minutes taken an hour earlier is not, even though the agent worked the identical total.
Anything the schedule put into the day counts while it happens where it was put: breaks, lunch, training, coaching sessions, team meetings, offline work such as dispatch cleanup, and after call work where the schedule carries it.
Three situations sit in the gap, and your tool has already decided each one for you. A call still live when the break starts is the common one, followed by an agent pulled into a side by side that nobody put on the schedule, and an outage that drops the floor into an unplanned aux code. Somebody has to code each of those as an exception, and where nobody does, the agent wears a variance that came out of a decision they did not make. The coding rule matters more than the threshold, and most programs have written down neither.
Where it meets occupancy and shrinkage
Shrinkage is the share of paid time the forecast sets aside for everything that is not taking calls. Adherence is the check on whether the day matched the plan that number built. Underestimate shrinkage and you publish a schedule nobody can hold, then the adherence report names eleven agents instead of naming the forecast.
The pressure runs the other way against occupancy. Push adherence up while volume stays flat and more people are reachable in the same intervals, which pulls occupancy down and looks like waste on a different dashboard. Where occupancy is already high, the same push asks for more presence inside a day with no slack, and what comes back is unscheduled breaks.
Before an adherence number becomes a conversation
Adherence turns punitive faster than anything else on a contact center dashboard. The number comes off the platform with no human judgment in it and sorts into a clean descending list of names, and the thing it measures is presence rather than work.
That ranking is worth distrusting. Over twenty shifts an agent is scheduled 9,600 minutes. Disappear for a single forty minute block on a Tuesday afternoon and you score about 99.6%. Run ninety seconds late at each of six state changes every day and you give up 180 minutes, which scores about 98.1%. The list puts the second agent underneath the first, while the forty minute hole is the one that broke service level in the interval where it landed.
Four things belong on the desk before the number becomes a conversation with somebody. The shape of the miss comes first, meaning whether it was one block or a daily drift, and which intervals absorbed it. Pull the exception log for that week next, because coaching, side by sides, and outages a supervisor scheduled arrive as the agent's variance when nobody codes them. Read the schedule after that, since a lunch assigned at the busiest interval of the day gets moved by the person living it. Then put the week's QA score and coverage rate beside it, since an agent in seat every minute and scoring 71 is a different conversation from one who runs loose on breaks and scores 92. Both of those come off the QA metrics that describe your program.
CallSniper does not produce an adherence figure. That one comes out of your workforce management or phone platform, and what we built scores the calls that happened inside those scheduled minutes. A call review scores a batch of your own recordings from the week adherence moved against your existing scorecard, so the presence number and the quality number are both on the table before anybody sits down.
Frequently asked questions
What is the schedule adherence formula?
Adherent minutes divided by scheduled minutes, multiplied by 100. A minute counts as adherent when the agent's actual state matched the scheduled activity at that moment, and every minute outside the plan reduces the result whether it fell early or late.
What is the difference between schedule adherence and conformance?
Conformance compares total time worked against total time scheduled and ignores the clock, while adherence compares the two moment by moment. An agent who works a full shift half an hour shifted posts perfect conformance and imperfect adherence.
Does being early count against schedule adherence?
Yes, on most configurations. The schedule places people in specific intervals to match forecast volume, so time in a seat the plan did not ask for is a variance the same way a late start is. Where early sign ins help, put them on the schedule.
What is a good schedule adherence percentage?
The target follows from your own settings rather than from a published figure. Your tolerance window, interval length, and exception coding rules decide what any percentage means, so a figure quoted without them describes nothing you can act on. Build the target off your own trailing baseline.
Should schedule adherence go on a QA scorecard?
Keep it off. It comes off a different system, measures presence rather than call handling, and putting it on a scorecard punishes the agent who stays on a live call past the start of a scheduled break.
Sources
- No published schedule adherence benchmark is quoted on this page, so no target percentage appears in it.
- The 480 minute worked example, the tolerance comparison, the 9,600 minute ranking example, and the supervisor checklist. Editorial judgment as of 2026-09-12.
- CallSniper scoring recordings after the call ends against the customer's existing scorecard. Trinsic feature sheet and call notes, 2026-08-27 and 2026-09-03. Verified internally.