What “locked” should actually mean

Ask three people in a WFM team what “locked” means and you will get three answers. Published. Approved. Can’t be edited by team leaders. Frozen for payroll. Each of these is a reasonable use of the word and they are not the same thing, which is why the concept fails so quietly: everyone believes the schedule is locked, and everyone is describing a different lock.

The 25-point problem

Take one agent on one eight-hour shift, 09:00 to 17:00.

Published schedule: on the phones 09:00–12:00, lunch 12:00–13:00, on the phones 13:00–17:00.

What the agent did: logged in five minutes late, worked through to 12:00, took lunch 12:00–13:00 as published, worked to 17:00. One small lateness, otherwise a textbook day.

What happened at 10:30: the interval was running hot, so the intraday desk moved lunches an hour earlier to protect the 12:00–13:00 peak. The agent’s schedule now says lunch 11:00–12:00. Nobody told the agent, or they were on a call when the notification appeared.

Now score adherence, minute by minute, over the 480-minute shift:

A 25-point swing, from identical behaviour. The agent is penalised twice for one change: once for being on the phone during their new lunch, and again for being at lunch during their new phone time.

Three different locks

The confusion is fixable by naming the locks separately. There are three, and they protect different things.

Measurement lock is the one that is almost always missing, and the one that causes the 25-point problem. Most systems have a publication concept and a payroll cut-off. Very few keep a distinct, immutable baseline for scoring; they simply score against whatever the schedule currently says, which means the yardstick is edited by the thing it is meant to be measuring.

Locked does not mean rigid

The objection to all this is immediate and fair: contact centers change during the day. An intraday desk that cannot move a break is not an intraday desk. This is correct, and it is not what a measurement lock prevents.

The distinction is between the plan of record and the operating schedule. The plan of record is what was published and what adherence is measured against. The operating schedule is what the operation is doing right now. They are allowed to diverge: that divergence is intraday management, and it is precisely what the Revised column in an Original / Revised / Projected / Actual loop exists to hold.

So the desk can still move that lunch. What changes is what the move implies:

• The change is recorded against Revised, with who made it and when. Original is untouched.
• Adherence continues to be scored against Original, so the agent is not penalised for a change they did not make.
• If the agent accepts the change (explicitly, in the tool) the baseline moves with their consent, and only from the moment of acceptance.
• If they do not see or do not accept it, the operation absorbs the variance. That is the correct place for it to land.

That last point is the one that meets resistance, because it means the operation carries the cost of its own short-notice changes. That is the intended effect. A desk that can move breaks for free will move them constantly; a desk that owns the variance moves them when it matters.

Consent, and what counts as it

“The agent was notified” is not consent, and treating it as consent is how systems that are technically correct still fail on the floor. A workable standard:

• Inside the shift: changes require explicit in-tool acceptance. Unaccepted changes apply to the operating schedule but never to the measurement baseline.
• Before the shift, short notice (typically under 24–48 hours): acknowledgement required, with a defined window; unacknowledged changes do not rebase adherence.
• Beyond the notice period: the schedule can be republished and the baseline moves with it, because the agent has genuine time to plan around it.
• Agent-initiated changes (swaps, approved leave, voluntary overtime): rebase immediately. The agent asked for it.

Pick the thresholds that suit your operation and any agreements you are bound by. The part that is not negotiable is that some class of change must fail to move the baseline, or the baseline is not a baseline.

What the record has to keep

Immutability is not achieved by disabling an edit button. It is achieved by never destroying the prior version. Concretely, for any interval on any day you should be able to answer:

• What was published, and when.
• What it is now, and every change in between, each with an actor, a timestamp and a reason.
• Which changes were accepted by the agent, and when.
• Which version adherence was scored against, and why that one.

That is an append-only history, not a mutable row. It is a modest amount of engineering and it is the difference between a schedule that can be discussed and a schedule that can only be argued about. When an agent disputes an adherence score (and they will) the entire conversation should take under a minute, because the record answers it.

Symptoms you have a lock problem

• Adherence disputes that cannot be settled from the system, only from memory.
• Adherence scores that change retrospectively for a day that has already closed.
• Team leaders keeping private spreadsheets of “what really happened”. This is the clearest signal available: people build shadow records when the real record is not trusted.
• Agents screenshotting their schedules. Same signal, from the other direction.
• Nobody able to say who moved a break, or when.
• Suspiciously good aggregate adherence alongside obvious floor-level frustration.

That last one is worth sitting with. A system where the schedule quietly follows reality will report excellent adherence indefinitely, because the target moves to wherever the outcome landed. High adherence is not evidence of a healthy operation unless the baseline is immutable; without that, it is a tautology with a percentage sign on it.

The short version

Lock the measurement baseline at publication and never move it without the agent’s consent. Let the operating schedule change as much as the day requires, and record every change with an actor and a reason. Keep the two apart, and adherence becomes a number people will argue from rather than about.