Planning for Schedule Variance

Line of people dividing - variance concept

Are you confused by schedule variance, or just not quite sure what to do to restore balance when everything goes off track? You are not alone.

That’s why we’ve asked WFM expert David Appleby to share the core reasons for schedule variance so you know what to look out for, alongside mitigating actions you can take to regain some control.

Variance Is Just One Facet of a WFM Strategy

As the old saying goes, ‘Look after the pennies and the pounds will look after themselves’. I’ll not stretch the metaphor too far; however, costing is obviously a salient factor when planning your schedules.

Firstly, I strongly urge you not to Google ‘Schedule Variance’, unless, of course, you wish to lose hours in the murky swamp water of PRINCE2 theory.

At the risk of teaching my grandmother to suck eggs, the idea is a simple one. Essentially, what do I currently have for staffing vs. what should I have according to the plan/rota?

The purpose here is to have a look at the core reasons for variance, and give some ideas as to mitigating actions you can take.

So where do we start? Well, longtime readers of this site cannot help but notice the plethora of articles around WFM (workforce planning/management) strategies – this is the core of your planning cycle.

Variance is just one facet of that and, unfortunately, yet another variable that needs to be factored into your planning.

I’ll start by taking a broad strokes approach, starting with my definition.

Variance: A time loss percentage variable dependent on multiple factors, of which the majority are measurable, although out of the business’s control.

Variance itself can – and should – be measured over a range of timescales. Dependent on the quality and availability of historical data, you can extract and chart data:

  • Annually
  • Seasonally
  • Monthly
  • Weekly
  • Daily
  • Hourly
  • Planning interval (15/30 minutes etc…)

Although the last case may be a level of granularity too far.

Shrinkage and Variance May Seem to Be the Same BUT They Are Only Second Cousins

At this point, it should be noted that whilst shrinkage and variance may seem to be the same, they are only second cousins.

Shrinkage is a factorable loss of time due to the unavailability of staff for a given interval. Variance is more a movement/drift of staffing level away from the plan due to circumstances outside the control of said plan.

With a single agent, this may not be an issue. However, the problem starts to rear its head when this starts to cause a drag on remaining agents in a team or has a knock-on effect across the board.

Looking at a simple one-variable change, and at a single team only, in this case a change in AHT (Average Handle Time), an example would be:

Agent A takes a call. From a planning perspective, we know that for this type of call, the team has an AHT of 600 seconds (I’m slow, let me stick to round numbers please).

The planning team has done its job correctly with 15-minute breaks factored into the shifts. Unfortunately, just before the scheduled break, Agent A takes an outlier.

This turns into a complex call and doubles the AHT. On finishing the call and wrap about 450 seconds after the scheduled break time, and following procedures, Agent A takes their break now at the first possible opportunity.

The agent returns from break. However, they have a 30-minute team meeting scheduled immediately and are now 9 minutes late (V=450 seconds).

This now hits across the team, as all 8 members are now in variance and V has jumped, in an instant, to 3600 seconds. We’ve now got an hour variance from the plan. This tells us that, in an ideal state, 6 calls will not have agents available.

Now, what happens if a second outlier hits? Given the work of Erlang, we know that queuing never fits the ideal state and the bunching will affect the results.

Therefore, three states are possible: i) The call volume is lower than forecast, giving a Variance Severity (Vs) of 0.00; ii) Call volume matches the forecast, making Vs ~ 0–0.29, i.e. minimal impact; iii) Call volume exceeds forecast, making Vs >= 0.3.

Adherence Tracking Is Far Easier Than Variance

The question now becomes one of monitoring and/or intervention. Adherence tracking is far easier than variance, given that adherence is a known factor and, more importantly, excluding Mjölnir hitting the building, tends to affect single agents. Variance can drag the whole operation down, which…

Adherence tracking is far easier than variance, given that adherence is a known factor and, more importantly, tends to affect single agents.

…comes to the crux of the matter, and the big questions. Can you plan variance in? Plus knowing when to get the real-time management team to intervene, or letting things drift back? Real-time intervention is not, in any scenario, a desired act.

Ideally, once a day has been set up, as little adjustment as possible should take place. It causes ripples across the pond and invariably upsets someone.

From the single variable issue example above, we can see how one small issue can affect a whole team. Now mirror it across a bank’s 500-seat contact centre with a login issue for the second shift.

Not an adherence issue, although it will affect it, but rather an unplanned variance that’s going to knock on throughout the day.

Lack of agents at time x is going to lead to an increased abandon (ABA) percentage, leading to redials, leading to an increased call volume.

What Can We Do With This?

David Appleby

Isn’t that a lovely open question? I have a few suggestions.

There is a military adage that states “No plan survives contact with the enemy.” I think anyone who has spent any time in planning will agree that this is a proven fact rather than a theorem.

The first challenge is identifying and categorizing the data. The key here is to start simple. Look at your adherence.

Now look at the data again, eliminating the standard factors, and look for the areas where adherence has dropped from your x+-%.

These shifts in adherence will tend toward unusual and unplannable factors such as IT issues, outlier calls, and negative feedback loops from increased ABA rates, driving repeat calls, and the rest.

I posit a radical proposition: variance cannot be directly planned for, but it can be reacted to efficiently.

A fine line needs to be trod between the balance you already have with X factors feeding your planning on the shrinkage and adherence side, and the variance management, which, although not necessarily a ‘Black Swan’ event, may be a slightly stroppy ‘Grey Cygnet’.

So, when setting the plan and schedule, allow for software/hardware roll-outs, factor in a percentage for drift, decide at what point schedule intervention is required, and make sure the right people have the authority to act autonomously. Cygnets grow up fast!

Written by: David Appleby

If you are looking for more articles and advice on workforce planning, read these articles next:

Author: David Appleby
Reviewed by: Megan Jones

Published On: 16th May 2024 - Last modified: 21st May 2024
Read more about - Workforce Planning, , , , ,

Follow Us on LinkedIn

Recommended Articles

A picture of someone trying to catch clocks (schedule adherence concept)
How to Manage and Improve Schedule Adherence
Concept of Schedule Adherence
How do I Calculate... Schedule Adherence?
Excel Schedule Adherence Tool
Excel Schedule Adherence Tool
A picture of someone chasing a clock
17 New Ways to Improve Schedule Adherence in the Contact Centre