Right on Queue - Management's role in killing queues and backlogs
Queues and backlogs seem to be part of life. Much effort in customer facing parts of businesses is focused on measuring, minimising and managing queues. Back-office managers worry about backlogs, call centres almost all measure “grade of service” which really measures queue time and airports seem to be full of queues. With all this attention, you’d think we’d be better at managing queues, but they seem as much a feature of life today as ever. Some of our government departments even accept that their customers will queue for five minutes on average for calls and one has a target of twenty-minute wait times!
Our assumption is that queues are hated by all. Customers have better things to do than queue.
When queues occur, they stress staff who also have to deal with the adverse customer reaction. They are a source of pressure on management, so clearly no one likes them. However, budgets are tight, therefore most organizations don’t want staff sitting around idle to prevent queues. They plan their staffing levels to keep their teams busy, which means that queues can form.
Is it just a question of organizations not having enough budget to add extra staff? That is the case sometimes, but in this paper, we’ll explore
The fundamentals of queues and backlogs
The persistence of queues
Queues, scale and flexibility
Tactics to get rid of queues once they exist
Tactics to prevent backlogs and queues
We’ll show that some queues are avoidable and that management can use better tactics to prevent queues and need an array of mechanisms to get rid of them. We’ll cover a range of options for queue avoidance and removal. Some of these solutions can appear simplistic, but queues and backlogs aren’t complicated. They are just bottlenecks in supply and demand.
The Fundamentals of Queues and Backlogs
It seems a simple equation:
Supply greater than or equal to Customer Demand = No queue or backlog
Customer Demand greater than resource Supply = Queues or Backlogs can develop
Few organisations have the luxury of having workforces with lots of spare capacity sitting around waiting for customers to visit, call or claim. They try and run their resourcing models to have demand and supply matched as closely as possible, so that customers aren’t waiting too long for staff and staff in turn aren’t waiting too long for their next item of work.
Companies actually use queues to their advantage to maximise utilisation. By knowing the psychology of customer waiting tolerance and the competitive environment, companies can reduce costs by having customers line up to be served, meaning there is little to no "idle” time of the employee. An advanced science has built around these concepts often called queuing theory and deals with rostering effectively to manage the work within these tolerances. Government departments only have a general community goal to minimise waiting whilst banks, for example, work in a competitive environment and need to meet or beat customer expectations to maintain an edge.
Budget constraints or poor forecasting mean that demand exceeds supply some or all of the time. When demand exceeds supply, a queue will develop. Non-competitive organizations can choose to accept queues as a matter of course and just try and keep them under control. Objectives like answering 80% of the calls in thirty seconds, recognise that for 20% of the times, queues will be longer than that. Most customer research says that customers will accept a ‘reasonable’ queue but not long and frequent ones. Government departments like Centrelink, although it is a monopoly and needs no competitive advantage, regularly move into the spotlight due to their poor performance in managing customer wait times. They have been critiqued several times by the Auditor General for the high volume of abandoned calls (a key indicator that queues are too long) which has been as high as twenty million in a year.
Persistence of Queues and Avoidable Queues
We see many instances of backlogs not growing or shrinking but staying the same for weeks at a time. In some call centres a queue builds in the morning and then stays around all day. One complaints body had a six-week backlog of cases that persisted for over six months but stayed at the same level. What is interesting about queues and backlogs that stay the same, is that for these periods, demand and supply are in balance but the queue persists. So, if a processing team has a five-day backlog but maintains that for months at a time, that means they are processing work at the rate at which it is arriving. Similarly, a call centre that holds the queue constant at 5 customers waiting, but holds it all day has also balanced their resources effectively but will show poor results for the day. Unfortunately, every customer still experiences the queue or backlog and the queue hangs around like a bad smell.
This illustrates a key problem with queues and backlogs in that they persist. Once a queue starts in a call centre, even if the supply of staff equals the call demand, the queue won’t go away. A backlog of work can last for weeks unless supply starts to exceed demand. Only that starts to eat into the queue. The persistence of queues and backlogs means that they have to be broken.
This persistence “property of queues” is important because it means that supply and demand only have to be out of balance for a short period for a queue to develop and then stay. For example, if a call centre is undermanned for the first hour of the day and then has sufficient staff to answer arriving calls for the remainder of the day, the queue will stick around. This is what makes the balancing act so hard. An operation only needs to get behind for a short period for a queue to develop and then persist. Once a queue forms you need to reverse the demand supply equation for a period to get rid of it.
Queues, Scale and Specialisation
It’s interesting to sit in a contact centre and see some “skills” with queues while other groups are waiting for calls. A recent client had over thirty different queues handled by less than a hundred staff. This kind of model creates “micro centres” that lack scale and become hard to manage. Scale can help a customer function ride out the bumps of variable demand, staff breaks and the like.
The smaller the pool of people handling a queue, the greater the risk that it will be impacted by one off events like a long call, an absent team member or lunch breaks. In an operation of one thousand, the absence of one staff member has little impact on queues. In a branch of three staff, having one absent team member can cause queues all day.
One answer to problems of scale, is to create flexibility in work. Some operations do this by cross skilling staff across a variety of work types. Another model is to have staff work on a mixture of call taking and processing. This means that at busy times staff are diverted from one type of work to another. Another option is to formalise arrangements to borrow staff from other parts of the business or third parties such as outsourcers. A fourth option is to have some flexibility in the size of the workforce through part timers or what we call “a flexible resource pool”. We’ve covered that in a range of other white papers.
Tactics to Get Rid of Queues and Backlogs
There are three tactics to consider that can eat into a backlog or queue. You can:
Add more staff i.e. increase supply
Increase the capacity of the staff i.e. find a way for them to handle more work such as making calls shorter or processing work faster
Defer some of the demand to a time at which capacity exists
Each of these tactics is what we call a contingency. They need to be designed and planned for. Management needs to consider which tactics can be applied and then who will decide when and how to apply them. In contact centres, for example, we advocate a clear “active management role” that monitors queues and applies appropriate contingencies. It’s not much good having contingencies if they never kick in or having a monitoring function that can’t take any actions. It helps to have agreement on which thresholds will trigger which contingencies. For example, five customers in a queue might trigger plan A but twenty in queue might trigger actions B, C and D.
Adding staff:
In a typical supermarket, when a queue develops, management may add an extra checkout person to break the queue by talking staff from shelf stacking or other less urgent processes. Adding staff to reduce a queue isn’t always easy but sometimes it may only need to be temporary. In one operation they used a tactic of adding a group of temps to handle part of the call. It would have taken too long to train new staff to handle all aspects of the call. However, we’d spotted that 15% of the calls were landing in the wrong teams. With two days training, the temps learnt to act as a human switchboard that worked better than the IVR and made sure the calls were landing in the right teams. This added capacity and used these added staff to reduce some misplaced demand. It helped get queues under control for a low and temporary investment.
Another company came up with an innovative model, they used their substantial quality checking team to be available to take calls when queues jumped as a contingency model. Also, they were used regularly at 9-11am Mondays, so these ex-agents were able to keep their skills up and be deployed when a queue spike looked to be forming.
In another backlog situation, the answer turned out to be to move some staff from call taking to processing claims despite call centre queues. The claims team were in backlog which in turn drove a huge increase in calls asking, "Where's my claim?", which caused queues every day in the call centre.
The counter intuitive solution was to accept longer queues in the call centre, but divert some staff from call taking to helping the claims team. They weren’t able to process the claim, but they were able to take some simple and time consuming tasks from the claims team which meant the backlog was soon reduced. Once the backlog of claims fell away, the call demand dropped and the call queues also fell away.
Another company had an innovative approach of employing a proportion of their staff on 25 hour a week rosters, selecting a specific profile of people who preferred that working model. When demand spiked, instead of queues forming, their 25-hr week could be quickly extended to 35 hours giving the organisation a substantial capacity boost when required.
Increase capacity
Another “contingency” that can be used to increase capacity is to change the process so that the staff can handle more. For example, it may be that there are parts of the process that an organisation can remove or reduce to speed the process during periods of high demand. In one call centre, the normal process involved “reconfirming” some information held on file on every call. In times of high demand, staff were trained to skip this process as it wasn’t essential. This added capacity and enabled the centre to break queues, but it needed careful management.
Another model to increase capacity quickly is to “split” the handle time when there is processing as part of a call. During the call the agent would say something like “this change is confirmed and will be recognised on our system by 8 pm tonight” and take the details down for a group to catch up with processing later in the day. The shorter call length busted the queue and as long as they planned to finish the processing that evening, customers and company were better off.
Other tactics we have seen include, prohibiting notes after calls, bring back ex staff from the centre who are working in other divisions, and a range of tactical techniques such as “trading breaks” for an earlier shift end, quickly flexing in SME and support staff to take calls. These methods are strictly temporary to quickly bust the peak of the queue and get supply and demand in balance, but they are not to be used over more than say a few hours.
Reduce demand
Deferring some work is another useful "queue breaker” if there is capacity to do this work later. So, if a call centre knows that they are short of staff in the morning but always have surplus in the afternoon, then it may be possible to use tactics like “case managing” some work. This means taking the bare minimum of details on the morning calls to make them shorter (increasing capacity). The agents then process these items later and may call the customer back later in the day when more capacity exists. This can be a useful tactic for a temporary queue. Many organisations use tactics like advertising when they have more capacity or warning of wait times to try and get customers to defer their demand.
Preventing Backlogs and Queues
One experienced contact centre manager told us recently that he had worked out that he needed to stop queues forming in the mornings. He understood the persistence of queues and recognised that he was better off being slightly over staffed early in the day. If a queue formed late in the afternoon, it would have less time to persist. A prevention tactic is to try and anticipate when queues might form and be ready to manage those.
A second tactic is to respond really fast with contingencies. The larger a queue or backlog becomes, the more action is required to break it. Expert workforce planners can often predict likely peaks and have contingencies ready to go to prevent queues. It really helps if there is a rapid response. It can be as simple as asking staff to defer breaks or deferring some planned meetings.
Responding quickly helps stay on top of queues and stops them forming and getting too large to break. In processing areas, many organisations recognise that “same day processing”, meaning everything gets done on the day it arrives, prevents backlogs forming. This sounds easy but means that workforce planning must align staff numbers with work arrival. Some organisations use things like over-time to ensure they prevent backlogs. Others have predictable cycles such as month or year ends and plan their staffing accordingly. For example, many health insurance businesses bring in temps for their annual “rate renewal peak” when volumes are far higher than the rest of the year.
The last strategy we’ll mention is about the design of the operation. Breaking a call centre down into hundreds of skills or fracturing a processing operation into many small teams, may have the apparent benefit of specialisation, but also creates sub scale teams where queues and backlogs are harder to control. Queues can be prevented by designing an operation differently. It’s much easier to manage a contact centre with two skills than twenty. Rostering and forecasting for twenty different queues is that much harder and more likely to be error prone. So, design plays a key role in queue and backlog prevention. Technology can help by doing things like overflowing calls between skills or using complex rules and algorithms to prioritise work but it can’t overcome the problem that some staff with narrow skills are idle, while others are flat out.
Summary
In this paper, we hope we have explained that queues and backlogs are frustrating but predictable. There are many tactics to predict, prevent and manage them. We’ve learnt over the years that good design of the supply model combined with effective management can make a big difference. If you’d like to know more about these ideas, please contact us at info@limebridge.com.au or call 03 9499 3550 or 0438 652 396.
