Why your business backlog keeps growing, even when everyone is busy.

If you run an established or growing business and the queue keeps getting longer despite overtime, recruitment or repeated process changes, this article is for you.

It explains how to identify operational bottlenecks, distinguish a genuine capacity shortage from avoidable work, and decide where to investigate before committing more money. That is the starting point for DeaconOps’s operational improvement work: establish what is constraining performance, then choose an intervention that fits the evidence.

Everyone is busy. The inbox is full. There’s a meeting about the backlog, followed by another meeting to find out why nobody has had time to reduce it.

Meanwhile, customers are still waiting.

A growing backlog is easy to describe: work is arriving faster than it is being completed. Explaining why takes a closer look. Sometimes there genuinely isn’t enough capacity. Sometimes the business has plenty of effort available, but a surprising amount of it is being spent creating, moving or repairing work.

Before asking people to do more, I’d want to understand what keeps preventing them from finishing.

1. How operational bottlenecks make backlogs grow

Imagine a café with three people taking orders and one person making coffee.

The queue is growing, so the owner tells the people at the till to take orders faster. They do. Sales look encouraging. The person making coffee is now surrounded by tickets and considering a career in gardening.

Taking another order hasn’t helped anyone get their drink sooner.

The same thing happens in offices. A team can prepare 120 cases a week, but the specialist who approves them can only complete 80. Increasing preparation to 140 produces a bigger pile of cases waiting for approval. It doesn’t produce more completed cases.

This is the central insight of Eliyahu Goldratt’s Theory of Constraints: the performance of the whole operation is limited by its current constraint. Improving activity elsewhere may achieve very little unless it helps work pass through that constraint.

That changes where I’d look first. Is the specialist actually spending their time making decisions? Or are they searching for documents, correcting submissions, answering progress enquiries and explaining the same requirements repeatedly?

Those supporting tasks may be necessary, but they don’t necessarily require that person’s expertise. Better preparation, clearer acceptance criteria and appropriately trained support can release scarce decision-making capacity before recruitment becomes the answer.

Then there’s the rate at which work enters the process. Upstream teams should supply the constraint reliably without burying it. In Goldratt’s terms, the wider system must be subordinated to the constraint: individual departments may need to change their priorities to improve overall completion.

That can be uncomfortable when a manager is rewarded for how much their team passes on. Their numbers might look excellent while the customer’s experience gets worse.

For DeaconOps, the diagnostic question is where completed work is being limited. I’d compare arrivals, completions and queue age at each stage, then examine how the scarce expertise is used. That helps establish whether you need more capacity, better preparation or a change in how work is released.

2. How failure demand adds avoidable work to your backlog

You order a sofa. The delivery date passes. You call for an update. Nobody knows, so you email. Someone replies asking for the order number you already provided. You call again.

The company has now handled several customer interactions. You still don’t have a sofa.

A dashboard might describe that as increased demand. From your side, it’s a fairly elaborate way of failing to deliver furniture.

John Seddon’s concept of failure demand gives this a useful name. It is demand arising because a service failed to do something, or failed to do it properly for the customer.

In practice, it includes chasing overdue responses, clarifying confusing instructions and reopening issues that should already have been resolved. These contacts consume capacity without representing a genuinely new customer need.

This distinction matters when assessing workload. If avoidable repeat contact is mixed into total demand, the business can end up planning additional resources to accommodate its own service failures.

There is also a feedback loop. Delays cause chasing; chasing takes time away from completion; reduced completion creates further delays. The backlog starts generating some of the work that keeps it growing.

I’d examine the reason behind repeat contact rather than classify every second conversation as failure demand. Customers may have new information or need a legitimate follow-up. The question is whether a better original service would have prevented that particular contact.

Answering that can point to a very different intervention from simply putting another person on the phone.

Before recommending recruitment, I’d sample the work arriving and trace why it exists. If a material share comes from chasing or correction, the next question is which service failure keeps generating it, and what that costs in time and customer effort.

3. What toil is, and how repetitive work consumes capacity

Someone downloads a spreadsheet, copies information into another spreadsheet, checks it against a system and emails the result to somebody who puts it into a third spreadsheet.

Tomorrow, they’ll do it again.

If the business doubles in size, they may need a colleague to help. Eventually, there could be an entire department ensuring the spreadsheets remain on speaking terms.

This is where the idea of toil becomes useful.

In Google’s Site Reliability Engineering literature, toil has characteristics such as repetition, manual effort, potential for automation and little enduring improvement. It often grows as the service grows. Applied to business operations, it helps identify work that repeatedly keeps the current process functioning without making that process better.

Examples might include copying the same information between systems, maintaining duplicate trackers, manually producing an unchanged status report or repeatedly correcting the same data mismatch.

The important distinction is the lasting effect. Fixing a mismatch every morning restores today’s position. Correcting the underlying integration can prevent the mismatch from returning.

That doesn’t make the person doing the daily reconciliation unproductive. They may be protecting the business from a problem that nobody has given them the time or authority to remove.

Nor does repetitive work automatically lack value. A control involving meaningful judgement may be essential. Customer conversations and investigations can look similar from a distance while requiring considerable expertise.

I’d assess the purpose, frequency, variation and consequences of each task. Then I’d consider whether the demand can be removed, the process simplified or the information connected more effectively. Automation follows where the benefit justifies building and maintaining it.

Reducing toil matters because it releases capacity repeatedly. It also gives people room to improve the operation. If every available hour is committed to keeping today running, tomorrow inherits exactly the same problems.

In a diagnostic, I’d examine the workarounds alongside the formal process. A recurring task needs a purpose, an owner and an explanation of what would happen if it stopped. That prevents an efficiency exercise from accidentally removing a useful control, while revealing which burdens can be reduced at source.

4. How work in progress limits help teams finish

You decide to tidy the house. You empty the kitchen cupboard, pull everything out of the wardrobe and start sorting the garage.

By lunchtime, you’ve worked extremely hard and made three rooms worse.

Opening another cupboard probably isn’t the breakthrough you need.

An operation can behave in much the same way. Everybody has plenty started, nobody has quite enough finished, and each item needs another conversation before it can move.

Work in progress limits place a deliberate boundary around the amount of work that has entered a stage but has not yet been completed. When the limit is reached, attention turns to finishing, resolving dependencies or removing an obstruction before more work is pulled in.

This makes blocked work harder to ignore. Without a limit, the easy response to a difficult case is often to pick up another one. The difficult case remains unfinished, and the number of competing commitments grows.

A limit needs to reflect actual conditions: the skills available, the variability of the work and the dependencies between stages. Set it, observe the effect and adjust it. An arbitrary number imposed without understanding the process can create a different problem.

Waiting work still counts as work in progress once it has entered that stage. Otherwise, a team can appear to respect its limit simply by relabelling everything difficult as “pending”.

Incoming demand must also remain visible. Limiting what is actively handled does not remove the obligation to record requests, manage waiting times or respond to urgent needs. Monitor the age of the whole queue, not just the items someone happens to have open.

5. How task switching and interruptions reduce effective capacity

You’re halfway through something that requires concentration.

“Quick question.”

You answer it, find your place and start again.

“Have you got two minutes?”

By the end of the afternoon, you’ve helped several people and completed very little of your own work. Everyone else’s two minutes appear to have acquired your entire day.

Task switching has a cost beyond the interruption itself. Cognitive research, including work by Joshua Rubinstein, David Meyer and Jeffrey Evans, describes the effort involved in changing goals and activating the rules for a different task.

In an operation, that can mean reconstructing a customer’s circumstances, remembering which checks have been completed or working out why a particular decision was being considered. Complex and unfamiliar tasks are especially vulnerable to this disruption.

The result is that available working time and usable completion capacity are not interchangeable. A diary containing several free hours may offer very little sustained attention.

I’d look for patterns. Who is repeatedly interrupted? What information are people seeking? Why can’t they obtain it elsewhere? Are routine status requests reaching the same specialist whose work already determines the pace of the operation?

Shared information, clearer decision rights and a reliable escalation route can reduce unnecessary switching. Protected working periods can help too, provided colleagues still have a sensible way to obtain urgent support.

The objective is to make interruptions deliberate. If every request immediately displaces the current priority, the business pays repeatedly for work being put down and picked up again.

6. How to keep the effort proportionate to the task

Someone asks for a quick update. It becomes a presentation. The presentation needs charts. The charts need commentary. Someone suggests changing the colours.

Three days later, the original decision could still have been made from a short email.

Most businesses have work that becomes more elaborate simply because nobody agreed where to stop.

Ryan Singer’s Shape Up approach uses the idea of an “appetite”: how much time a problem is worth investing in before committing to the solution. That is a different question from estimating how long a proposed solution will take.

An estimate considers the effort required by the current approach. An appetite asks whether that approach is proportionate to the outcome.

For discretionary work, I’d establish the decision it supports, the likely value, the consequences of getting it wrong and the minimum adequate result. A routine update may warrant a concise explanation. A consequential customer decision may require substantial evidence and review.

This is where professional judgement matters. Reducing unnecessary scope must preserve the quality, controls and obligations the work genuinely requires.

It also means recognising diminishing returns. Another review, another chart or another round of refinement may add very little while delaying something more valuable.

Agreeing what “enough” looks like gives people permission to finish. Without that boundary, conscientious teams can spend considerable time improving a piece of work beyond what anyone needs.

7. How stopping defects prevents rework and further delays

If a printer starts producing every page upside down, continuing for another thousand copies is rarely described as commitment.

Yet businesses sometimes do the operational equivalent.

A form repeatedly arrives without essential information. The next team fills in the gaps. Incorrect data moves through a process, and somebody further along quietly repairs it. Everyone keeps working because stopping would affect the daily numbers.

The numbers survive. The problem gets a permanent job.

Toyota’s principle of jidoka involves detecting abnormalities and stopping the affected process so defects do not continue unchecked. Its relevance extends beyond manufacturing: a service operation also needs a way to recognise when work should not proceed in its current condition.

That requires a clear standard. People need to know what a complete handover looks like, what information is essential and which exceptions require intervention. Without that, “abnormal” becomes a matter of personal interpretation.

The response should be proportionate. Contain the affected work, make the issue visible, assign responsibility and agree how service will continue. Pausing one defective step may be sufficient; shutting down the entire operation usually isn’t necessary.

Crucially, detection must lead to investigation. Repairing individual cases may protect customers today, but a recurring defect needs its cause addressed if tomorrow’s correction queue is to shrink.

People also need confidence that raising a problem will be supported. If they are penalised for interrupting output, they have a strong incentive to pass the defect along. The operation loses the warning precisely when it needs it.

8. How to measure whether your backlog is actually reducing

Moving a case from one team’s queue into another can make the first team’s figures look better.

The customer is unlikely to celebrate.

They’re waiting for a decision, a delivery or a resolved problem. They don’t particularly care which department currently owns the delay.

That is the distinction I’d keep returning to when measuring improvement.

Track completed outcomes alongside new demand, failure demand, rework and the age of unfinished items. Look at time spent waiting between stages as well as the effort spent processing them. A case that takes twenty minutes to handle can still spend three weeks in the business.

Then connect those measures to the constraint. Is its effective capacity improving? Are submissions more complete? Is less time being lost to avoidable interruptions and recurring manual work? Is the oldest work moving?

These questions help separate genuine improvement from a redistribution of the backlog. They also provide a firmer basis for investment.

Sometimes the evidence will support more people, different expertise or additional specialist capacity. Sometimes it will show that the operation is spending too much of its existing capacity on preventable work. Both can be true at once.

When an independent operational diagnostic is useful

You don’t need a consultant to recognise that people are waiting. The difficult part is establishing why, particularly when the problem crosses several teams.

An independent review can be useful when managers have different explanations, earlier fixes have moved the pressure elsewhere, or the business is considering recruitment, automation or a new system without a clear account of the underlying problem. It can also help when the people who understand the operation are too occupied keeping it running to investigate across departments.

This is where DeaconOps’s Operational Diagnostic fits. We speak with your teams, review available data and follow how work actually moves through systems, decisions and handovers. The purpose is to connect the visible symptoms to the operational conditions creating them.

You receive a practical report explaining the findings, the evidence behind them and their impact on customers, teams and performance. It includes prioritised recommendations and a sequenced action plan, discussed in a findings session.

A standard Diagnostic starts from £2,250 + VAT and includes three consultancy days delivered over two weeks. Larger or more complex operations are scoped and priced before work begins. The review establishes what needs to change; implementation can then be scoped separately through Improvement Projects.

If the backlog is growing and you’re weighing up your next move, start a conversation about what is happening. Bring the symptoms you can see. We can discuss whether a Diagnostic is the appropriate next step.

Questions business owners ask about growing backlogs

Why does our backlog keep growing even when everyone is busy?

A backlog grows when work enters a defined queue faster than it leaves. Busy teams may still be constrained by specialist capacity, repeated corrections, avoidable customer contact, interruptions or excessive unfinished work. Examine completed outcomes and queue age alongside activity levels to understand which causes apply.

Should we recruit more people to clear the backlog?

Recruitment can help when evidence shows a genuine capacity or skills gap. First identify the stage limiting completion and how its capacity is used. Extra resources elsewhere may increase unfinished work without improving the rate at which customers receive an outcome.

What does an Operational Diagnostic provide?

DeaconOps’s Operational Diagnostic provides a practical report setting out the findings, supporting evidence, operational impact, prioritised recommendations and a sequenced action plan, followed by a findings session. It helps establish what needs to change before you commit to implementation.

---

About the author

Dean Cain is the founder of DeaconOps, an independent operations consultancy supporting established and growing businesses across the UK. He has more than 15 years’ experience across operational delivery, leadership and regulated environments, with experience in performance management, process improvement, technology and change. Read about Dean and the DeaconOps approach.