Purchase orders pile up. Vendor emails wait for replies. Shipment updates arrive late, and managers spend their afternoons checking spreadsheets and chasing approvals. The usual first response is to hire another coordinator.
Sometimes that is exactly right. Sometimes it adds cost without removing the delay, because the problem was never the number of people. This article helps you tell the difference, and choose between hiring, outsourcing and automation based on what is actually slowing the work down.
What a bottleneck is
A bottleneck is a point in a workflow where work arrives faster than it can leave. It is not simply a busy department. Take a purchase order: it is created, checked, approved, sent to a vendor, updated in the system, matched against an invoice and tracked until delivery. Each step is simple, but a delay at any one of them spreads.
That is why the fix matters. Treat the wrong cause and the chain keeps breaking, only with a bigger team watching it happen.
Three causes, three different fixes
Most operations bottlenecks come from one or more of three causes. Each points to a different first move.
Not enough hours
The process works and is written down, error rates are low, but volume has outgrown the people available.
- Backlog grows steadily with volume
- Quality holds when people have time
- Work waits in a queue, not in confusion
Unclear or inconsistent work
People do the same task different ways. Ownership, approval rules or exception handling are unclear, so work loops back.
- High rework and corrections
- “Who handles this?” comes up often
- Only one person knows how it’s done
Information stuck between tools
The work is clear, but people spend their time retyping data, copying between systems and chasing status by email.
- Same data entered in several places
- Status lives in inboxes and spreadsheets
- Delays come from waiting for information
Diagnose your bottleneck
Answer each statement for the workflow that is slowing you down. This is a thinking aid, not a formal assessment, and it only points to where to look first.
The backlog grows when volume rises and shrinks when it falls.
The work is done correctly when people have enough time.
More than one in ten items needs correcting or redoing.
There is no written procedure someone new could follow.
It’s often unclear who owns an exception or approval.
The same data is typed into two or more systems.
Staff spend much of their time chasing status or missing information.
Inputs arrive in varied formats: PDFs, emails, portals.
Hire, outsource or automate
Each option solves a different problem. None is always cheaper or always better.
| Hire | Outsource | Automate | |
|---|---|---|---|
| Best when | Work needs judgment, local presence or close relationships, and volume is steady | Work is repeatable and documented, or volume varies, and it doesn’t need to be done on site | Steps are high-volume, rule-based and stable, with data moving between systems |
| Needs first | A clear role and someone with time to train and supervise | Written procedures, access rules, quality checks and a named contact | A stable process, clean data, clear exceptions and a person reviewing output |
| Time to effect | Recruiting, onboarding and ramp-up | Documentation and handover, then a trial period | Mapping, build and testing, then a monitored pilot |
| Watch out for | A new person inheriting an unclear process | Handing over a process nobody has written down | Automating a broken process, which makes errors faster |
The options also combine well. A common pattern is to document the process, move the repeatable part to outsourced support, automate one stable step such as capturing data from incoming documents, and keep in-house staff on exceptions, vendors and decisions.
A worked example
This is a hypothetical example with invented numbers, to show how the diagnosis changes the decision. A wholesale business processes 900 purchase orders a month. Two coordinators can give about 60 hours each a month to PO work; the rest of their time goes on other duties. Each PO takes 14 minutes, and one in four needs rework.
Hours needed = POs × minutes × (1 + rework rate) ÷ 60. Hypothetical figures for illustration only; your own times, volumes and rework rates will differ.
In this example, hiring alone does close the gap on paper, and that is worth saying plainly: hiring can help. But one in four POs is still reworked, and new staff learn an undocumented process from people who are already stretched. Fixing the process first shrinks the problem, and the remaining gap can then be closed with outsourced support or by automating one step. The right choice depends on your costs, risks and how quickly volume is growing.
When hiring is the right answer
Local or in-house hiring is often the better choice when the work:
- needs judgment, negotiation or authority to make decisions;
- depends on close, in-person relationships with customers or suppliers;
- must happen on site, such as receiving, inspection or warehouse work;
- involves information that can’t be shared outside the company;
- is a steady, long-term workload rather than a short spike.
Hiring also works well once the process is clear. The risk isn’t hiring itself; it is hiring before you know why the work is slow. If five people are working around an unclear process, adding two more usually means seven people working around it.
Questions to ask before adding headcount
Name the step, not the department.
Repetitive work can be documented and delegated; decisions often stay in-house.
If not, that is the first job, whoever ends up doing the work.
Volume, time per item and rework rate turn opinions into a decision.
If not, outsourcing is an option to compare.
If so, some of it may be a candidate for automation.
Mistakes to avoid
- Adding people to a process that no one has written down.
- Outsourcing work without procedures, quality checks and a named owner.
- Automating a process that is still changing or full of exceptions.
- Measuring success by headcount or cost alone instead of turnaround, errors and backlog.
- Leaving skilled in-house staff on routine data entry while decisions wait.