How to Find and Fix the Operational Bottlenecks Slowing Your Business Down

Your business isn’t as fast as its quickest person. It’s as fast as its slowest step. Somewhere between a customer saying “yes” and the work getting done and paid for, there’s a step — or a sequence of steps — that’s constraining everything else. That’s your bottleneck. And until you find it and fix it, nothing else you optimise will matter much.

This isn’t theory. It’s the most practical operational concept there is. Because a bottleneck doesn’t just slow down one task — it creates a queue that backs up everything behind it. Fix the bottleneck and the entire system speeds up. Ignore it and you’ll keep wondering why hiring more people or buying better software doesn’t seem to help.

Where Bottlenecks Hide

Bottlenecks are rarely where you expect them. They don’t announce themselves. They hide behind habits, workarounds, and “that’s just how long it takes” assumptions. Here are the most common places I find them in businesses doing $2M-$20M.

Approval queues. A quote sits for two days because the owner needs to review it. A purchase order waits because the operations manager is on site. A job can’t start because someone hasn’t signed off on the scope. Every approval that requires a specific person is a potential bottleneck — because that person has a finite number of hours.

Handoffs between teams. Work moves from sales to operations. From field to office. From project management to accounts. Each handoff is a potential drop point where information gets lost, delayed, or distorted. The work itself might take 30 minutes, but the handoff adds 24-48 hours of latency.

Data entry and rework. Someone can’t do their job until someone else has entered information into a system. If the data entry person is busy with other tasks, everything downstream waits. And if the data comes through with errors, the downstream team does the work wrong and has to redo it.

Single points of knowledge. Only one person knows how to use the quoting software. Only one person can process warranty claims. Only one person understands the pricing structure for a particular product line. When that person is unavailable, the work stops.

Technology constraints. The software can only handle one user at a time. The system crashes when too many records are loaded. The integration only syncs once per day. The report takes 4 hours to generate. Sometimes the bottleneck isn’t a person — it’s a tool.

How to Measure Throughput (Simply)

You don’t need complex metrics to find your bottlenecks. You need two numbers for each step in your core processes:

Cycle time: How long does this step actually take when someone is working on it? A quote might take 20 minutes to build once someone sits down to do it.

Lead time: How long from when the step could start to when it’s actually done? That same quote might have a lead time of 3 days — because it sat in a queue for 2 days and 23 hours before someone got to it.

The gap between cycle time and lead time is your bottleneck indicator. A 20-minute task with a 3-day lead time means the step itself isn’t the problem — the waiting is. Something upstream is either overloaded or poorly prioritised.

A simple exercise: Pick your most common job type. Track a single job through your business, recording two timestamps for every step: when it became ready to be worked on, and when it was completed. You’ll see exactly where the waiting happens.

The Five-Step Bottleneck Fix

Step 1: Map the Process End-to-End

You can’t fix what you can’t see. Get your core process on a whiteboard — every step, every handoff, every decision point. Not the idealised version from the training manual. The real version, including the workarounds, the “oh and then I also have to call Dave” steps, and the things that happen differently on Fridays.

Step 2: Measure Each Step

For every step in the process, record the cycle time and lead time. Do this for 10-20 jobs to get reliable averages. You’re looking for the step where the gap between cycle time and lead time is largest — that’s your primary bottleneck.

Step 3: Understand Why

Once you’ve identified the bottleneck, ask why it exists. The answer usually falls into one of five categories:

• Capacity: The person or system handling this step simply can’t keep up with the volume
• Dependency: This step can’t happen until something else happens first, and that “something else” is unreliable
• Complexity: The step itself is unnecessarily complicated and takes longer than it should
• Variability: Sometimes this step is fast, sometimes it’s slow, and the unpredictability creates chaos
• Policy: A rule or requirement forces work through a narrow gateway that wasn’t designed for the current volume

Step 4: Fix the Constraint

The fix depends on the cause:

• Capacity bottleneck: Add capacity (another person, a faster system), reduce demand (eliminate unnecessary work that hits this step), or automate the routine portion so the constrained resource only handles exceptions.
• Dependency bottleneck: Remove the dependency if possible. If a step waits for information, can that information be collected earlier in the process? If it waits for an approval, can the approval criteria be turned into a rule?
• Complexity bottleneck: Simplify the step. Break it into smaller parts. Standardise the inputs so the step is more predictable.
• Variability bottleneck: Standardise the process so it runs the same way every time. Variability creates queues because you can’t plan around something unpredictable.
• Policy bottleneck: Change the policy. If every quote over $1,000 needs owner approval and 80% of your quotes are over $1,000, you don’t have a quality control process — you have a bottleneck disguised as one.

Step 5: Monitor and Repeat

After fixing the primary bottleneck, measure again. Your throughput should improve. A new bottleneck will emerge somewhere else. Fix that one too. This is an ongoing discipline, not a one-time project.

Automation as a Bottleneck Killer

Many of the bottlenecks I see in businesses are fundamentally caused by manual processes that can’t keep pace with business volume. The quoting step is slow because someone has to manually build every quote. The invoicing step is late because it depends on someone typing up field notes. The scheduling step is a bottleneck because one person manually juggles all the moving pieces.

Automation doesn’t just speed up these steps — it removes the capacity constraint entirely. An automated quoting system doesn’t get overwhelmed by 50 quotes in a week the way a person does. An automated invoice generation process doesn’t fall behind on Friday afternoons. An automated scheduling system can optimise across dozens of variables simultaneously.

The key insight: automate the bottleneck first. Not the easiest thing to automate. Not the most interesting thing to automate. The step that’s actually constraining your throughput. That’s where automation delivers the highest return.

Every business has a constraint. The best businesses know exactly where theirs is and they’re actively working to eliminate it. The worst businesses have bottlenecks they’ve never identified, buried under years of workarounds and “that’s just how we do things.” Find yours. Measure it. Fix it. Then find the next one.