> ## Content Index
> Fetch the complete content index at: https://www.cognativ.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Bottleneck Analysis Template: Capacity, Waiting and a Worked Example
- URL: https://www.cognativ.com/blog/top-bottleneck-analysis-template-for-efficient-process-improvement/
- Published: 2026-10-25T19:44:57.000Z
- Updated: 2026-10-25T19:44:57.000Z
- Description: Bottleneck analysis identifies the stage that limits a process under its current demand, workload and operating conditions. A long queue is a signal to investigate, not proof that the busiest-looking team is the constraint.
- Author: Kevin Anderson
- Tags: #Import 2026-10-25 13:20

Bottleneck analysis identifies the stage that limits a process under its current demand, workload and operating conditions. A long queue is a signal to investigate, not proof that the busiest-looking team is the constraint.

[Download the blank capacity worksheet (CSV)](https://www.cognativ.com/content/files/2026/10/cognativ-582-e762b3d2901d-bottleneck-blank.csv) and [the filled example (CSV)](https://www.cognativ.com/content/files/2026/10/cognativ-582-6318f0e4884e-bottleneck-example.csv). The example uses invented, deliberately simple data so the calculations can be checked. It is not a reported consulting outcome.

## Worked Example: A Review Queue Growing by 20 Cases a Day

A back-office team receives 80 comparable cases per business day. Each case goes through intake, review and release. Each stage has 240 productive minutes available daily, after breaks and other assigned work. Assume one pass through each stage, no batching, no rework and sufficient incoming work.

__Illustrative single-pass capacity calculation__
| Stage   | Available minutes/day | Minutes/case | Capacity cases/day | Demand/capacity |
| ------- | --------------------- | ------------ | ------------------ | --------------- |
| Intake  | 240                   | 2            | 120                | 66.7%           |
| Review  | 240                   | 4            | 60                 | 133.3%          |
| Release | 240                   | 2            | 120                | 66.7%           |

Capacity equals available productive minutes divided by active minutes per case. Review can handle 60 cases against demand of 80, leaving a shortfall of 20 per day. Under these assumptions, the review queue grows by 100 cases over five days. The 133.3% figure is required load relative to capacity, not observed employee utilization above 100%.

The nominal process ceiling is 60 completed cases per day because every case needs review. Actual throughput can be lower because of interruptions, missing inputs or downstream blockage. Increasing intake capacity to 160 does not remove the review constraint; it can deliver work to the queue faster.

![The worked example separates incoming demand, waiting work and completion capacity.](https://storage.ghost.io/c/cc/dc/ccdca949-0fe4-4119-8990-70a44ad75060/content/images/2026/10/cognativ-582-ae24f77908f1-brand-v2-batch-worked-napkin.png)

The worked example separates incoming demand, waiting work and completion capacity.

## Collect Timing Data That Distinguishes Work from Waiting

For each case, capture arrival, [work-start and completion timestamps](https://www.cognativ.com/blog/process-mapping-template-for-streamlined-business-workflows/) at each stage. Record the case category, priority, rework loop and exclusion reason. If the system only records creation and final closure, it cannot reveal which intermediate queue consumed the time.

Keep units consistent. Active handling time in minutes is not the same as elapsed business days. A reviewer may spend four minutes on a case that waits two days for a missing attachment. Adding staff without fixing the attachment failure could leave the dominant delay intact.

The [Kanban Guide](https://kanbanguides.org/the-kanban-guide?ref=cognativ.com) distinguishes work in progress, throughput, work-item age and cycle time. Adopt explicit start and finish events for those measures. Do not compare one team's time from assignment with another team's time from customer submission and call the difference a productivity gap.

- **Demand:** incoming eligible cases per time period, including bursts.
- **Capacity:** productive time and staffing actually available to the measured work.
- **Waiting:** time between readiness and work starting, with a reason where known.
- **Rework:** repeated visits and corrections, not just final accepted completions.
- **Open work:** unfinished cases at the cutoff, so stalled work does not disappear from the report.

![Distinct timestamps separate processing time from waiting and rework.](https://storage.ghost.io/c/cc/dc/ccdca949-0fe4-4119-8990-70a44ad75060/content/images/2026/10/cognativ-582-2cdff3a5a625-brand-v2-batch-inputs-napkin.png)

Distinct timestamps separate processing time from waiting and rework.

## Test Alternative Explanations Before Calling a Bottleneck

Observe several representative operating periods. A temporary absence can create a queue without establishing a persistent capacity constraint. Conversely, a stage with an empty queue can still limit output if upstream work is released in batches or the stage repeatedly waits for approval.

__Evidence that changes the diagnosis__
| Observation                              | Possible explanation                              | What to inspect                                            |
| ---------------------------------------- | ------------------------------------------------- | ---------------------------------------------------------- |
| Large review queue, normal handling time | Insufficient review capacity or bursty arrivals.  | Arrival pattern, staffing schedule and queue age.          |
| Long elapsed time, little active work    | Missing input, batch release or delayed decision. | Ready-to-work timestamps and exception reasons.            |
| High completion count, growing rework    | Output is being counted before acceptance.        | First-pass acceptance and reopened cases.                  |
| One case type dominates the delay        | Different work content or specialist dependency.  | Segmented times and routing, not only the overall average. |

A mean handling time is useful for an initial capacity estimate but can hide a long tail. Review the distribution and difficult cases. If a small number of complex cases require a specialist, record that dependency instead of treating all cases as interchangeable units.

![Test alternative explanations before declaring a bottleneck.](https://storage.ghost.io/c/cc/dc/ccdca949-0fe4-4119-8990-70a44ad75060/content/images/2026/10/cognativ-582-53851a696390-brand-v2-batch-test-napkin.png)

Test alternative explanations before declaring a bottleneck.

## Choose an Intervention and Predict Its Effect

In the example, reducing review handling time from four to three minutes would raise nominal review capacity from 60 to 80 cases per day. This exactly matches average demand, with no spare capacity for variability. It is a hypothesis worth testing, not a stable-service guarantee.

Possible interventions address different causes: validate inputs earlier, remove a redundant check with the control owner's approval, schedule review time, train a backup reviewer or change batch frequency. Select the intervention that addresses the observed mechanism. Automation is not automatically preferable to removing unnecessary work.

State a quality guardrail. A faster review that misses required checks is not improvement. Record first-pass acceptance, corrections and downstream exceptions alongside throughput. Preserve the prior process as a fallback for cases the pilot cannot handle.

## Recheck the Whole Flow After the Change

Compare demand, case mix, available minutes and completed work over equivalent periods. If demand falls from 80 to 50 while the queue clears, do not attribute the whole improvement to the intervention. Keep a dated change log and explain major differences.

Recalculate every stage. Review may cease to be limiting, and release may become constrained by a different schedule. A successful local intervention does not end analysis; it changes which condition deserves attention next.

The blank worksheet intentionally contains no universal “healthy utilization” threshold. Its purpose is to make assumptions visible. Add observed throughput, queue age and quality results beside the capacity estimate, then record a decision: continue, revise or investigate another constraint. A presentation chart can summarize that evidence, but cannot replace it.

![A local improvement must be checked against the whole flow.](https://storage.ghost.io/c/cc/dc/ccdca949-0fe4-4119-8990-70a44ad75060/content/images/2026/10/cognativ-582-df6d7acc344b-brand-v2-batch-recheck-napkin.png)

A local improvement must be checked against the whole flow.