Unanswered service emails
Count the backlog, messages arriving, messages resolved and reopened messages. Sending a reply is not necessarily resolving a customer's need.
Fictional shop. Illustrative numbers.
More repairs,
longer queue?
Stocks & Flows · System Dynamics
First, make the work visible.
20unfinished repairs
at the start
at the start
10new repairs accepted
per working day
per working day
8repairs of capacity
per working day, before
per working day, before
The queue includes waiting and in-progress repairs. A job leaves when its repair is finished—not when its owner collects the bike. Shop floor occupancy is a different quantity.
Unfinished work carries forward. New accepted work adds to it; finished repairs remove work from it. The unfinished jobs are a stock. Arrivals and completions are flows.
First new jobs arrive, then available jobs can be finished up to that day's capacity. All jobs count equally and can finish in the same day; there is no minimum repair duration.
Compare stock, level, inventory and capacity →Try a change
What happens next?
The first proposal adds one repair per day to capacity. It does not assume that one new mechanic produces one extra repair. Hiring and output need evidence outside this model.
Accepted jobs, not all customer demand. Whole numbers, 0–16.
Watch · model results
Previous run — choose a new prediction to try the changed rates.
Original rates · 10 in / 8 capacityYour run
Make a prediction, then run your proposed rates. At the original rates, the queue goes from 20 to 40 unfinished repairs.
Daily values
| Day | Accepted | Finished | Unfinished |
|---|---|---|---|
| 0 | 0 | 0 | 20 |
| 1 | 10 | 8 | 22 |
| 2 | 10 | 8 | 24 |
| 3 | 10 | 8 | 26 |
| 4 | 10 | 8 | 28 |
| 5 | 10 | 8 | 30 |
| 6 | 10 | 8 | 32 |
| 7 | 10 | 8 | 34 |
| 8 | 10 | 8 | 36 |
| 9 | 10 | 8 | 38 |
| 10 | 10 | 8 | 40 |
These are model results, not observations from a real shop. The ten-day arithmetic cannot establish fairness, profitability, quality or whether hiring was wise.
Revise
What would you revise?
Your prediction, the rates, or the model itself?
- Match accepted work to completion capacity and growth stops; the old queue remains.
- Finish more than is accepted and the queue can clear, if that capacity is real.
- Accept fewer jobs and waiting may move to customers outside this model. Where did that work go?
Doubt the model
What did we leave out?
Same-sized jobs. Constant rates. All parts available.
What we left out
- Jobs differ in effort and required skills.
- Training can absorb time before increasing output.
- Parts waits, interruptions, rework and cancellations are absent.
- Rejected demand, waiting before acceptance, costs and quality are outside the boundary.
- Completing a repair is not the same as returning a bike to its owner.
- If actual completions fall when more jobs arrive, our constant-capacity assumption needs revisiting.
Take the lens with you
Try this on your system
Every field is optional. Your notes stay in this browser tab; copy them before closing it.
The same lens, another situation
A civic permit queue
Count applications accepted and decisions completed. The same accounting does not imply the same goals, constraints or fair intervention.
Where this comes from
- Introduction to Project Dynamics · slides 28–30
James Lyneis · MIT OpenCourseWare · 2012. Stock/level and flow/rate distinctions.
- Systems thinking and practice
The Open University · OpenLearn. Perspectives and provisional boundaries inform our reflection prompts.
The bike-shop case and numerical model are original illustrative exercises from Renaissance Solutions. These institutions have not endorsed this exercise.
All sources and credits