All analyses
AN-2026-03Sample analysis · synthetic data

The analysis where we would have advised doing nothing

A distributor wanted to automate complaint handling. The measurement shows the money would have been burnt: the bottleneck sits outside the company.

In short

The bottleneck sits at the supplier, outside the company. Automation for €18,000 would change nothing — it would even make the bottleneck worse.

How long it takes

12.8working days

In 8 of 10 cases between 4.1 and 34.0 days.

Where it sticks

Klärung mit Lieferant

In 74 of 100 simulated runs this step was the hold-up.

What it costs

€61,000

Per year. Estimated between €47,000 and €79,000.

What the fix would return

€4,000

Per year — against €18,000 of one-off cost. It does not pay off.

01Starting point

Complaint handling

Industry
Technical wholesale
Size
120 Mitarbeitende
Cases per year
1,900
Fully loaded rate
€62 / h
Simulated runs
500
Date
May 2026

Complaints take a median of almost 13 working days. Customer service is frustrated, and so are customers. On the table: a vendor quote for workflow automation at roughly €18,000 plus ongoing licence.

Before signing, the process is modelled: five steps, four roles, 1,900 cases a year, €62 fully loaded rate.

02The model

Six steps, four roles, one measurable path

The process modelled in FlowVisual. ↯ marks a media break — the point where data is retyped from one system into the next.

Every value in the model, as a table.
StepRoleSystemDuration P10–P90Bottleneck
01Reklamation erfassenKundenserviceTicketsystem820 min4 %
02Interne PrüfungQualitätssicherungERP2060 min13 %
03Klärung mit LieferantBottleneckEinkaufE-Mail (extern)1545 min74 %
04EntscheidungQualitätssicherungERP1030 min5 %
05Gutschrift / ErsatzBuchhaltungERP1025 min4 %
03The measurement

500 runs, one clear answer

Every chart below shows before against after. Values are labelled directly — the colour is a second signal, never the only one.

Abb. 1Probability of each step being the bottleneck in a run. Before against after.
Abb. 2Lead time as a P10–P90 range. The light tick marks the median (P50).
Abb. 3Utilisation per role. Everything right of the red line is structural overload.
Throughput

38

Fälle/Woche

Days over capacity

17 %

Range P10–P90

4.1–34.0working days

04The finding

Klärung mit Lieferant — 74 %

In 74 % of runs the bottleneck is clarification with the supplier. That step costs 15 to 45 minutes of work — and then waits an average of seven days for a reply nobody in the company controls.

All four internal steps together account for under 20 % of lead time. No role is overloaded; purchasing sits at 59 %.

Complaints cost the company around €61,000 a year. But the cost driver is waiting on the supplier, not internal handling.

05The intervention

The simulated automation scenario

Instead of buying and hoping, the proposed automation was run through the model: portal intake, rule-based pre-check, automatic credit note.

All internal processing times were set as optimistically as the vendor promises — the best case.

06The re-measurement

Median lead time down 7 %

How long it takes
12.811.9 working days
Throughput
3840
Days over capacity
17 % → 14 %
Saving per year
€4,000

Even in the best case, median lead time drops only from 12.8 to 11.9 days. That is 7 % — for €18,000 of implementation cost. The labour saving is about €4,000 a year. Payback would sit beyond four years.

The reason is already in the model: you cannot automate away a step that happens outside your own company. The supplier step's bottleneck probability actually rises from 74 % to 82 % — the bottleneck sharpens, it does not weaken.

The recommendation would therefore be: shelve the automation. Instead, agree a response deadline with the three suppliers causing 80 % of complaints. That would cost nothing but three conversations — and it attacks exactly where the time is lost.

This case is in the archive because it shows the uncomfortable half of the truth: sometimes the result of a measurement is that you do not need to buy anything.

07Limits of this analysis

What this analysis cannot tell you

Every measurement has limits. A measurement that hides them is advertising.

  1. 01

    This is a sample analysis. Process and figures are constructed, not collected at a client.

  2. 02

    Supplier response time is modelled as a long-tailed distribution. That is plausible but not evidenced without real log data.

  3. 03

    The automation scenario was deliberately calculated optimistically. A pessimistic assumption would look worse still — the conclusion does not flip.

  4. 04

    Soft effects such as customer satisfaction or staff relief are not priced in euros. They may still influence the decision.

Your process will look different.

This analysis is a sample. Your numbers are not. With FlowVisual you model your own process and get the same evaluation — on your machine, with your values.