Not every AI idea is a viable use case.

Our three-level method for separating genuine from apparent AI potential in SAP EWM and SAP MFS.

“Do you have recurring error patterns in your warehouse?”

The answer is almost always “yes.” And then what?

For us, a blanket “yes” is not yet a recommendation – it is only the starting point of a structured assessment with three levels that build on each other.

Only those who work through all three receive a recommendation in the end – not after the first answer.

The path to a robust recommendation

Every topic area goes through the same three steps. Only when all three are passed do we talk about implementation.

Level 1 – In-depth diagnostics

Instead of a blanket question, we ask concrete, technically sound questions for each topic area – for example on error detection, prioritization, workforce planning, bottleneck management or quality assurance. For example: How many warehouse task errors actually occur per week, and how many of them still tie up power-user time manually today?

Level 2 – Qualification

We assess each topic area across six dimensions – as a traffic light, not an abstract score, so the assessment stays easy to follow in conversation. Only pursue a topic further if at least four of the six dimensions are “yellow” or better.

Level 3 – Reality filter

This last condition is non-negotiable: A topic that the customer does not perceive as a problem gets no recommendation – regardless of the objective score. For example, when a manager says: 'This actually costs us money – but it is not on our agenda right now.' Then we wait, even if the objective score would speak in favor.

The six assessment dimensions at a glance

The six dimensions we use to assess each topic area. Do a rough self-check of where your situation currently stands – red, yellow or green.

Dimension
Red
Yellow
Green
Frequency/volume
Isolated case, less than once a month
Several times a month
Weekly or more often
Economic impact per incident
< €1,000 estimated impact
€1,000–10,000
> €10,000 or customer retention/reputation risk
Data maturity
No structured data available
Partly available, < 6 months of history
Structured, ≥ 12 months of history
SAP-native coverage
A standard agent already covers this
Covers some aspects
No suitable standard agent available
Reversibility/risk
Not reversible, high governance requirements
Partly reversible
Reversible, low risk
Customer pain
Customer does not see this as a problem
Mentioned as annoying when asked
Raised unprompted as a wish topic
Red
Frequency/volume
Isolated case, less than once a month
Economic impact per incident
< €1,000 estimated impact
Data maturity
No structured data available
SAP-native coverage
A standard agent already covers this
Reversibility/risk
Not reversible, high governance requirements
Customer pain
Customer does not see this as a problem
Yellow
Frequency/volume
Several times a month
Economic impact per incident
€1,000–10,000
Data maturity
Partly available, < 6 months of history
SAP-native coverage
Covers some aspects
Reversibility/risk
Partly reversible
Customer pain
Mentioned as annoying when asked
Green
Frequency/volume
Weekly or more often
Economic impact per incident
> €10,000 or customer retention/reputation risk
Data maturity
Structured, ≥ 12 months of history
SAP-native coverage
No suitable standard agent available
Reversibility/risk
Reversible, low risk
Customer pain
Raised unprompted as a wish topic

Mostly yellow or green on at least four of the six dimensions? Then this topic area is worth a closer look.

Go deeper

You will find the technical background in our whitepaper "AI in SAP EWM".

No AI at any price

The result of our method is not a technology decision on principle, but a transparent recommendation with a business-case perspective – or, just as often, the realization that a topic is not yet ready for implementation.

Glass filter funnel with five checkpoints: many AI ideas enter, only a few orange-marked use cases pass every stage
A recommendation based only on the first answer is not a recommendation – it is a guess.
Portrait of Marc Hofheinz

Marc Hofheinz

Managing Partner | Qinlox Consulting GmbH

Employee monitoring automated warehouse systems on screens in a control room

Where would your warehouse stand in this assessment?

The best way to find out is a short reality-check call – non-binding and with no sales pressure.