KI-Integration im Handel 2026 - Zwischen Aufbruch und Umsetzung
Cetin Acar, Philipp Hübner, Imke Hahn, Markus Winter · EHI Retail Institute · 10 Nov 2026
Source finding
75.3 percent cite older systems, 73.1 percent data silos and 72.4 percent data quality as challenges when using data for AI. These values are not parts of a whole.
Our interpretation
Trace the data path for one use case before adding more AI tools. The survey identifies perceived barriers, not measured benefits of a specific solution.
Method and evidence quality
156 leaders in DACH retail; online survey in September/October 2026. Medium evidence for views within the research network, not a representative market rate.
Limitations and interests
Self-report, not a field experiment. The denominator for the data-barrier question is not publicly stated. Company count differs: 132 in the release, 133 in the sample report. No separate SME results. EHI sells research, consulting and training; technology partners contributed to the guide. Only the primary release and sample report were reviewed, not the paid full report.
What businesses can do
Before the next AI initiative, trace the data for one specific task. Identify missing, conflicting or outdated facts, assign owners and test again after correcting them.
Where to start
Data-quality section in the primary release, followed by the sample report methodology.
Connections
Architecture and Modularity · Workflow Redesign
Try it in your business
Our application idea, separate from the study finding
Our application idea, not an observed study case: A retailer wants AI to answer product questions. The team checks whether product attributes and availability can be combined from the responsible systems and kept current.
First steps, measures and team check
Who this is for
Business owner · E-commerce lead · Data owner · Retail and e-commerce · Customer service
Direct DACH retail evidence, but no separate SME findings. The proposed check is our own application idea.
Misuse to avoid
Misuse to avoid: Buying a new AI tool while assuming existing product data are already connected, current and reliable.
Steps for this week
- Select ten recurring product questions and name the data fields each requires
- Trace each field to its source and check freshness, permissions and inconsistencies across systems
- Test the questions using the same data and record missing evidence, incorrect facts and rework
What to measure
Share of test questions answered correctly with current evidence and without manual data searches
Count incorrect product facts and outdated availability separately
Check the same questions before and after correcting the data. Include manual search and correction in total time; do not infer a target from the survey
Team check
- Which data are missing for our ten selected product questions?
- Who owns data freshness and the connection between systems?
- How many answers are correct, current and supported after the correction?
Evidence and publisher interests
Industry survey of perceived data barriers. No observed error rate and no evidence that a particular data intervention increases productivity or revenue.
EHI sells research, consulting and training. Retailers and technology partners contributed to the guide. We reviewed the primary release and sample report, not the paid full report.