Sandra Dirks
Why the most effective lever against returns lies in your data foundation
In short: Clean product data from PIM and DAM reduces avoidable returns and helps AI systems such as ChatGPT, Gemini, or Perplexity understand and recommend products correctly.
Preventing returns in the first place is an effective lever for online retail. Better product data leads to fewer wrong purchases, and with them less packaging, transport, and processing effort. At the same time, product data increasingly determines how visible a product is in AI-driven buying processes.
Table of contents
What does a return cost in online retail?
The EHI survey 2024 of 146 online retailers in the DACH region shows how widely the cost per return varies:
- 30% of the retailers surveyed estimate it at € 5-10
- 26% estimate it at € 10-20
- individual retailers report up to € 50 or more
Taken together, 56% of retailers put the cost of a return between €5 and €20. As a planning assumption, around €10 per return is a reasonable starting point. Depending on the assortment, shipping method, and processing workflow, the actual figure can be higher or lower.
Example: With one million items shipped per year, a return rate of 30%, and €10 per return, reducing the return rate by just one percentage point avoids about 10,000 returns and saves roughly €100,000 per year.
Which product data reduces the return rate?
How recent product data fuels avoidable returns is shown by a recent consumer survey by EHI together with KPMG among 500 online shoppers in December 2025. The top three reasons for returns are inaccurate size information (76.4%), inaccurate descriptions (74.6%), and discrepancies from product images (72.4%).
Therefore, avoidable return reasons can be addressed specifically through better product information. The most important levers are:
- Size and fit information: detailed size charts, fit notes, model sizes
- Product images: meaningful, detailed, accurate product photos instead of generic imagery
- Product videos and 360° or 3D views: a realistic impression of shape, volume, and details
- Material and color details: precise rather than embellished descriptions
- Technical specifications: clear, error-free technical data
What role does product data play in AI-driven purchase recommendations?
Product data now works on two levels at once. Good product data directly influences people’s purchase decisions. At the same time, more and more customers rely on AI systems such as ChatGPT, Gemini, or Perplexity when deciding what to buy, whether as a research tool, a source of recommendations, or, in the future, as an agent acting on their behalf in agentic commerce.
AI search systems such as ChatGPT or Perplexity combine information from multiple sources into a single answer instead of showing ten links to choose from. For a product to be classified and recommended correctly within that answer, an AI system needs reliable, structured, and consistent data across all channels, for example:
- clear product attributes instead of marketing copy alone
- accurate dimensions and material details
- current availability
- meaningful image metadata
If this information is missing or contradicts itself across the online shop, marketplace, and data sheet, a language model will likely fill the gaps, with the risk of false or inappropriate details. The result is not only weaker purchase recommendations but also exactly the wrong purchases that later come back as returns. Product data quality is therefore no longer just a concern for your own shop. It is also a prerequisite for being mentioned, and represented correctly, in AI-driven buying processes.
Why are PIM and DAM the foundation for reducing returns and AI visibility?
Product data cannot be fixed touchpoint by touchpoint. It has to be right at the source. A Product Information Management system (PIM) consolidates product data in a structured way and provides it consistently to all channels. A Digital Asset Management system (DAM) does the same for images, videos, and their associated metadata. Anyone who builds on incomplete or inconsistent data will, with AI tools, mainly automate errors, faster and at greater scale than before. Clean, detailed product data from PIM and DAM is therefore the real foundation for reducing returns and the first step for everyone investing now in AI-driven buying processes and content creation.
Is your product data ready?
Is your product data ready to prevent returns and to be understood by AI systems? We help you set up your PIM and DAM systems, along with your database-driven Dynamic Publishing, in an optimal way.
FAQ
How much does a return cost in online retail?
Costs vary widely. In the EHI survey 2024, 56% of retailers reported € 5-20 per return. As a planning assumption, around € 10 applies, though it can be higher or lower depending on the assortment, shipping method, and processing workflow. The basis is the EHI survey 2024 of 146 online retailers in the DACH region.
Which product data reduces the return rate the most?
Above all, precise size and fit information, accurate product images, 360° or 3D views, and clear material, color, and compatibility details.
Why are PIM and DAM relevant for visibility in AI systems?
Because AI systems such as ChatGPT, Gemini, or Perplexity rely on structured, consistent product data to make purchase recommendations. PIM and DAM deliver exactly this data foundation across all channels.
