German online retail is back in the plus: €83.1 billion in goods revenue for 2025, up 3.2 percent per the bevh industry association, with 3.8 percent expected for 2026. But the growth arrived unevenly. The Chinese platforms Temu, Shein and AliExpress grew 27.2 percent to €3.7 billion, capturing around 30 percent of the entire market's growth, marketplaces rose to a 56 percent revenue share, and multichannel merchants, the classic mid-sized business with store and shop, were the only shrinking sender group in 2025. One insolvency monitor counted roughly 2,490 retailer insolvencies in twelve months, just short of the record. In Austria, 51 percent of online spending flows to foreign providers per the trade association's study, over 40 percent to Amazon alone.
Since 1 July 2026 there is at least a relief signal: the EU scrapped the €150 duty-free threshold, small parcels from third countries now carry a flat €3 customs charge per item, with a handling fee to follow in November. At 4.6 billion small parcels a year, 91 percent from China, that is not nothing, but it changes nothing about the homework: the independent merchant competes with machines and has to get his own cost side under control.
Returns are a scale problem, and you are on the wrong side of it
Germany is Europe's returns champion: per the University of Bamberg's last large merchant survey (data year 2021), nearly every fourth parcel goes back, around 530 million shipments carrying 1.3 billion items. Bamberg's cost figures are older but unique: the average return already cost €19.51 in 2018, and the 2012 survey shows the structural gradient that has since only steepened: small senders paid €17.70 per return, large ones €5.18. The corporation automates its returns line; the SME merchant pays for his by hand, and postage and wages have risen sharply since 2018. What automates here are above all the edges of the process: return-reason capture and triage, status communication, restocking, and preventively the product-data quality, since wrong expectations are the classic return driver.
The service inbox: plenty of folklore, one real lever
On service load, mostly vendor numbers circulate: "30 to 40 percent of all inquiries are where-is-my-order," the helpdesk vendors report in unison; no independent German study exists, and we say so plainly. What is solid is the customer side: in the 2025 Digital Commerce Report a third of German respondents complain about service problems, and nearly 40 percent regularly reach nobody. The lever is banal and effective: order-status inquiries can be answered from data that already sits in the merchandise-management and shipping systems, as automatic replies with tracking status, and everything else as a draft for sign-off. From 2 August 2026 the AI Act's labeling duty applies: a service chatbot must identify itself as AI, and a note in the legal page does not suffice. And because the opposite circulates stubbornly: ordinary AI-generated product texts do not need labeling; the duty covers chatbots and deceptively realistic media content.
The compliance treadmill
Hardly any industry got as much compliance upkeep loaded on in eighteen months: since December 2024 the product-safety regulation GPSR requires manufacturer details, a responsible EU person, warnings and a product image for every single listing, technically enforced by the marketplaces. Since June 2025 the accessibility act applies to B2C shops, exempting only service providers with under ten employees and at most two million euros in revenue, both criteria together. From August 2026 the packaging regulation PPWR adds conformity duties. Each of these obligations must be maintained per item and per channel, and exactly this field upkeep across shop, merchandise management and three marketplaces is a textbook automation case: a system that checks listings against the mandatory fields and reports the gaps before the marketplace suspends.
Why 70 percent AI engagement yields 9 percent effect
The Shopware merchant survey 2026 puts the gap in one line: over 70 percent of merchants actively engage with AI, but only 9 percent see a measurable revenue effect. The reason sits on the tool shelf: the shipped AI of the shop systems, Shopware AI in the paid plans, JTL's AI extensions, Amazon's and eBay's listing generators, automates single steps, above all product texts. Use those built-in tools; they are good and mostly already paid for. But the expensive processes of a merchant's day, the service inbox, returns, cross-channel data care, compliance fields, run across systems, and there no platform feature delivers. The 2026 lever lies in connected workflows within the tool landscape German merchants already own: JTL, Shopware, the marketplace APIs. That is integration work, not a feature list, and exactly the pattern we found in ten other industries.
Whether your business has an automation case, the 60-second check will indicate; what projects cost is in the pricing overview, with sources, and how sign-offs work with us in draft plus sign-off.
As of 10 July 2026, not legal advice. Sources: bevh annual press conference (January 2026) and H1 release (July 2025), HDE online monitor 2025, University of Bamberg returns research (data years 2012, 2018, 2021, each labeled), Austrian trade association/KMU Forschung Austria (June 2025), HDE/Safaric AI study 2025, Shopware merchant survey 2026 (n=360), Publicis Sapient Digital Commerce Report 2025, EU Commission on small parcels, chamber of commerce on the customs threshold (July 2026), GPSR, accessibility act, PPWR, EU AI Act Art. 50, vendor documentation Shopware/JTL/Amazon/eBay (2025/26). Debunks: "406,000 shops", "70 percent WISMO", "AI product texts need labeling".