What a 64-Company Audit Found About B2B E-Commerce
What does a systematic audit of B2B e-commerce sites actually find wrong with them? creativestyle's B2B Commerce Report Health tested 64 manufacturers and wholesalers in the German-speaking health market against 45 criteria. The gaps run from missing documentation to sites that aren't readable by AI agents [1].
The findings are industry-specific, but the categories translate directly to any B2B vertical selling technical or regulated products: industrial supplies, MRO parts, chemicals, packaging, medical devices. This is the second half of a two-part breakdown. The first covered visibility, pricing display and search [1].
This one covers what happens after a buyer finds the product: the product page itself, the order, service, loyalty, the technical stack, and whether the site is ready for AI purchasing agents [1].
Give Buyers the Documents They Actually Need
A B2B purchase usually clears more than one desk. Procurement checks price and terms; quality management checks certificates and safety data; the end user checks fit and function. Datasheets, instructions and compliance evidence belong on the product page itself, without a login [1].
Technical specifications belong in a structured attribute table rather than free text [1].
The audit shows how wide the gap is: only 13 of the 37 manufacturers tested display evidence of efficacy or regulatory compliance for their products [1]. More than half show no application video on the product page at all [1]. For a buyer who cannot get past compliance review without that paperwork, its absence is a lost order, not a minor inconvenience.
Fewer Wrong Orders Start With Clear Units
Unclear packaging units, piece, box or pallet, are the single most common cause of mis-orders according to the report, which is why unit, content and variant need to be unambiguous on both the listing and the detail page [1]. Ordering itself should not require a detour either.
Only a third of the wholesalers tested offer a working quick-order function where a buyer enters an item number and quantity and is done, and roughly half leave delivery timing unclear until after the order is placed [1]. Shipping cost and minimum order value belong before the checkout button, not buried after it [1].
Manufacturers who don't sell directly should show where to buy on the product page, not in the fine print [1].
Contact and AI Assistance at the Point of Doubt
Service and contact score among the two weakest chapters in the audit. Questions arise on the product page, not on the imprint or contact-us page, yet only 7 of the 64 providers tested offer a live chat or chatbot at all, and where one exists, availability often isn't stated clearly [1].
AI-assisted advice is rarer still: 55 of 64 providers use no AI function anywhere in the buying process, and only four have implemented one fully [1]. For a buyer trying to confirm a technical fit at 9pm before a deadline, the absence of any assisted channel is the difference between finishing the order and abandoning it.
Most Wholesalers Still Don't Reward Repeat Buyers
Retention in B2B e-commerce is treated as an afterthought by most of the sites tested. Most of the wholesalers audited offer no loyalty or repeat-customer program of any kind, despite B2B relationships being built on repeat, high-frequency purchasing rather than one-off transactions [1].
The report frames retention as extending past the transaction itself: a service area for maintenance, spare parts and appointments, training content, and guide material that positions a supplier as a technical authority rather than just a catalog [1]. None of that requires new commerce infrastructure so much as a deliberate content and loyalty strategy layered on top of what already exists.
The Technical Basics Are Still Not Solved
Three in four websites tested load too slowly, and only 17 of the 64 sites reach at least a 90 out of 100 score in the automated performance test used by the report [1]. Speed compounds with two other gaps: more than half the providers run tracking before users have given consent, which requires a legal basis under German telecommunications and data protection law.
Two of five score zero on the security standards checked [1]. Mobile usability and accessibility round out the technical chapter, both increasingly relevant as buyers order from a phone on the shop floor rather than from a desktop [1]. Pure B2B shops that clearly address only business customers generally fall outside the accessibility law's mandatory scope, but the report treats it as good practice regardless [1].
| B2B commerce lever | Share of tested sites falling short |
|---|---|
| Compliance or efficacy evidence on the product page | 24 of 37 manufacturers show none [1] |
| Working quick-order by item number | About two-thirds lack one [1] |
| Live chat or AI chatbot at the point of purchase | 57 of 64 offer neither [1] |
| Loyalty or repeat-customer program | Most have none [1] |
| Automated performance score of 90 or higher | 47 of 64 fall short [1] |
| Full readiness for AI purchasing agents | 100% fail at least one criterion [1] |
Are B2B E-Commerce Sites Ready for AI Purchasing Agents?
No, not yet. The report measured this with an agentic-readiness scanner built by the shop software vendor Shopware, checking:
“B2B buyers are under immense pressure to justify investments and minimize risk.”
No wholesaler in the audited group meets every criterion for AI-agent readiness [1]. The category is young and not yet standardized industry-wide, so the low scores partly reflect market immaturity rather than individual failure.
What's harder to dismiss is the demand signal already building on the buyer side. Forrester's most recent buyer research finds that generative AI is fundamentally reshaping how business buyers discover, evaluate and purchase products and services [3].
Gartner separately expects a large share of enterprise software to include agentic AI by 2028, up sharply from a negligible base in 2024, though it also warns that a substantial share of current agentic AI projects will be canceled by the end of 2027 on cost, value or risk-control grounds [4].
That combination, rising buyer complexity plus immature agent standards, is exactly why structuring product data now pays off even before agentic ordering becomes mainstream: the same clean, structured catalog that an agent needs also improves search, filtering, comparison and visibility in AI-generated answers today [1].
- Structured, machine-readable product data
- Technical interfaces for agent access
- Explicit rules for how agents may access the site, including support for protocols such as llms.txt [1]
What This Means for B2B Commerce Teams
None of these fifteen levers requires a platform migration to fix. Most are content, data and configuration gaps: missing PDFs, unclear units, a checkout that hides shipping cost, a loyalty program that was never built. Each one sits on the path between a buyer's search and their order, and each is measurable.
The context makes the stakes bigger than one health-market audit. B2B sellers that offer e-commerce already generate 34% of their revenue online, yet a global buyer survey found 73% of B2B buyers prefer buying online while 85% remain frustrated with the experience, and 75% say they would switch suppliers for a better one [2].
Fixing product data, ordering friction and service gaps is not a nice-to-have ahead of the agentic-commerce transition. For most B2B sellers, it is the transition.
Sources
- [1]etailment, Teil 2: 15 Hebel, mit denen B2B-Shops von der Suche bis zur Bestellung besser werden — etailment.de
- [2]etailment, Warum Agentic Commerce die größte Chance für den B2B-Handel seit Jahren ist — etailment.de
- [3]Forrester, Forrester's 2026 Buyer Insights: GenAI Is Upending B2B Buying As Leaders Face Mounting Pressure To Justify Every Dollar Spent — forrester.com
- [4]Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 — gartner.com
