The B2B ecommerce trends that matter heading into 2027 point the same way: business buyers want to research, compare and reorder without waiting for a sales rep, and more of that work is done by software, from AI assistants that summarize suppliers to procurement systems and AI agents that place orders. Ecommerce is already the largest revenue channel for B2B sellers that offer it, according to McKinsey, and Gartner forecasts that most B2B buying will be intermediated by AI agents by 2028.
Below are eight shifts, each backed by research you can open, with what each one asks of your product data: specifications as typed fields, classification such as ETIM, identifiers, order units and documents. The last sections turn them into a B2B ecommerce strategy for the next 12 months and flag the trend claims worth reading with caution. If you are still choosing a platform, our list of B2B ecommerce features covers pricing, approvals and punchout in detail, so this post does not repeat it.
Gartner, McKinsey and Forrester: the research behind this list
B2B commerce trend lists come from two kinds of publishers. Analyst firms survey buyers and sellers, and Gartner, Forrester and McKinsey all publish summaries of their B2B buying research. Commerce platform vendors such as Sana Commerce and OroCommerce publish yearly trend posts too. Those are useful for seeing what platforms are building, but they are vendor views and often quote figures without a traceable source. The figures in this post come from the first group, plus EU legislation and the ETIM standards body:
- Gartner surveyed 645 B2B buyers in August and September 2025 on rep-free buying and their use of generative AI (Gartner, May 2026).
- McKinsey's 2024 B2B Pulse survey gathered responses from nearly 4,000 B2B decision makers across 34 sectors in 13 countries (McKinsey, September 2024).
- Forrester summarized its State of Business Buying, 2026 research on how buyers use AI (Forrester, January 2026).
Eight B2B ecommerce trends and what each asks of your product data
1. Rep-free buying becomes the default for routine purchases
In Gartner's survey, 67% of B2B buyers said they prefer a rep-free experience. That does not mean buyers never want a person: the same research found that 69% prefer to validate AI-generated insights with a sales rep (Gartner). Routine work such as finding a specification, checking a price break or reordering moves to self-service, while complex or high-risk decisions still involve people.
What it asks of your product data: every question a rep used to answer by phone now needs an answer on the product page. For a technical product that means typed specifications with units, the order unit and pack size, compatible accessories and the current data sheet. A specification that only lives in a sales engineer's head is a reason to pick up the phone, or to buy elsewhere.
2. Ecommerce becomes the largest B2B revenue channel
McKinsey's 2024 B2B Pulse found that 71% of B2B respondents offer some form of ecommerce and that online sales account for 34% of their revenue, which makes ecommerce the top revenue-generating channel ahead of in-person sales (McKinsey). Buyers' comfort with remote and self-serve spending also rose, notably for orders above $500,000.
What it asks of your product data: a B2B webshop that started as an extra for small orders now carries a third of revenue, so the long tail matters. A common pattern is that the top 500 articles get attention and the remaining tens of thousands keep a one-line supplier description and no attributes. Large orders often include exactly those slow movers, and an order that cannot be completed online goes back to email.
3. Buyers research suppliers with generative AI
Forrester reported that 94% of business buyers use AI in their buying process, up from 89% a year earlier, and that 61% use private AI tools provided by their organization (Forrester). Gartner found that 45% of buyers used generative AI during a recent purchase, mainly to gather information on vendors and products, and that 51% think they are more likely to encounter misleading information from GenAI.
What it asks of your product data: an AI assistant can only summarize what it can read. When your website says 6 kA, the data sheet says 10 kA and a distributor's listing says nothing, the buyer gets a guess. Consistent values across your site, feeds and documents are the practical defense. A quick test: ask a general AI assistant for the specifications of three of your best sellers and compare the answers with your own records.
4. AI agents move from research to ordering
Gartner's predictions for 2026 and beyond include the forecast that by 2028, 90% of B2B buying will be AI agent intermediated, pushing over $15 trillion of B2B spend through AI agent exchanges (Gartner, October 2025). Gartner ties this to a data feed economy in which verifiable operational data is a condition for taking part. It is a forecast, not a measurement, but it describes what an agentic commerce buyer needs: data a machine can trust without a person checking it.
What it asks of your product data: identifiers that match across systems (your article number, the manufacturer part number and the GTIN), attributes stored as name, value and unit rather than prose, current prices and availability, and order rules a machine can apply, such as minimum order quantity, increment and pack size. Free text like "16A B-curve 1P 6kA" in a description is readable for a person and close to useless for an agent comparing three suppliers.
5. Buyers switch between ten channels and expect the same answer in each
McKinsey's survey found that B2B customers use an average of ten interaction channels in their buying journey, up from five in 2016, and are likely to switch suppliers when the experience across them is poor. For a wholesaler those channels include the webshop, a customer portal, a marketplace or sector platform, punchout sessions, catalog files for buying groups, PDF price lists and the sales rep's tablet.
What it asks of your product data: one maintained record per article, with channel-specific output generated from it instead of a separate copy per channel. Ten copies drift apart within a quarter. Our post on unified commerce covers the system side; for product data, the rule is that a correction made once reaches every channel.
6. Procurement teams ask for punchout and catalog files earlier
Larger customers buy through their own procurement systems, so your catalog has to arrive there: as a live punchout catalog session over cXML or OCI, or as a catalog file such as BMEcat that the buyer imports. As buying groups grow and procurement joins the process early, these requirements show up in the first conversation rather than after the contract is signed.
What it asks of your product data: procurement systems are strict. A catalog import rejects an article without an order unit, a classification or a price, and a punchout cart fails to match when your article number differs from the one in the buyer's contract. Clean identifiers, units and classification are the entry ticket. The B2B features post linked above explains the protocols themselves.
7. Classification standards set the pace: ETIM 10 and the next release
ETIM International released ETIM 10.0 in December 2024 and publishes official releases about every three years, with dynamic releases in between (ETIM International). In electrical, HVAC and plumbing wholesale, manufacturer data and customer requests are often tied to a specific ETIM release. ECLASS and UNSPSC play a similar role in other sectors and in procurement.
What it asks of your product data: give each article a class, fill the class features with values from the permitted list and in the right unit, and record which release you mapped against. Then treat the move to the next release as a project, because classes and features change between releases. Our ETIM guide explains how classes, features and values fit together.
8. Regulation turns product data into compliance records
The EU Ecodesign for Sustainable Products Regulation introduces the Digital Product Passport, a digital record of a product's materials, compliance and repair information, and leaves the details per product group to later acts. Batteries come first in practice: from 18 February 2027, all batteries must carry a QR code, and for LMT batteries, industrial batteries above 2 kWh and electric vehicle batteries it must give access to a battery passport (EUR-Lex, Regulation (EU) 2023/1542).
What it asks of your product data: information that used to sit in a compliance folder, such as material composition, substances of concern, repair and spare part information and declarations, becomes a set of product attributes that must be versioned and shared. Distributors will be asked for it by their own customers. Our Digital Product Passport entry summarizes what a passport contains.
One cable gland, eight trends: a worked example
To make this concrete, take an article an electrical wholesaler sells by the thousand: a gray polyamide cable gland with an M20 thread, rated IP68. Here is the record it needs and the trends each field serves.
The M20 cable gland record, field by field
| Identifiers | Article number, manufacturer part number and GTIN | 4, 6 | The agent, the punchout cart and the buyer's ERP must all resolve to the same item |
|---|---|---|---|
| Specifications | Thread M20 x 1.5, clamping range 6 to 12 mm, IP68, material polyamide, color gray, as separate fields with units | 1, 3, 7 | They drive filters, AI summaries and the ETIM class features |
| Classification | ETIM class and release | 7 | One class code and the release number it was mapped against |
| Order units | Sales unit piece, packed per 100, minimum order 100, increment 100 | 2, 4 | Without them, a price break or an agent's order quantity comes out wrong |
| Prices | Unit price and quantity breaks | 2, 4 | Current, and expressed in the same unit as the order unit |
| Documents | Data sheet and declaration of conformity | 1, 8 | The current revision, attached per language |
| Accessories | M20 lock nut and sealing washer | 1, 2 | A self-service buyer completes the order without calling |
Now count how many of those fields are typed, current and identical across your channels today. In many wholesale catalogs, only the identifiers and the price are reliably filled. In that case the self-service, AI and procurement trends all stall on the same gap, and fixing it once serves all of them.
A B2B ecommerce strategy for the next 12 months, built on these trends
A trend list is only useful if it changes what you do next quarter. This sequence works for a wholesaler or manufacturer with a few thousand to a few hundred thousand articles:
A 12-month B2B product data plan
Each step produces something a buyer or a buying system notices. The order matters: portals, AI features and agents built on top of thin data mostly expose the thinness faster.
B2B trend claims to read with caution
Not every trend figure in circulation holds up. Before you put a number in a business case, check it against these patterns:
- Forecasts presented as facts. Gartner's 90% figure for AI agent intermediated buying is a prediction for 2028. Quote it as one.
- Market size totals without a source. Global B2B ecommerce market figures in the trillions are often copied from article to article without a report you can open.
- Vendor surveys about vendor categories. A platform vendor's survey on how much buyers want that vendor's feature is not neutral. It can still be useful; label it.
- AI uplift percentages. Claims that AI raises conversion or margin by a fixed percentage rarely state the baseline, the sample or the time frame.
- Generational claims. Figures about millennial and Gen Z buyers vary with the survey and the age definition. Check who was asked.
Classifying, checking and exporting B2B product data in a PIM
Most of the work above is product data work, which is where a PIM fits. In WISEPIM, products can be classified in ETIM or GS1 GPC with the ETIM release pinned, and each product shows a conformance check of which required class features are valid, missing or invalid, with an option to fill missing features with AI. Feed Hub exports BMEcat 2005 files with ETIM features, Channel Readiness shows which products miss required fields per channel, and AI enrichment extracts attributes from supplier text for you to review before they go out.
Company accounts, quotes, approvals and punchout sessions stay in your commerce platform or ERP. If you are comparing systems for the product data side, see PIM for B2B or our guide to PIM for distributors.
