By John Wheeler
For a long time, the path to a sale was fairly easy to understand.
A customer walked into a store, opened a catalog, called a salesperson, or ordered through a distributor. The seller controlled most of the information. Sales handled the customer. Marketing created the message. Operations made or moved the product. Finance measured the result.
Companies were built around that path.
Then the way people shopped began to change. Websites, online catalogs, social media, mobile apps, digital marketplaces, ratings, reviews, and ecommerce systems all became part of the buying process. Now AI is entering the journey as well.
The customer adapted quickly. Most companies did not.
Many businesses responded by adding one new team after another. They added ecommerce, product data, digital marketing, data science, and now AI. Each new group was asked to solve the latest problem while the rest of the company remained much the same.
That model is reaching its limit.
Companies should stop treating ecommerce, product data, and AI as separate projects. They must reinvent how work gets done around how people actually discover, evaluate, buy, and use products.
The Normal: A Clear Path From Seller to Buyer
The old business model matched the old buying journey.
The customer usually entered through a known door. A retailer managed the store shelf. A distributor representative managed the account. A salesperson knew which products to present. The manufacturer trained the sales team, printed sales material, attended trade shows, and worked through established channel relationships.
Information moved at a human pace.
Product changes might be reviewed a few times a year. Catalogs had deadlines. Sales plans followed annual cycles. A missing description or outdated image could create a problem, but it did not always stop the product from being discussed. A strong salesperson or broker could fill in missing details during a conversation.
Companies built departments and systems to support this world. The ERP system held item numbers, prices, inventory, and orders. Marketing kept images and sales copy. Quality maintained ingredients, allergens, and claims. Operations knew case dimensions, storage needs, and production limits. Sales knew the customer and the channel.
This structure was not foolish. It reflected how the market worked.
But the market no longer works that way.
The Explosion: The Buying Journey Breaks Apart
The modern customer can enter from almost anywhere.
A person may first hear about a product from a friend, an online review, a social post, a search result, a distributor app, or an AI assistant. They may research it on the manufacturer’s website, compare it somewhere else, speak with a salesperson, and finally purchase through a third party.
Discovery, evaluation, purchase, and use no longer happen in one place.
The Digital Shelf Institute describes this as an omnichannel world. Its report, Reinventing the Organization for Omnichannel Success: Beyond “Where Ecommerce Sits,” argues that companies should stop asking where ecommerce belongs on the organization chart. The more useful question is how the whole business should operate around the way customers now shop.
That is a much bigger idea than ecommerce.
Ecommerce is not only the place where a customer clicks “buy.” Digital information may shape a purchase that is later completed through a salesperson, distributor, dealer, or physical location. A traditional sale may now begin with a digital search. A digital order may still depend on a trusted human relationship.
AI adds another path. It can search, compare, explain, and narrow the choices before the customer speaks with anyone from the company.
The number of places where a product can be found has exploded. So has the amount of information needed to support the sale.
Yet inside many companies, that information is still split across departments and systems built for the old normal.
One Restaurant Operator, One Business Problem
Imagine a restaurant operator looking for an allergen-friendly sauce.
The operator is not trying to complete an ecommerce process. They are trying to solve a business problem.
They may ask an AI assistant for ideas. They may search a distributor’s catalog, check a manufacturer’s website, compare ingredient and allergen statements, review pack sizes and preparation instructions, and ask a distributor representative whether the product is available. A broker may provide a sample. The operator may calculate the cost per serving, test the product in the kitchen, add it to the menu, and decide whether to reorder it.
To the operator, this is one journey.
To the manufacturer, it may involve sales, marketing, quality, operations, finance, a broker, a distributor, and several software systems.
The operator expects every part of the story to agree. The product identity, ingredients, allergens, pack size, storage instructions, preparation method, availability, and cost all need to make sense together.
If the distributor catalog says one thing, the manufacturer’s website says another, and the salesperson gives a third answer, the operator does not see three departments making separate mistakes.
The operator sees one company that cannot provide a dependable answer.
This is why the customer journey must become the unit around which the work is designed.
Product Data Is a Set of Promises
Product data is often treated as a technical matter. It may be assigned to IT, ecommerce, marketing, or the person responsible for entering fields into customer portals.
But product data is not just a collection of fields.
It is a set of promises.
An ingredient statement is a promise. An allergen statement is a promise. A case dimension is a promise. A preparation instruction is a promise. A product image, storage rule, availability statement, and performance claim are all promises.
Those promises affect whether a product can be found, whether it fits the customer’s need, whether it can move through the supply chain, and whether people can use it safely.
Different teams may create and approve different facts. That is reasonable. The company still needs one governed method for deciding which facts are trusted, who owns them, where they are stored, how they stay in sync, and where they are allowed to go.
An ERP system such as Microsoft Dynamics 365 Business Central can hold approved operational facts. GS1 standards can help identify and describe products in a common way. A product information management system can manage richer content and channel needs. Syndication networks can deliver approved information to distributors, retailers, and other partners.
These systems have different jobs. Buying all of them does not automatically create one trusted product story.
Management must still define ownership, approval, business meaning, and the process for resolving conflicts. The goal is not to force every fact into one database. The goal is to create one governed set of product promises that can travel through the full buying journey.
AI Is Becoming Part of How Products Are Discovered
US Foods offers a useful foodservice example.
Its platform is called MOXē. US Foods describes MOXē as an all-in-one ecommerce and business application where customers can browse products, manage inventory, place or manage orders, and track deliveries. It is not simply an AI chatbot. It is becoming a digital place where operators run more of their purchasing and menu work.
US Foods has also added AI-powered tools to that environment. In February 2026, the company announced Menu IQ, a feature built into MOXē that lets operators upload recipes, calculate food costs, track menu-item margins, and identify possible improvements. Menu IQ can use actual food costs from MOXē Inventory.
This changes the path to product discovery.
The restaurant operator may no longer begin by searching for a product by name. The operator may begin with a business question: Which menu items are losing margin? What ingredient is driving the cost? What product could improve the economics of this dish?
The system can then help connect that need to products, costs, and inventory.
US Foods has also worked with Coveo on AI search and product discovery within its digital platform. The stated goal is to help buyers find what they need even when product names, spelling, and customer catalogs vary. Coveo reports better conversion and fewer searches with no results, although those performance figures come from the technology provider and should be read as a vendor case study rather than independent proof.
The larger direction is still clear.
The distributor is moving closer to the operator’s decision. It is not only taking the order. It is helping the operator understand recipes, costs, margins, inventory, and product choices.
For a food manufacturer, that creates a new question: Will the manufacturer’s product facts be complete and useful enough for a system like MOXē to find, understand, and present the product at the right moment?
An AI system cannot taste a sauce. It cannot personally watch it perform in a kitchen. It works from the information and signals it can reach. It may consider product identity, descriptions, ingredients, pack size, availability, cost, customer fit, and prior performance.
The manufacturer’s story must therefore become understandable to both people and machines.
This does not make human selling less important. Brokers, distributor representatives, chefs, and manufacturer salespeople still provide taste, judgment, trust, and real-world experience. But the machine may help decide which products reach the human conversation.
That is the shift manufacturers cannot ignore.
The New Normal: Build the Company Around the Journey
The answer is not to create a bigger ecommerce department, a separate product data empire, and an AI team working off to the side.
The answer is to connect the company around the customer’s job.
That does not require every small or midsize company to replace its organization chart. It means redesigning the work that crosses the chart.
A manufacturer might begin with one journey: bringing a new product from internal approval to customer discovery and channel availability.
The company would follow the product through every step. Who approves the product facts? Where are those facts recorded? How are GS1 records maintained? Who prepares distributor content? What happens when a channel asks for information the company does not have? How does sales know the product is ready? How does management learn whether the product was found, ordered, used, and reordered?
This view will expose the real breaks in the system. Facts conflict. Ownership is unclear. Employees enter the same data several times. Approvals happen in email. Important work depends on one person’s memory. Each department measures its own activity while no one measures whether the complete journey worked.
Those are not just data problems. They are design problems.
The company can then create one shared way of working around a clear customer outcome. Each fact gets an owner. Each system gets a defined role. Each handoff has a rule. Exceptions go to the right person. Results are measured across the full journey.
That is what reinvention looks like in practice.
AI Workers Can Connect the Work
Once the journey is clear and the information is governed, AI can become useful business capacity.
A Product Data Coordinator agent could check for missing attributes, compare approved records, monitor channel requirements, prepare updates, and send exceptions to the person who owns the answer.
A Sales Intelligence agent could connect product availability, customer history, channel performance, and market signals before recommending the next action.
A Customer Support agent could prepare an answer from approved product facts and send uncertain or high-risk questions to a person.
These AI workers can help cross old department lines because their roles are built around an outcome.
But capability does not create authority.
Every AI role still needs approved sources, limited access, tests, stopping rules, monitoring, and a named human owner. Important actions should require approval. High-risk work should receive independent review. The company must be able to stop the worker and revoke access quickly.
The goal is not AI everywhere. The goal is a sound division of work among people, software, and AI.
The Intelligent Company Learns From the Whole Journey
An intelligent company does more than collect data or install AI.
It records what happened in dependable business systems. It combines those records with shared definitions and company knowledge. People and AI help understand what the facts mean. Decisions are made within clear authority. Work moves through controlled processes. Results are measured and used to improve what happens next.
In simple terms:
Record. Understand. Decide. Act. Learn.
That loop should follow the customer journey from discovery through use and reorder.
Ecommerce provides places for the customer to interact. Product data supplies trusted promises. Operational systems record orders, inventory, cost, and delivery. AI helps people understand signals and prepare work. Human judgment and governance keep the system responsible.
These should not be separate strategies. They are parts of one company learning how to serve the customer better.
Where Leaders Can Begin
The first step is to map how customers actually discover, evaluate, buy, use, and reorder one important product. Include every meaningful touch point, not just the company’s website.
Next, identify the facts required at each stage. Determine where each fact lives, who owns it, who approves it, and whether every channel receives the same answer.
Then choose one journey that crosses several departments and give it a shared business outcome. Do not measure success only by whether each department completed its task.
After the work and facts are clear, select one repeated task that an AI worker could prepare or perform within a controlled role. Define the inputs, output, authority, tests, stopping rules, and human owner before choosing the tool.
Finally, measure the whole result. Track missing information, exceptions, cycle time, human effort, accepted outputs, customer response, revenue, cost, and what the company learned.
Start small enough to control, but large enough to matter.
Reinvention Starts With a Better Question
The old question was: Where should ecommerce sit?
The next question may be: Who owns AI?
Both questions begin with the company and its tools.
The better question begins with the customer:
How do people actually shop, and what must our company become to serve that journey well?
The companies that win will not simply have the best ecommerce team, the largest product database, or the most AI agents.
They will be the companies that present one dependable face to the customer. Their product promises will agree. Their systems will connect. Their people and AI workers will have clear roles. Their decisions will use evidence. Their organization will learn from what happens after the sale.
That is more than ecommerce.
It is more than product data.
It is more than AI.
It is the work of reinventing the company around how people actually shop.
Source and Evidence Notes
The organizational argument in this article was informed by the Digital Shelf Institute report, Reinventing the Organization for Omnichannel Success: Beyond “Where Ecommerce Sits”. The report draws on 33 interviews and a survey of more than 90 brand leaders. DSI is connected to Salsify, so its findings are best treated as useful industry research rather than neutral academic proof. The applications to foodservice, governed AI workers, and the Intelligent Company are Wheeler IS interpretations.
The description of MOXē and Menu IQ is based on US Foods’ February 23, 2026 announcement and the US Foods MOXē application description. The product-discovery example also draws on Coveo’s US Foods case study. Coveo is the technology provider, so its reported performance results are vendor-supplied evidence rather than an independent study.