Building a B2B Wholesale Store: Why Retail E-Commerce Rules Do Not Apply

Most e-commerce advice assumes retail: anyone can browse, everyone sees the same price, and the goal is to get a stranger to buy in one visit.

Wholesale breaks all three assumptions. When I built an online store for a mobile parts wholesaler, almost every default in a standard e-commerce setup was wrong for them — and understanding why mattered more than any technical decision I made afterwards.

The problem the business actually had

The business sells phone parts — screens, LCDs, batteries, accessories — to repair shops, not to phone owners. Orders were arriving over WhatsApp: a shop owner would message asking what was in stock, someone would reply with prices, and the order would be assembled by hand from a chat thread.

That works until it does not. Stock questions get answered from memory. Prices get quoted slightly differently to different shops. Orders get missed because they arrived at midnight in a thread with forty other messages.

They did not need “a website”. They needed the ordering process to stop living in someone’s head.

Why wholesale prices cannot be public

This is the requirement that changes everything.

A wholesaler’s prices are commercially sensitive. Publishing them means every competitor knows your margins, and every retail customer who finds the site asks why the shop down the road charges triple.

So the store has two states. A visitor who has not been approved sees the catalogue — products, photos, specifications, stock availability — and where the price should be, a prompt to register and get approved. An approved buyer sees real wholesale pricing on the same pages.

That single rule ripples outward. Registration is not a formality that gets you a faster checkout; it is a gate a human decides on. There is an approval step where the business confirms this is a real repair shop. Only then do prices appear.

It also means “logged in” and “allowed to see prices” are two different things, and the system has to keep them separate everywhere — product pages, search results, the cart, order confirmations.

Stock is a promise, not a number

In retail, an out-of-stock item is a mild annoyance. In wholesale it is a broken promise to a business that has already told a customer their phone will be fixed today.

So stock information had to be present everywhere a decision gets made, not just on the product page. Every listing shows whether an item is in stock, running low, or gone. Quantity selectors will not let you add more than actually exists. The same numbers appear on the listing and on the product page, because a buyer who sees “in stock” in search and “sold out” after clicking has learned not to trust the site.

Low stock gets its own state rather than being lumped in with “available”. A shop ordering twenty screens needs to know that only three are left before they build their week around it.

The order does not end at checkout

The part I underestimated: for this business, checkout is the middle of the process, not the end.

After an order is placed, somebody has to pick it, pack it, confirm it with the buyer, and arrange delivery. That conversation still happens on WhatsApp, because that is where the buyers are and no software was going to change that.

So rather than fight it, the admin side was built to feed it. From an order, staff can share the full details straight to WhatsApp, and export order data for their own records. The admin panel also shows every order a particular customer has ever placed, because the useful question is rarely “what is in this order” — it is “is this the shop that had the problem last month?”

Fighting the customer’s existing habits is usually a losing move. Meeting them where they already are costs less and works better.

Being found, now that search has changed

A wholesaler’s customers do not browse for fun. They search for a specific part, usually urgently, often while a customer waits at their counter.

For years that meant search engine optimisation: titles, descriptions, structured data, page speed. That still matters. But an increasing share of those questions now go to an assistant instead of a search box — “where can I get a screen for this model in Lahore” — and the answer is assembled rather than listed.

Which changes what the work is. An answer engine cannot rank you; it can only quote you. So the site has to state things plainly enough to be quoted: what is stocked, which brands, which city, what the minimum order is. Content written to game a ranking reads badly to a summariser, because it hedges and repeats. Content written to answer a question directly does well in both.

The most valuable thing I did for search on that site was not a meta tag. It was deleting a 15.5 MB image — a 200-megapixel logo file that nothing on the site actually used, sitting there slowing every page that loaded near it. Page speed is a ranking factor and a bounce factor, and the biggest wins are usually removals rather than additions.

What I would tell someone starting a similar project

Ask who is allowed to see what, before anything else. In retail that question is boring. In B2B it defines the entire data model, and retrofitting it is painful.

Find out where the real conversation happens. For this business it was WhatsApp. Any feature that ignored that was going to be worked around.

Treat stock as a trust problem. The number itself is easy. Showing it consistently everywhere, so buyers believe it, is the actual work.

Expect the admin side to be half the project. The storefront is what you show people. The admin panel is what the business uses eight hours a day, and it is where the time goes if you are honest about the scope.

The store is live and taking orders. The measure of success is not the technology — it is that nobody has to scroll back through a chat thread to work out what somebody ordered.