“We want to sell online.” One sentence, but behind it sit two projects that have little in common: a store open to the public, and a portal where your existing customers reorder what they already buy.
They cost differently, they are designed differently, and they fail for different reasons. Choosing wrong does not show up in the quote. It shows up six months later, when the platform is finished and nobody uses it.
The ratio that decides which one is yours
One question settles this in a minute. Of ten orders that would come in through the web, how many would come from customers who already buy from you every month, and how many from someone who has never bought from you?
If most are the regulars, yours is a reorder portal. The work is not in attracting anyone. It is in letting a customer who already decided to buy from you skip the email and the wait. If most would be strangers, then it really is a store, and the actual project is not the platform but how people will find it.
Both are legitimate. What does not work is building a store for customers you already have, because everything that makes a store good — a catalog that attracts, a cart for visitors, card checkout — is exactly what gets in your regular customer’s way.
If you sell to large maquiladoras, you may need neither
Worth saying up front, because it saves money.
If your main customers are large plants, they are not going to log into your portal. They have their own, or they will ask you to connect over EDI, and their purchase will keep coming out of their system, not yours. A portal adds nothing there.
A portal pays off with the rest of your book: shops, distributors, parts houses, mid-size companies around Juárez and El Paso, the repeat customer who buys three times a year. These are people who call or write today, who consume somebody’s time ordering the same thing as always, and who would buy more often if they did not have to wait for an answer.
Work out what that rest is worth in revenue. If it is a small share, the project can wait.
Pricing is where it stalls, and it is not a technical problem
This is the part that stops more projects than anyone admits.
Ask: do all your customers pay the same price for the same part? At an industrial company the answer is almost always no. There are lists, there are discounts negotiated years ago, there is one customer with a special price because the owner agreed to it over lunch.
To publish a price you have to be able to calculate it. If every order is quoted by phone, what is missing is not a platform. It is written pricing rules. That is process, and no system can guess it.
The good news is a portal tolerates this better than a store does. In a portal each customer signs in and sees their price, so you do not need a public price — you need the lists written down, current, and in one place. If they live in the rep’s head today, that was already a problem, and the portal will only put it on display.
The stock number you are about to publish
Second uncomfortable question: the availability figure the customer will see — where does it come from, and how often is it refreshed?
A spreadsheet somebody updates on Mondays does not work for this. The moment you publish a number it becomes a promise, and a promise that fails online fails louder than one that fails over the phone: it is in writing and the customer can see it at any hour.
There are three honest answers, and all three are defensible. Publish real stock, if you have one reliable source. Publish no number and confirm within hours. Or publish a band — in stock, low, made to order — which is what most serious sellers do when they do not want to lie.
What is not defensible is publishing a number you already know is wrong. And if the problem is that the system says you have it and you do not, the fix comes earlier: when inventory gets committed.
How you get paid is not a detail
In consumer retail, payment is a gateway and that is the end of it. Not here.
Your customers probably pay by transfer, or on thirty-day terms. If it is credit, the portal has to know who approves the limit, who checks the balance, and what happens when a customer with an outstanding balance wants to order again. That logic is half the project, and it rarely appears in the first quote, because whoever quotes stores does not carry it in their catalog.
If you plan to take cards, that is a commercial decision before a technical one: the fee eats margin that is thin in industrial. Decide it with a number, not out of habit.
The four things to settle first
None of them is software. All four can be written in an afternoon:
- Who you sell to online: open to the public, or account holders only. If that is undecided, no platform can guess it.
- What price can be published, and if there is no public price, which lists exist and where they live.
- One single source of stock a system can query without anyone copying it by hand.
- Who fills the online order. An order arriving through the web needs the same owner and the same date commitment as a plant order. If it lands in an inbox nobody checks until Tuesday, the portal will make you look worse than the phone did.
The fourth is the most forgotten and the one that kills the most projects.
What to build first if you choose a portal
A useful reorder portal is smaller than it sounds. The first version needs four things: the customer signs in and sees what he has bought, he can repeat a previous order in two clicks, he sees his price, and he can check the status of what he already ordered without calling.
That last one — seeing where his order stands — is usually worth more than the ordering itself, because it is what consumes your people’s calls today. Measure it first: count how many calls or emails a week exist only to ask “has it shipped?” That number is the project’s return, and almost nobody collects it.
Full catalog, search, comparisons and recommendations come later, if they come at all.
Where to start
Take last quarter’s orders and sort them into two piles: repeat customers and new ones. That ratio tells you which of the two projects is yours, and it is data you already have.
Then count a week’s worth of follow-up calls. With those two numbers you can decide whether a custom platform pays for itself or whether it is still too early.
And if this is one of several fronts open at once, the order matters: a portal publishing false stock does more damage than no portal at all.