The short answer
- Push beats pull for anything time-sensitive — a portal is a place people do not visit
- Behind a login: order history, artwork, reorders, invoices, store administration
- Never behind a login: order status, proofs awaiting approval, delivery dates
- Portals earn their keep with repeat accounts, not one-off buyers
- Measure the calls it removed, not the logins it recorded
Customer portals are a standard feature and a standard disappointment. A shop adds one expecting to reduce phone calls, the calls continue, and the conclusion is that customers do not want self-service.
The real reason is narrower: a portal only helps for things a customer is willing to go and look for. Almost nothing time-sensitive qualifies.
Push versus pull
A portal is pull — the customer must remember it exists, find the link, recall a password, log in and navigate. That is a lot of friction in front of a question that takes ten words to answer.
Push is you sending the information without being asked. For anything urgent, push wins every time, which is why four proactive messages remove more calls than any portal ever will.
| Information | Push or pull |
|---|---|
| Order confirmed | Push |
| Proof ready for approval | Push |
| In production / shipped | Push |
| Order history | Pull |
| Past artwork | Pull |
| Invoices and statements | Pull |
| Placing a reorder | Pull |
The split is essentially by urgency. Anything the customer needs now should arrive; anything they might want someday can wait behind a login.
A portal is where customers go when they already trust you to have the answer. It is not where they go when they are worried.
What genuinely belongs behind a login
Order history with enough detail to recognise. Not a list of numbers — the garment, the design, the quantity and the date, so a customer can point at the one they mean.
Their artwork. Customers lose their own logos constantly. Being able to download the vector file you prepared is genuinely useful to them and costs you nothing.
Reordering. The highest-value thing a portal does. Repeating a past order should take about thirty seconds, which is what makes the second order easier than the first.
Invoices and statements. Accounts departments ask for copies constantly, and self-service removes an entire category of small interruption.
Store administration for anyone running a company store — orders placed, by whom, against which budget.
What should never be behind a login
Anything a worried customer wants. If someone is checking whether their order will arrive in time, making them log in to find out is the moment they decide to phone instead — and now you have a portal and a phone call.
Push these, always
- Order confirmation, with the delivery date
- Proof ready, with a one-click approve that needs no login
- In production, date restated
- Shipped, with tracking
- Anything that has changed — a delay, a substitution, a short count
Proof approval deserves emphasis. A proof requiring a login to approve is a proof that will be approved slowly, and approval delay is a real production problem. Send it as a link that works without an account, and record the approval against the job regardless.
Company store administrators
The one group who genuinely will log in, because it is part of their job. A facilities manager or office administrator running a staff store needs a view you should design deliberately rather than leave as a by-product.
What an administrator needs
- Every order placed on their store, by person
- Spend to date against any budget or allowance
- The ability to approve or block an order before it is produced
- Delivery status across all open orders
- Their own artwork and approved product list
- Nothing belonging to any other customer
That last line is not a formality. Scoping is the one part of a portal that must be right on day one, and it is worth testing explicitly rather than assuming. Setting it up properly at launch is part of how our B2B stores are configured, and it is far easier than retrofitting later.
Getting people to use it
Not by announcing it. Portals get adopted when the link to them arrives at the moment it is useful, embedded in something the customer was already reading.
Where the link should appear
- In the order confirmation — "see this order and your past orders"
- In the shipped notification — "reorder this in one click"
- On every invoice
- In the reorder prompt you send on their cycle
- Never as a standalone "we have a new portal" email
The reorder link in the shipped notification is the highest-converting placement, because the customer is thinking about the order at exactly that moment. A generic announcement is the lowest, because it asks somebody to do something for no immediate reason.
Measuring whether it worked
Not by logins. A login is not a benefit; it is a step somebody took. What matters is whether it removed work.
| Measure | Tells you |
|---|---|
| Requests for invoice copies | Whether self-service billing is working |
| Requests for past artwork | Whether artwork access is discoverable |
| Reorders placed without contacting you | The main prize |
| Status calls | Should be flat — push fixes these, not the portal |
| Store admin questions | Whether the admin view is adequate |
The status row is the one to watch for a misleading conclusion. If status calls have not fallen, that is not a portal failure — status was never a portal problem. Fixing it means sending the four updates, and judging the portal on that measure will lead you to abandon something that was working on the things it can actually affect. The rest of the software archive assumes you push what is urgent and store what is not.
Passwords are where portals die
The commonest reason a portal goes unused is not the design. It is that the customer cannot get in, tries once, and never tries again — and you never learn it happened.
Occasional buyers will not remember a password set eight months ago. Assume that, and design around it rather than treating it as their failing: a magic-link sign-in, or a link in an email that authenticates on its own, removes the barrier entirely for the people most likely to hit it.
Where a password is genuinely needed — a company store administrator with approval rights — make the reset obvious and instant. A reset that requires contacting you has converted a self-service feature into a support ticket, which is the opposite of the point.


