Business

The portal that stops the phone ringing

A customer portal pays off when there is a repeated question customers could answer themselves. To work it must show real, live data rather than a manually updated summary: otherwise it becomes another channel to staff. Start from a single piece of information — the one generating the most calls.

14 Jan 2026 · 8 min read · updated on 27 Aug 2026

In almost every company serving other companies there is a phone number that rings for one reason only: someone wants to know where things stand. The order, the case, the repair, the shipment. Whoever answers opens the business system, looks, and reports back.

It is work that produces nothing: it moves information that already exists from a database to a person. And it costs twice, because it interrupts someone who was doing something else.

First of all: count the calls

For one week, ask whoever answers to note two things: how many calls they get and which question is asked. Nothing else. By the end of the week you will have a short list where the top two or three questions cover the majority.

If at the top there is a question the customer could answer themselves — where it stands, how much I have spent, when it arrives, where my documents are — then a portal makes sense. If instead the top questions require judgement ("do you think it's worth doing now?"), a portal will remove no calls: those calls are your job.

A portal doesn't reduce calls because it exists. It reduces them if it answers, better than the phone, the question customers actually ask.The test we use before building one

The two decisions that determine everything

First: where the data comes from. A portal showing information someone in the office updates by hand removes no work: it moves it and doubles it. It has to read from the business system automatically and show what is there — even when that is uncomfortable. If the data isn't reliable, the problem isn't the portal: it is the data, and it must be fixed first.

Second: what each customer sees. A customer must see their things and only theirs. It sounds obvious and it is where hastily built portals go wrong: the customer filter lives in the code, someone writes a new function and forgets it, and for one afternoon one company sees another's documents. That is why isolation must be enforced by the database and not left to the good intentions of the code.

What to put in it, in order of value

The status of what is in progress, with the expected date and the last update. That is eighty per cent of the value.
Documents: invoices, delivery notes, certificates, job sheets. Downloadable, without having to ask.
History: what was done, when, at what price. It reduces disputes more than the first two combined.
New requests: a form that opens a case in the business system, instead of an email someone has to transcribe.
Notifications: telling people when something changes is what turns a portal from a place to visit into a service working for you.

The first two are built in a few weeks if the business system exposes the data. The rest is added later, if needed — and often it isn't.

Three ways to make it fail

Asking for a complicated sign-up. If getting in takes three steps and a password nobody remembers, the customer phones. Access must be faster than a call, or it loses from the start.
Showing stale data. Once is enough: if a customer sees a wrong status they will never come back, and will phone even when the portal is right.
Telling nobody. The portal belongs in the email signature, on invoices, and should be mentioned on every call — "I'll send it through the portal, so you always find it there". Habits are built, not announced once.

The return, measured

Measure the calls again after two months, with the same method. It is the only way to know whether it worked — and to notice something that often happens: calls drop, but the remaining ones are harder, because the portal took the easy ones away.

That is a good sign, even if it doesn't feel like one: it means whoever answers is spending their time on things that deserve a person. It is also the moment to review who answers and how, because the work has changed.

If you don't know how many calls you get and what for, start there: one week and a sheet of paper. The same measurement will tell you whether a portal is needed, what it must show and — just as usefully — whether it isn't needed at all.

What does a customer portal cost?

Almost entirely on what the business system exposes. If the data is easy to read, a first version with status and documents is measured in weeks. If the system is closed, the expensive part is building the connection, not the portal.

Will customers use it?

They use it if it solves an annoyance of theirs and if getting in is faster than phoning. You already have the best signal: if someone asks you the same question every week, that person will use the portal on day one.

Is giving access to our data risky?

The risk isn't giving it, it is giving it badly. With personal accounts, database-enforced isolation and a record of who saw what, a portal is safer than an attachment sent by email — which is the route you are using now.

App or website?

Almost always a website that works well on a phone. An app makes sense when you need offline operation or heavy use of camera and notifications: for checking status and documents it is an extra cost and a barrier to access.