External website; opens in a new tab.

Software & Apps

What must a customer portal protect before real members use it?

A convincing demo is not enough: current membership, server-side separation, recovery and customer-safe information need independent proof.

Software & Apps

Source review 21 November 2026

A portal can look complete while exposing the wrong customer’s records. The important questions sit behind the screens: how identity is verified, which organisation/project the person currently belongs to, what each operation is allowed to do and what happens when access is revoked. A selectable synthetic demo identity is not a production login system.

Separate identity from permission

Use a suitable supported identity service and secure session handling, with MFA and recovery appropriate to the users and risk. Identity answers who has authenticated; authorisation decides whether that person can perform this action on this organisation/project now. A remembered organisation ID in the browser is not sufficient permission.

Enforce membership and object scope on the server and database boundary as appropriate. Test a valid user attempting to read, edit or reference another customer’s project, document and evidence record. Hiding a navigation item does not protect an API. A copied link or old session must not preserve authority after removal from the project.

Show the right information

Customer views should contain deliberate safe projections, not a filtered dump of internal administration records. Internal costs, margins, unrelated customers, privileged notes and secrets should have no path into a customer response. An external accounting link is different from an authoritative claim that an invoice has been paid. State the source and freshness of a verified financial observation.

Progress should be tied to the accepted scope and evidence. Engineering verification, delivery acceptance and customer contractual acceptance are different facts. A manually adjusted forecast may be useful, but it must not silently become evidence of completed or accepted work. The same care applies to a mission status: an old “running” record is not proof a worker is currently alive.

Prove failure and recovery

Exercise revoked sessions, expired grants, password/account recovery, denied cross-customer operations, interrupted writes, duplicate events and restore. A clean unit suite is valuable but does not prove independent process identity, secret custody or a live provider configuration. Record limitations explicitly and keep real-member admission separate until those gates are demonstrated.

Give users understandable errors and an accessible route to report problems. Avoid including secrets, stack traces or another customer’s details in error messages and logs. Plan support ownership, backup keys, data retention and access removal before collecting live project information.

A public information site and a privileged portal do not need to share every runtime, database credential or session audience. The simplest proportionate architecture still preserves the necessary trust boundaries. Choose implementation and hosting from verified requirements, rather than adding infrastructure solely to make a diagram look secure.

Sources and review

Prepared by Yellow Brick Tech. General guidance; the source and test conditions for your own situation still matter.

Next review: 4 May 2027

Related services

How Yellow Brick Tech can help.

Next step

Need to apply this to a real situation?

Ask a question or choose the closest project route. A short conversation can often clarify the next step.