Short answer: Build client portals from a template with a fixed structure — status summary, deliverables, timeline, shared documents, and contacts — generated per client from a database. Share individual pages, never the parent workspace, and audit permissions on a schedule. The maintenance burden, not the build, is what determines whether portals survive.
Two jobs: reducing the number of times a client has to ask where things are, and creating a single agreed record of what was decided. If it doesn't do both, it's a shared folder with extra steps.
That framing determines the content. A portal should answer the questions clients actually ask — what's the status, what do you need from me, where's that document, when is the next thing — rather than showcasing project management.
One page per client, generated from a template, containing in order:
Notion sharing is inherited, so a mistake here exposes more than the page in question. The rules:
Client portals typically contain personal data, so they belong in your record of processing activities, with retention decided in advance rather than by accident.
Portals die of staleness. A portal showing last month's status is worse than none, because it teaches the client not to look. Three defences:
If clients need to interact with structured data — submitting records, filtering large datasets, uploading in volume — an Airtable interface or a purpose-built portal fits better. Notion excels at readable, document-shaped client communication and is mediocre at client data entry.
Guests need a free account. Public links avoid that but remove access control, which is rarely the right trade for client work.
With generation and automation in place, twenty to thirty. Fully manual, about five before they start going stale.
Notion's branding options are limited on standard plans. If brand presentation matters commercially, look at a dedicated portal tool.
We build client-facing systems that stay current without becoming a job. See examples.