← Back to Blog
Web AppAWSEngineering

Customer Portals: The Build SMBs Underestimate

Patrick "Wiloak" Wilson · November 5, 2026 · 5 min read

A customer portal looks like a weekend project. Log in, see your data, click some buttons. Most SMBs scope it that way and then watch the timeline triple. The screens are not the work. The work is everything behind the screens.

What the demo hides

The demo is a list and a form. That part is real, and it is maybe 20% of the build. The other 80% is the stuff nobody shows in a wireframe.

Auth is the first wall. You need login, sessions, password reset, and probably SSO before long. Then you need to decide what each user can see. That is authorization, and it is not the same problem as login.

Multi-tenancy is the second wall. Two customers cannot see each other's records. Ever. One leaky query and you have a breach, not a bug.

Where the time actually goes

Permissions are the quiet budget killer. An admin sees everything. A manager sees their team. A regular user sees their own rows. Now add a customer who only sees their own account. Each rule touches every query you write.

Get this wrong and the failure is silent. The page renders. The data is just the wrong tenant's data. You will not see it in a happy-path demo.

Audit logging is the next thing people skip. Who changed this record, and when? Customers ask the day after they need it, not before. Bolting it on later means touching every write path twice.

Then there is the boring list that always runs long. Email that does not land in spam. File uploads that reject the 200MB PDF gracefully. Pagination that survives 50,000 rows. Search that returns in under a second. Rate limits so one script does not take the portal down.

None of these are hard alone. Together they are the project.

The failure mode

The classic SMB portal dies at the 70% mark. Screens look done. The demo works. Everyone thinks they are almost shipped.

Then real users arrive. The forgot-password flow breaks. An admin sees a customer's billing. Search times out at scale. The last 30% holds all the security and all the edge cases, and it is the part that was never scoped.

The second failure mode is the freelancer who ships it and vanishes. The portal works until it doesn't. Now you own code you cannot read and a person you cannot reach.

How we scope it

We treat the portal as three systems, not one. Identity, authorization, and the app. We design the tenancy model and the permission rules before we write a screen, because they are the expensive part to change later.

That work goes faster when the person scoping it has done it before. Our team is two people, one of them an ex-AWS Federal Solutions Architect, and we have shipped 17 production products, 10 of them in our first 7 months. We hold 18 cloud certifications, 13 of them AWS. The portal sits on the same patterns we use everywhere: managed auth, row-level isolation, real audit trails.

Pricing is fixed, not hourly. A Web App is a fixed price in the $6K-$25K range, delivered in 3-5+ months. You get the scope written down before we start, so the last 30% is in the plan instead of in a surprise invoice.

You own the code. We hand it over and exit. No retainer to keep the lights on, no vendor you cannot fire.

Before you sign anyone

Ask three questions of whoever quotes your portal. How will you isolate tenants. How will permissions work for each role. Where does audit logging live. If the answers are vague, the quote is vague, and the overrun is already baked in.

A portal is one of the most useful things an SMB can ship. It cuts support tickets, gives customers self-service, and replaces the email-and-spreadsheet shuffle. It is worth building right the first time.

If you are scoping one and want a straight read on the real work, book a free 30-minute call. We will tell you where the time actually goes.

Further reading

Want to Talk About Your Project?

We write about what we do every day. If any of this resonates, let's chat.

Request a free 30-min call See services and prices