How much does it cost to build a customer portal

Building a customer portal in 2026 typically costs $15,000 to $500,000+ depending on what it does. A read-only portal (orders, invoices, downloads) runs $15k to $40k. A transactional portal (manage subscriptions, support tickets, upgrades) runs $50k to $150k. A full B2B portal with multi-user roles, audit logs, SSO, and white-label runs $200k to $500k+.

The cost gap is real, and it comes from one decision: how much of your business logic the portal exposes to the customer. A page that shows past invoices is a SQL query and a PDF link. A page that lets an admin at a customer org invite teammates with scoped permissions is a small product inside your product. Same word, two completely different builds.

This post breaks down what you actually pay for, how to choose between building and buying, and where to spend the money where it earns its keep.

The three customer portal tiers (and what each one actually does)

Most "customer portal" projects fit into one of three buckets. Pricing each accurately requires knowing which bucket you are in before you write a single ticket.

Tier 1: Read-only portal ($15k to $40k)

A read-only portal lets a logged-in customer see their own data. No mutations beyond updating their profile and changing a password. Typical features:

View past orders, invoices, shipments
Download receipts, statements, contracts
See current subscription status
Update billing address, contact email, password
Maybe a basic ticket-submission form that posts into your existing helpdesk

This is the cheapest portal because the heavy lifting is read queries and PDF rendering. There is no permissioning beyond "is this row owned by this user." You can ship this in 3 to 6 weeks with one mid-level engineer and Clerk (or Auth0) for auth.

Tier 2: Transactional portal ($50k to $150k)

Now the customer can change things that cost you money or risk. Typical features:

Upgrade, downgrade, cancel subscription (with proration)
Manage payment methods, swap cards, retry failed charges
Open, update, close support tickets with attachments
Manage usage limits, API keys, webhooks
Self-serve refunds within policy bounds
Email and in-app notifications on state change

This is where engineering cost jumps roughly 4x. Every mutation needs validation, an audit trail, an idempotency key, and a careful rollback path. Stripe billing portal handles a chunk of the subscription work, but custom logic around upgrade rules, dunning, and team plans still lives in your code. You are looking at 8 to 16 weeks with one senior engineer plus a junior on test coverage and edge cases.

Tier 3: Full B2B portal ($200k to $500k+)

Now multiple humans at a customer org log into the same account, with different roles and audit trails. Typical features:

Multi-user accounts with org-level permissioning (admin, editor, billing-only, viewer)
SSO via SAML or OIDC (Okta, Azure AD, Google Workspace)
SCIM for automatic provisioning and deprovisioning
Audit log of every action (who did what, when, from where)
White-label theming, custom domain, custom email senders
Granular API key management, per-key scopes and quotas
Custom contract terms surfaced (volume tiers, SLA dashboards)
Approval workflows for high-impact actions

This is a real product. You are not building a "portal," you are building the customer-facing surface of an enterprise SaaS. Plan 4 to 9 months with a lead engineer architecting, a senior shipping the org and roles model, and a mid or junior on glue work. Most of the cost goes to multi-tenant data isolation, the audit log infrastructure, and SSO + SCIM testing across the four or five identity providers your customers actually use.

Cost breakdown by approach

The portal pillar has a wider cost range than almost any other software build, because the right approach depends entirely on whether the portal is your business or a feature next to your business. Here is the honest comparison:

Approach	Cost	Timeline	Pros	Cons
US full-time hire (senior)	$180k+/yr	8 to 14 weeks to ship Tier 2	Owns it long-term, full context	3-month hiring loop, equity, $30k+ in benefits, can quit
Dev agency (US/EU)	$80k to $250k fixed bid	12 to 20 weeks	Project management included, contractual	Scope creep is expensive, agency talent rotates, low ownership of edge cases
Freelancer (Upwork / Toptal)	$50 to $150/hr ($40k to $130k total)	8 to 16 weeks	Fast to start, no long-term commitment	Quality varies wildly, often disappears mid-project, you manage the loop
Salesforce Experience Cloud (formerly Community Cloud)	$25 to $250/user/month + $50k to $150k implementation	3 to 6 months	Enterprise-grade out of the box, SSO and audit log built-in	Locks you to the Salesforce data model, per-user pricing scales painfully, customization needs Salesforce devs ($150+/hr)
HubSpot Service Hub portal	$150 to $1,200/month + customization	2 to 8 weeks	Cheap to start, integrates with HubSpot CRM	Limited customization, not a true portal for product accounts, branding caps
Build on Clerk + Next.js + Supabase	$20k to $200k engineer time + ~$300/month infra	4 to 16 weeks	Full control, your data, modern stack	You own the maintenance, you build SSO from scratch (or pay Clerk B2B)
Cadence	$500 to $2,000/wk	48-hour trial, then week one ship	AI-native engineers by default, weekly billing, replace any week, no hiring loop	Less suited to large enterprise procurement workflows

The buy-it route (Salesforce, HubSpot) is genuinely the fastest path if your customer portal is a checkbox in a sales cycle, not a core product surface. If customers will live in it daily, build it. The lock-in math on Salesforce gets ugly above 200 portal users.

Feature-by-feature cost breakdown

Real numbers for the components you will actually wire together. We have shipped enough of these on Cadence (~340 customer-portal-adjacent bookings in the last 12 months) to have honest priors.

Authentication and identity ($0 to $25k engineer time)

Clerk: free up to 10k MAU, then ~$25/month + $0.02 per MAU. Drop-in for B2C. Their B2B add-on (orgs, invites, SSO) is ~$100/month + per-org pricing.
WorkOS: the SSO and SCIM specialist. Free until you have SSO customers, then ~$125 per SSO connection per month. Worth every cent if you sell to mid-market and up.
Auth0: $35/month for 1,000 MAU, scales steeply ($240/month at 5k MAU on the B2C plan, more for B2B). Still the default for many F500-adjacent sales.
Supabase Auth: free, built-in if you are already on Supabase. Email/password and OAuth are solid, SSO is paid plan only.
Build it yourself: roughly 2 to 4 weeks of senior engineer time ($3k to $8k on Cadence) for password + email + OAuth + sessions. Add 4 to 8 weeks ($6k to $16k) if you need SAML.

Recommendation: never build auth from scratch in 2026. The break-even against Clerk or WorkOS is somewhere north of 100k MAU.

Billing self-service ($0 to $15k)

Stripe Customer Portal: free with Stripe, drop-in, handles cards, invoices, subscription cancel/upgrade. ~1 day to integrate.
Stripe Billing API direct: 2 to 5 weeks to build a custom UI on top ($3k to $10k). Worth it if you need custom upgrade flows, team-plan logic, or non-standard proration.
Paddle, Lemon Squeezy: similar story, hosted portal options exist.

Multi-user roles and orgs ($10k to $80k)

This is where Tier 3 portals burn money. You need a roles model (RBAC or ABAC), org-scoped row-level security, an invite flow with magic-link emails, and an audit log. The Clerk and WorkOS B2B SDKs handle the schema and UI for orgs and basic roles. Custom permissions on top (per-resource, per-action) is 4 to 12 weeks of senior work.

Audit log ($5k to $30k)

A real audit log writes immutably, records actor and IP and action, and ships to S3 or a SIEM. Use AuditBoard, Cribl, or roll your own with append-only Postgres + S3 archive. Most teams underspec this at v1 and pay later when SOC 2 audit asks for 12 months of retention.

SSO and SCIM add-on ($10k to $40k if building, $1,500/month minimum if buying)

WorkOS SSO: $125 per active SSO connection per month, includes SAML and OIDC for every major IdP.
Building SAML from scratch: 4 to 8 weeks senior ($6k to $16k). Then you maintain it forever as IdPs change their behavior.
SCIM: 2 to 4 more weeks if you build, included with WorkOS Directory Sync at ~$125/connection/month.

A useful rule: if you have fewer than 5 enterprise customers on SSO, build cheap (or use WorkOS pass-through). If you have more than 20, the math on building still does not beat WorkOS unless you have a security engineer in-house.

White-label and custom domains ($5k to $25k)

Subdomain per customer, custom branding, custom email from-address. The DNS and TLS automation (Cloudflare for SaaS or Vercel Custom Domains API) is the real work. Allow 2 to 4 weeks of senior time.

Support tickets ($0 to $20k)

Embed Intercom, HelpScout, or Zendesk via their web SDK ($50 to $500/month + per-seat) instead of building. A native ticket system inside the portal is 4 to 8 weeks and rarely worth it unless support is your differentiator.

For more on this pattern of building only the differentiators, see our breakdown on the cost to build an admin dashboard, which shares most of the architectural decisions.

How to reduce customer portal cost without cutting corners

The biggest wins come from being honest about what you can hand off to SaaS and what is actually core to your product.

Use Stripe Customer Portal for billing UI. Building a card-management UI in 2026 is wasted engineering. The portal embed is two hours of work.
Use Clerk or WorkOS for auth and orgs. Saves 6 to 12 weeks on a Tier 3 build. The per-MAU pricing is a rounding error against engineer time.
Defer SSO until you have a paying customer asking for it. Do not pre-build SAML. WorkOS lets you turn it on in a day when the contract is on the table.
Audit log: pick the simplest pattern that survives SOC 2. Append-only Postgres table plus daily S3 archive plus a retention job. Do not build a search UI until someone asks.
Use AI-native engineers for the glue work. Half of a Tier 2 portal is CRUD scaffolding, form validation, and email templating, which is where modern Cursor + Claude Code workflows compress timelines by 2 to 4x compared to manual coding. Every engineer on Cadence is AI-native by default, vetted on Cursor and Claude Code fluency before they unlock bookings.
Stage the build. Ship Tier 1 in 4 weeks. Then upgrade to Tier 2 features as customers ask. Skipping straight to Tier 3 is the most common reason portal projects miss timelines.

This staging pattern is the same one we recommend for building a Next.js application end-to-end: ship the smallest valuable thing, then layer features against real usage data.

Build vs Buy vs Hire: how to actually decide

A useful framework:

If the portal is for a sales-cycle checkbox (procurement says "they need a customer portal"), buy HubSpot Service Hub or Salesforce Experience Cloud. The total cost will be lower over 24 months, and the build will not block your roadmap.
If customers will log in daily and the portal experience is part of your product story, build it. SaaS portals look like SaaS portals. Yours should feel like yours.
If you are building Tier 2 and have no in-house engineer, booking a senior on Cadence for 8 to 12 weeks lands you at $12k to $18k for the build, with no hiring loop or notice period. If the project lands and you want to keep them, you keep them. If something feels off after the 48-hour trial, you replace them at no cost.

For founders without an in-house engineer who need a clean estimate before committing, the fastest path is to scope the portal as Tier 1, Tier 2, or Tier 3 (using this post), then book a Cadence senior or lead for the first sprint. You will know inside two weeks whether the scope and the engineer are right.

The 48-hour read on Cadence: book a senior or lead engineer with a portal-shaped spec, get matched in 2 minutes, run the first 48 hours free. If they ship the Tier 1 scaffolding in week one, you have your answer. Start a booking on Cadence.

The fastest path from idea to portal

Three steps that work for almost every founder we have shipped portals with:

Decide your tier. Read-only, transactional, or full B2B. Write it on one line. Do not let scope drift into Tier 3 features during week one of a Tier 1 build.
Buy the commodity pieces. Clerk or WorkOS for auth, Stripe Customer Portal for billing, Intercom or HelpScout for support, Cloudflare for SaaS for custom domains. These are not where your product wins.
Build the differentiator with a senior engineer who can ship in week one. If you do not have one on staff, book a senior or lead engineer on Cadence and use the 48-hour trial to confirm scope before paying for a full week.

FAQ

How long does it take to build a customer portal?

A Tier 1 read-only portal takes 3 to 6 weeks with one mid-level engineer. A Tier 2 transactional portal takes 8 to 16 weeks with a senior. A Tier 3 full B2B portal takes 4 to 9 months with a lead plus 1 to 2 supporting engineers. AI-native workflows on tools like Cursor and Claude Code compress these by 30 to 50% versus manual coding on the same scope.

What tech stack should I use for a customer portal in 2026?

The default for greenfield portals is Next.js (App Router) on Vercel, Postgres on Supabase or Neon, Clerk or WorkOS for auth, Stripe for billing, Resend or Loops for transactional email, and Cloudflare for SaaS or Vercel Custom Domains for white-label. This stack ships a Tier 1 portal in days and scales to Tier 3 without a rewrite.

Should I build a customer portal or use Salesforce Experience Cloud?

Buy Salesforce if the portal is a procurement checkbox and your customers will use it monthly, not daily. Build it (with Clerk + Next.js + Stripe) if portal usage is core to your product experience or if you have more than 200 portal users (where Salesforce per-user pricing gets painful). HubSpot Service Hub is a cheaper middle ground if you are already on HubSpot.

What does SSO cost for a customer portal?

If you build SAML from scratch, expect 4 to 8 weeks of senior engineer time ($6k to $16k on Cadence) plus ongoing maintenance. If you use WorkOS, expect ~$125 per active SSO connection per month and a one-day integration. The break-even is around 10 enterprise SSO customers, where building plus maintaining slightly edges WorkOS, but most teams should buy.

Can a non-technical founder ship a customer portal solo?

A Tier 1 read-only portal, yes. Use a no-code or low-code stack: Softr, Glide, or Stacker on top of Airtable or Supabase. Expect $50 to $200/month and a week of setup. Anything in Tier 2 or Tier 3 needs at least one senior engineer because the surface area for billing, permissions, and audit bugs is too large for no-code to cover safely. A useful comparison is our breakdown on the cost to build a Shopify app, where the same buy-vs-build math applies.

All posts