TL;DR

  • SaaS platforms (Memberful, Circle, Kajabi, Skool) cost $1,200-$7,200/year in subscription fees, but transaction cuts and feature ceilings stack up fast past 2,000 members.
  • A custom membership site built on Next.js 16 + Supabase + Stripe Billing runs $24K-$80K for an MVP, $80K-$179K for a full platform, with monthly operating costs under $500 up to 25K members.
  • At 5,000+ paying members, custom build TCO drops below SaaS within 24-36 months and the gap widens every year after.
  • If you don't have product-market fit yet, use SaaS. If you do, run the 5-year math before you re-sign.
  • The most expensive mistake I see is not picking the wrong option at the start. It's staying on the wrong option two years too long.

Last quarter a professional education company asked me to audit their Kajabi setup. They had 6,800 paying members at $49/month, which comes to roughly $333K in annual recurring revenue. Kajabi was taking $0 in transaction fees on their Growth plan, but the $199/month subscription, the $11K they'd spent on Zapier workarounds, and the custom Bubble app they'd built to handle certification workflows added up to $27K/year in platform costs alone. That's 8.1% of gross revenue going to infrastructure duct tape. We moved them to a custom stack for $74K total build cost, and their annual operating cost dropped to $4,200. The build paid for itself in 11 months.

That story is not universal. I've also watched three founders torch $40K rebuilding what Memberful would have done for $99/month and zero engineering hours. The question isn't which option is "better." The question is which option is correct for your stage, your member count, and where you're headed in the next three years.

I've been building membership platforms for 12 years. Here's what the numbers actually look like in 2026.

What Counts as a "Membership Site"

A membership website is any site where users pay for recurring access to content, community, tools, or services behind an authentication wall. That's it. The definition is simple. The implementations are wildly different.

In practice, I see three archetypes show up repeatedly:

Archetype 1: Content Paywall. A creator or publication gates articles, videos, podcasts, or newsletters behind a monthly subscription. Think Substack but self-hosted, or a niche media brand. Features needed: auth, Stripe checkout, content gating, email delivery. This is the simplest version.

Archetype 2: Course + Community Platform. An educator or professional organization sells access to structured learning (video modules, quizzes, certificates) bundled with a community forum or discussion space. Features needed: everything from Archetype 1, plus progress tracking, drip content, community threads, possibly live events. Mighty Networks and Circle live here.

Archetype 3: Full Membership Product. The membership IS the product. Custom dashboards, API access, usage-based billing tiers, multi-seat team plans, third-party integrations, admin analytics, possibly white-labeling. This is where SaaS platforms hit the wall and where custom builds earn their cost back.

Your archetype determines your budget. A content paywall has no business spending $80K on a custom build. A membership product with 10,000 users has no business staying on Kajabi. The rest of this guide will make the boundaries clear.

SaaS Platforms: Real Cost Breakdown

Here's what the major platforms actually charge in 2026, including the numbers they don't put in the hero section of their pricing page.

Platform Monthly Fee Transaction Cut Member Limit (Plan) What Hits the Wall
MemberPress (WP plugin) $33/mo (billed annually) 0% (Stripe fees only) Unlimited WordPress performance, no native community, plugin conflicts
MemberSpace $29/mo 3% (Starter) Unlimited Limited design control, 3% adds up fast
Memberful $49/mo (Pro) 4.9% + Stripe fees Unlimited No community, no course tools, that 4.9% is brutal at scale
Mighty Networks $41/mo (Community) 0% 1 space only Need $99/mo+ for courses, no custom CSS below Business plan
Circle $89/mo (Professional) 0% Unlimited members No native payments on lower tiers, limited course features
Skool $99/mo 0% Unlimited No custom domain on free, very rigid structure, no API
Patreon $0/mo 8% (Pro) or 12% (Premium) Unlimited 8-12% of revenue, limited site customization, you don't own the audience
Kajabi $149/mo (Growth) 0% 10,000 members Hard cap at 10K, slow page loads, limited API, expensive to upgrade
Teachable $99/mo (Pro) 0% Unlimited Course-first, weak community, limited membership flexibility
Thinkific $74/mo (Basic) 0% Unlimited Course-only, no real membership tier logic, limited community
Podia $75/mo (Shaker) 0% Unlimited Basic feature set, no real API, design constraints
Substack $0/mo 10% of paid sub revenue Unlimited 10% forever, minimal customization, no ownership

The Hidden Math on Transaction Fees

Transaction fees are where the real cost lives. Let's say you have 2,000 paying members at $29/month. That's $58,000/month in gross revenue.

  • Stripe processing (everyone pays this): 2.9% + $0.30 = ~$2,012/month
  • Memberful's 4.9% cut on top of Stripe: $2,842/month additional
  • Patreon Pro's 8%: $4,640/month additional
  • Substack's 10%: $5,800/month additional

On Memberful, you'd pay $4,854/month in combined fees on $58K revenue. That's $58,248/year just in transaction costs, plus the $588 annual subscription. At that scale, a custom build with only Stripe's base fees saves you over $34,000 per year.

What "Hits the Wall" Means in Practice

Every SaaS platform has a ceiling. It's not always a member count limit. More often it's one of these:

  • No API access (Skool, lower Kajabi tiers), so you can't connect your CRM, analytics, or internal tools.
  • Rigid tier logic. You want to offer annual + monthly + team plans + usage-based add-ons. The platform supports two tiers and a coupon code.
  • Design constraints. Your brand needs a custom member portal, not a template that looks like every other Circle community.
  • Performance degradation. MemberPress on WordPress with 15 plugins starts crawling at 5K+ concurrent sessions. I've seen Time to First Byte hit 4.2 seconds.
  • Data portability. Try exporting your full member history, content relationships, and community threads from Mighty Networks. You'll get a CSV with email addresses if you're lucky.

Custom Build: Real Cost Breakdown

When I say "custom build" in 2026, I'm talking about a specific production stack, not a theoretical architecture. Here's what we deploy at socialanimal.dev for SaaS and membership projects:

Core Stack:

  • Next.js 16 on Vercel (app router, server components, server actions)
  • Supabase for auth, database (Postgres), row-level security, real-time subscriptions
  • Stripe Billing with Subscription Items and Meters for usage-based pricing
  • Resend for transactional and marketing email
  • PostHog for product analytics and feature flags

This stack handles Archetypes 1-3 with zero vendor lock-in on your content or member data. You own the database. You own the code. You can move hosting providers in an afternoon.

Build Phases and Cost Ranges

Phase What's Included Timeline Cost Range
MVP (Archetype 1-2) Auth, Stripe checkout, content gating, member dashboard, basic admin, email notifications, responsive design 6-10 weeks $24,000-$80,000
Full Platform (Archetype 3) Everything in MVP + community features, course/progress engine, team plans, usage-based billing, admin analytics, API layer, migration tooling 12-20 weeks $80,000-$179,000
Enterprise / Multi-Tenant White-labeling, per-tenant config, SSO/SAML, audit logs, SLA-grade infra, advanced permissions 20-32 weeks $150,000-$320,000

These numbers assume a senior Next.js development team (2-3 engineers, 1 designer, 1 PM). If you're hiring an agency like ours, expect the higher end. If you have strong in-house engineers and just need architecture guidance, you can hit the lower end.

In-House vs Agency Math

In-house team (2 senior engineers + 1 designer):

  • Fully loaded cost: ~$45K-$65K/month in the US (salary, benefits, tools)
  • MVP in 8 weeks: $90K-$130K in labor
  • But you keep the team and maintain velocity post-launch

Agency build:

  • Fixed-bid MVP: $24K-$80K (depending on scope)
  • You don't carry payroll after launch
  • Ongoing maintenance retainer: $2K-$6K/month typical

Freelancer patchwork (I don't recommend this for membership builds):

  • Looks cheap at $60-$120/hour
  • Coordination overhead eats 30-40% of productive hours
  • No one owns the architecture, so you accumulate tech debt from day one

Monthly Operating Costs (Custom Build)

Service Free Tier Ceiling Typical Cost at 5K Members Cost at 25K Members
Vercel (Pro) 100GB bandwidth $20/mo $20-$150/mo
Supabase (Pro) 500MB DB, 1GB storage $25/mo $75-$200/mo
Stripe Billing N/A 2.9% + $0.30/txn 2.9% + $0.30/txn (volume discounts at $80K+/mo)
Resend 3,000 emails/mo $20/mo $80/mo
PostHog 1M events/mo $0 $0-$50/mo
Domain + DNS N/A $15/year $15/year
Total (excl. Stripe) -- ~$66/mo ~$350-$480/mo

Compare that to Kajabi at $199/month with a 10K member cap, or Circle at $89/month with no course tools. The infrastructure cost for a custom build is trivially low. What you're paying for is the upfront engineering.

5-Year TCO Comparison

This is the table that changes minds. I'm modeling a membership site charging $29/month per member, with 5% monthly churn (industry median for content memberships), growing from launch to the stated member count by month 18 and holding.

Cost Category SaaS (1K Members) SaaS (5K Members) SaaS (25K Members) Custom (1K Members) Custom (5K Members) Custom (25K Members)
Platform/Build $9,000 $18,000 $36,000 $36,000 $74,000 $145,000
Transaction Fees (5yr) $12,400 $62,000 $310,000 $6,100 $30,500 $152,500
SaaS Transaction Cut (5yr) $6,200 $31,000 $155,000 $0 $0 $0
Hosting/Infra (5yr) Included Included Included $4,000 $4,800 $24,000
Maintenance/Updates (5yr) $3,000 $8,000 $15,000 $18,000 $36,000 $72,000
Plugin/Workaround Tax (5yr) $2,400 $12,000 $48,000 $0 $0 $0
5-Year TCO $33,000 $131,000 $564,000 $64,100 $145,300 $393,500

Key takeaways from this table:

At 1,000 members, SaaS wins by $31K over five years. Custom doesn't make financial sense unless you have specific product requirements that SaaS can't serve.

At 5,000 members, it's nearly a wash. Custom is $14K more expensive, but you own the codebase, the data, and the exit value. If you're planning to sell the business, the custom asset adds 0.5-1.5x to your ARR multiple.

At 25,000 members, custom saves $170K+ over five years. The SaaS transaction cuts and workaround tax become enormous. The "Plugin/Workaround Tax" line is real. It's the Zapier plans, the custom Bubble apps, the developer you hire to hack around platform limitations, the landing page tool because the built-in one is insufficient.

The Hidden Cost Nobody Models: Migration

When a SaaS platform hits the ceiling, you migrate. Migration is not free.

Typical migration costs from my project history:

  • Mighty Networks to custom: $18K-$35K (community thread data is especially painful)
  • Kajabi to custom: $22K-$45K (course structures, student progress data, email sequences)
  • MemberPress/WordPress to Next.js: $15K-$40K (we've done enough of these to build a dedicated migration practice)
  • Teachable to custom: $20K-$38K (completion data, certificate records, student accounts)

These migration costs should be factored into the SaaS TCO, because if you're successful, you WILL outgrow the platform. The question is when, not if. You can run rough numbers through our migration ROI calculator to see where your break-even sits.

When SaaS Wins (Honest)

I build custom membership platforms for a living. It would be easy for me to say "always go custom." That would be dishonest and bad advice. SaaS wins in these situations:

1. You don't have product-market fit yet. If you're testing whether people will pay $19/month for your content, spend $99/month on Circle or Memberful and find out in 30 days. Do not spend $40K on a custom build to test a hypothesis.

2. Your member count is under 2,000 and you have no custom logic requirements. At this scale, even platforms with transaction fees are cheaper than the interest on the capital you'd spend on a custom build.

3. Simple paywall, no community, no courses. If all you need is gated articles and a Stripe checkout, Memberful or Ghost Pro will do this beautifully for under $100/month.

4. You're a solo creator with no engineering support. MemberPress on WordPress or Podia can be set up in a weekend. A custom build requires ongoing engineering attention. If you don't have access to engineers (or budget for a maintenance retainer), SaaS is the correct choice.

5. Speed to market matters more than unit economics. A Skool community can be live tonight. A custom platform takes 6-10 weeks minimum. If you're launching alongside a book release or a course presale, time-to-market beats optimization.

When Custom Wins (The Math)

Custom makes financial sense when the ongoing SaaS tax exceeds the amortized build cost. Here are the triggers I watch for:

Past 5,000 paying members. At $29/month average revenue per member and Memberful's 4.9% cut, you're losing $8,500/month to the platform on top of Stripe fees. That's $102K/year. A full custom build pays for itself in under 18 months.

Custom membership tiers and billing logic. When you need annual + monthly + team seats + usage-based add-ons + promotional pricing + upgrade/downgrade proration, Stripe Billing with Subscription Items and Meters handles this natively. SaaS platforms give you 3-5 fixed tiers and a coupon field.

Third-party integrations are critical. Your membership needs to sync with Salesforce, trigger webhooks to internal tools, feed data to a mobile app, or connect to an LMS. SaaS APIs (where they exist) are limited and rate-throttled. With a custom build on Supabase, you have direct database access and can build any integration.

Brand is the product. If your membership experience IS your brand differentiator (think: a premium professional community, a financial research platform, a health coaching program), you cannot afford to look like every other Circle or Kajabi site. We cover this in depth in our writeup on the best WordPress alternatives in 2026.

You're building toward an exit. Acquirers pay more for proprietary technology. A membership business on Kajabi is valued at 2-4x ARR. The same business on a custom platform with owned infrastructure, clean data models, and an API layer commands 4-7x ARR. On $500K ARR, that's the difference between a $1.5M and a $3.5M exit.

The Migration Decision

Migration is the most stressful project I take on, because there's a live business with paying members who expect zero downtime and no broken experiences. Here's how I approach it.

When to migrate:

  • Your SaaS platform costs exceed 6-8% of gross revenue (all-in, including workarounds)
  • You've hit a hard feature wall that the platform won't address in their roadmap
  • Page load times exceed 3 seconds and you can't fix it
  • You need data portability for compliance (GDPR, SOC 2) and the platform won't export cleanly
  • You're preparing for a fundraise or acquisition and need owned infrastructure

What breaks during migration:

  • Member sessions. Password resets are almost always required. Plan for a "Welcome to our new platform" email flow with magic link re-authentication.
  • Billing continuity. If the old platform managed Stripe Connect, you may need to migrate Stripe subscriptions to your own Stripe account. This requires manual intervention or Stripe's subscription migration API.
  • Content URLs. Every URL that changes costs you SEO authority. Map redirects obsessively. I keep a spreadsheet of every URL on the old platform and its new equivalent.
  • Community history. Forum threads, comments, DMs. Some platforms export these, many don't. Mighty Networks, in my experience, is particularly difficult for community data export.

A real example: Last year we migrated a 3,200-member professional development community from WordPress + MemberPress + BuddyBoss to a custom Next.js + Supabase stack. The project took 14 weeks, cost $62K, and reduced their monthly operating cost from $2,100/month (hosting, plugins, patch management, performance firefighting) to $180/month. Monthly churn dropped from 7.2% to 4.8% after migration because the new member experience was faster, cleaner, and had fewer bugs. At their revenue level, the 2.4-point churn reduction was worth $14K/month in retained revenue. The build paid for itself in under 5 months.

What I'd Pick (My Honest Recommendation by Stage)

Testing an idea (0-500 members): Memberful if content-only, Circle if community-first, Skool if you want speed and simplicity. Don't spend more than $100/month. Don't hire an engineer. Validate first.

Early traction (500-2,000 members): Stay on SaaS but start watching your costs. If you're on Patreon or Substack, migrate to Memberful or Circle to reduce your transaction cut. Start planning your data export strategy, because you'll need it.

Growth stage (2,000-5,000 members): This is the decision zone. Run the 5-year TCO math with your actual numbers. If you have custom feature needs or your SaaS workarounds exceed $500/month, start scoping a custom build. Talk to an agency (like ours) or start hiring.

Scale (5,000+ members): Go custom. The math almost always works. Every month you delay, you're paying a SaaS tax that could be funding product development instead. Build on Next.js 16 + Supabase + Stripe Billing. Own your stack. Own your data. Own your exit multiple.

Already stuck on WordPress with 3K+ members and performance problems: We've done this migration enough times that we wrote a full guide to WordPress-to-Next.js migration. The short version: yes, it's worth it, and no, you don't have to lose your SEO rankings if you handle redirects correctly.

FAQ

How much does it cost to build a membership website from scratch in 2026?

An MVP membership site on a modern stack (Next.js 16, Supabase, Stripe Billing) costs $24,000-$80,000 through an agency, depending on scope. If you need community features, course tools, team billing, and admin analytics, expect $80,000-$179,000. Enterprise multi-tenant platforms run $150,000-$320,000. Monthly operating costs (hosting, email, analytics) are typically $66-$480/month depending on member count. These figures assume a US-based senior engineering team. Offshore teams can reduce build costs by 40-60%, but in my experience, the architecture and UX quality tradeoffs increase long-term maintenance costs.

What's the cheapest way to launch a membership site?

Skool at $99/month or Memberful at $49/month are the lowest-friction starting points. You can be live in a single day. If you already have a WordPress site, MemberPress at $33/month (billed annually) adds membership gating to your existing content. Ghost Pro at $25/month works well for newsletter-style memberships. Total Year 1 cost for a basic SaaS setup: $600-$3,000 plus Stripe transaction fees. Don't over-engineer at this stage. Validate that people will pay before you invest in infrastructure.

How do SaaS membership platform transaction fees compare?

Stripe's base processing fee (2.9% + $0.30 per transaction) applies to every platform. On top of that, Patreon takes 8-12%, Substack takes 10%, Memberful takes 4.9%, and MemberSpace takes 3% on their Starter plan. Circle, Kajabi, Skool, and MemberPress take 0% beyond Stripe's cut. At 5,000 members paying $29/month, Memberful's 4.9% adds $8,500/month in fees. Over five years, that's over $500K in combined transaction costs. Platforms with 0% additional transaction fees shift costs into higher monthly subscriptions and feature-tier limitations instead.

When should I migrate from a SaaS platform to a custom build?

The most reliable trigger is when your all-in platform costs (subscription + transaction fees + workaround tools like Zapier, custom landing pages, external community tools) exceed 6-8% of gross revenue. Other signals: you've hit the platform's member or feature cap, page load times exceed 3 seconds, you need integrations the platform doesn't support, or you're preparing for an acquisition. Migration typically costs $15K-$45K depending on the source platform and data complexity. Run your specific numbers through a migration ROI calculator before committing.

Is WordPress with MemberPress still viable in 2026?

For small membership sites under 1,000-2,000 members, yes. MemberPress is mature, well-documented, and handles basic membership gating effectively. The problems start at scale: WordPress plugin conflicts, performance degradation under concurrent sessions, security patch management, and the accumulated weight of 8-15 plugins that most MemberPress sites end up running. If you're currently on WordPress + MemberPress and experiencing performance issues, a migration to a modern stack (Next.js + Supabase) typically reduces page load times by 60-80% and eliminates the plugin maintenance burden. We've documented the full process in our WordPress to Next.js migration guide.

What tech stack should I use for a custom membership site?

In 2026, I'd recommend Next.js 16 on Vercel for the frontend and API layer, Supabase for authentication, database (Postgres with row-level security), and real-time features, Stripe Billing with Subscription Items for payment management, Resend for transactional email, and PostHog for analytics. This stack gives you type safety end-to-end, edge rendering for performance, built-in auth with social providers, and direct Postgres access for any integration you need. Monthly infrastructure costs stay under $500 even at 25,000 members. The stack is also portable. Nothing here locks you to a single vendor.

How does a custom membership site affect business valuation?

SaaS-hosted membership businesses typically sell at 2-4x annual recurring revenue. Businesses with proprietary custom platforms, owned data infrastructure, and clean APIs command 4-7x ARR. The difference comes from reduced platform risk (no dependency on a third-party vendor's pricing changes or shutdowns), defensibility (your member experience can't be replicated by buying the same SaaS subscription), and data quality (acquirers value clean, portable datasets). On a $500K ARR membership business, the difference between 3x and 6x is $1.5 million in exit value. The custom build cost is a rounding error at that scale.

How long does it take to build a custom membership platform?

An MVP (auth, billing, content gating, member dashboard, admin panel) takes 6-10 weeks with a focused team of 2-3 engineers and a designer. A full platform with community features, course tools, team billing, usage-based pricing, and integrations takes 12-20 weeks. Enterprise multi-tenant builds run 20-32 weeks. These timelines assume clear requirements and an experienced team. If you're still figuring out what features you need, add 2-4 weeks for a discovery and scoping phase. I always recommend starting with the MVP scope and layering features based on real member feedback rather than building everything before launch.