Payload vs Strapi in 2026: Next.js, Drizzle RLS & CMS TCO
Payload vs Strapi in 2026, with Directus and Sanity: the MIT license after Payload joined Figma, the Next.js Local API overrideAccess default, PostgreSQL row-level security, and a modelled 36-month TCO for 35 seats.

On this page

Payload vs Strapi comes down to where the CMS runs. Payload is an open-source (MIT) TypeScript headless CMS that installs inside your Next.js app and stores content in Postgres or MongoDB. Strapi is an open-source Node.js headless CMS that runs as its own server and exposes REST and GraphQL APIs. Most teams weighing Strapi vs Payload also shortlist Directus and Sanity, so this headless CMS comparison covers all four.
Our CMS vs DXP guide covers the earlier question: whether you need a $500k+ monolithic Digital Experience Platform suite (Adobe Experience Manager, Sitecore, Optimizely) or a composable MACH stack. This guide starts after you've picked headless. Four things separate the options from there: Next.js runtime topology, MIT versus BSL and Enterprise SSO paywalls, Drizzle ORM relational control, and multi-tenant database security.
At Idealogic, our JavaScript development, custom software development, SaaS development, and progressive web app development squads ship content and portal platforms across fintech, real estate and proptech, and social and creator economy products. You get three things here: a 2026 architecture comparison (Payload after Figma, Strapi v5's Document Service), a production PostgreSQL FORCE RLS guard around the Payload v3 Local API that closes the overrideAccess default, and a 36-month TCO model for 35 seats across 12 localized B2B brand portals (40M monthly views).
The short version
- Payload v3 fits Next.js TypeScript teams. Payload CMS (MIT on GitHub) mounts inside Next.js
/appand queries Postgres in-process through Drizzle ORM, with no HTTP hop. It stayed open source under the MIT license when Payload joined Figma in June 2025. - Strapi v5 (MIT core plus EE) has better APIs than v4: it replaced Entity Service with the Document Service API. It still needs a separate Koa server, keeps Audit Logs and Review Workflows on Strapi CMS Enterprise, and sells SSO with Enterprise or as a paid Growth add-on.
- Directus and Sanity both get expensive at mid-market size. Directus wraps SQL cleanly, but free production use stops at $5,000,000 in total annual finances. Sanity ($15 per seat per month on Growth) pairs a hosted Content Lake with GROQ, and you need Sanity Enterprise for SAML SSO and audit trails.
- At 35 seats and 12 portals (40M views a month), owned Payload v3 comes out $51.0k to $89.7k cheaper in our model. Over 36 months, owned Payload v3 with Postgres FORCE RLS and WorkOS SSO totals $87,900 at CEE rates. The modelled quotes for the others are $138,900 for Directus Commercial, $172,500 for Strapi v5 Enterprise, and $177,600 for Sanity Enterprise.
Payload vs Strapi vs Directus vs Sanity: 2026 architecture and licensing
Feature checklists won't tell you much here. Comparing Payload CMS vs Sanity or Strapi vs Payload by ticking boxes misses why two 35-seat content stacks end up about $90,000 apart over three years, and why one deploys as a single service while the other needs two.
Payload CMS v3 (MIT, part of Figma): Next.js colocation, Drizzle ORM and Lexical
Payload is an open-source (MIT), TypeScript headless CMS and application framework. It installs inside a Next.js app and stores content in Postgres or MongoDB.
- Next.js App Router integration with no sidecar. Payload installs inside your Next.js monorepo (
app/(payload)/admin/[[...segments]]/page.tsx), so oneDockerfiledeploys your marketing site, B2B portal, and CMS admin together. Inside React Server Components, Payload's Local API talks to PostgreSQL from the same Node process. That removes the network hop, REST/GraphQL over-fetching, and codegen drift. - Relational tables and migrations through Drizzle ORM. Under
@payloadcms/db-postgres, collections compile into PostgreSQL tables, indexes, and Drizzle migrations (see Drizzle vs Prisma). Rich text is stored as Lexical AST blocks that render typed React components. - MIT license after Figma. When Payload joined Figma in June 2025, both Figma and Payload said it stays open source and self-hostable, and the repository license is still MIT. There's no per-seat or SSO fee in that license. Custom auth strategies let you plug in WorkOS or any OIDC/SAML provider, but you build and maintain the integration yourself. Teams deploy Payload onto AWS ECS Fargate, Vercel, or Kubernetes VPCs.
Strapi v5 (MIT core plus proprietary EE): standalone Koa server, Document Service and EE paywalls
Strapi is an open-source Node.js headless CMS that runs as its own server, with a browser-based content-type builder and REST and GraphQL APIs.
- Two clusters and a serialization tax. Strapi runs its own Koa/Node server and React admin SPA. Every Next.js build or ISR revalidation serializes JSON over HTTP or GraphQL between two services, so you coordinate schema deploys, run two container clusters, and generate GraphQL types.
- Document Service improvements. Strapi v5 dropped v4's deeply nested
data.attributesJSON wrapper, introduced stabledocumentIdidentifiers across locales and drafts through the Document Service API, and added Draft & Publish history. The v5 breaking changes list has the details. - Cloud and CMS licenses are billed separately. Strapi Cloud hosting lists Starter at $35, Pro at $90, and Business at $450 per project per month. Strapi CMS feature tiers come on top: Community is $0, Growth is $45 a month with 3 users included and $15 for each extra user, and Enterprise is a custom contract that adds Audit Logs and Review Workflows. Admin SSO comes with Enterprise or as a paid Growth add-on.
Directus ($5M finances threshold) and Sanity (hosted Content Lake)
- Directus (
directus/directus): Built in TypeScript on Node, Directus on GitHub introspects existing PostgreSQL schemas without owning migrations. The catch is the license. BSL 1.1 allows free production use only while the whole organization has under $5,000,000 in total annual finances (revenue, funding, and grants). Series A+ and mid-market firms above that line need a commercial contract. - Sanity (
sanity.io): Sanity separates its React Studio from a proprietary cloud Content Lake queried with GROQ. Real-time collaborative editing is the main reason teams pick it. The Free plan covers up to 20 seats and Growth costs $15 per seat per month up to 50 seats, which fits startups. SAML SSO, custom RBAC roles, and audit logs need Sanity Enterprise, which is priced by custom contract and keeps content outside your own VPC Postgres.
Production blueprint: closing Payload's overrideAccess default with Postgres FORCE RLS
Payload's Local API is fast because it skips the HTTP layer. It also skips access control unless you ask for it: in the server-side Local API, overrideAccess defaults to true. That hurts most in multi-tenant SaaS and regulated multi-brand portals built around API integration.
Read Payload's Access Control documentation and the Local API access control page closely. If a developer calls payload.find() with a session user inside a Next.js Server Action or Route Handler and forgets to pass overrideAccess: false, Payload ignores that user's permissions and skips the collection access-control functions. Drafts from other tenants leak unless a tenant filter or Payload's multi-tenant plugin enforces the boundary. Raw SQL is worse. When another engineer writes a custom join through payload.db.drizzle for a sitemap or dashboard, the CMS-level TypeScript hooks never run at all.
We close both gaps with two layers: a TypeScript wrapper in the app, and PostgreSQL row-level security (RLS) underneath it. The wrapper is the part specific to Payload. Server code never calls the Local API directly. It goes through a small helper that takes the user and tenant ID, always passes overrideAccess: false so callers cannot switch access control back off, and merges a tenant_id equals filter into every where clause.
The rest of the blueprint is design decisions more than code. Here is where each guarantee lives:
Two caveats. RLS isn't a PostgreSQL 17 feature. It has shipped since 9.5, and 17 is just the version we run. And FORCE makes the policies apply to the table owner too, but superusers and roles with BYPASSRLS still skip them, so the app has to connect as a dedicated role that has neither (see multi-tenant SaaS architecture). Binding the tenant setting to Payload's transaction also means reaching into internal adapter state that can change between minor versions. Pin your Payload version and re-test on every upgrade.
Pair this with WorkOS OIDC/SAML inside Payload's custom auth.strategies array (see WorkOS vs Auth0) and Lexical blocks governed by your design system. You get SAML SSO, an immutable audit log for SOC 2 evidence, optimistic locking, and tag-based ISR cache invalidation, and you pay no per-seat CMS fee.
Worked 36-month TCO model: 35 seats across 12 localized B2B brand portals (40M monthly views)
At 35 seats, owned Payload v3 totals $87,900 over 36 months, against $138,900 to $177,600 for the other three, under the assumptions below. Read the vendor lines with care. The Sanity, Strapi, and Directus enterprise prices are modelled quotes, not list prices, because all three sell Enterprise by custom contract. The AWS lines are our own list-price estimates and move with instance class and region.
The model is a mid-market multi-brand group (more than $10M in annual revenue, so above Directus's $5M free threshold) over 36 months: 35 editorial, legal, and regional marketing seats, 12 localized B2B brand portals, 40,000,000 monthly page and API views, mandatory Okta/Entra SAML SSO, immutable audit logs for SOC 2 Type II and PCI DSS v4.x adjacent portals, and live visual Draft Mode.
Engineering and DevOps labor is priced at Idealogic's $75/hr senior CEE rate against a $165/hr US agency rate, a 2.2x ratio. The US agency figure is our own market assumption. For wage context, see Eurostat's hourly labour cost levels (lc_lci_lev) and the O*NET Software Developers profile (15-1252.00):
- Option A: Sanity Enterprise (35 SAML SSO seats + Audit Logs + 12 Brand Datasets + Next.js Hosting):
- Setup (160h @ $75/hr CEE = $12,000; $165/hr US = $26,400): Sanity Studio workspaces, GROQ typegen, Presentation live preview, and Next.js webhook revalidation.
- 36-mo SaaS + Hosting ($4,150/mo x 36 = $149,400): Sanity Enterprise contract, modelled quote (35 SSO seats, custom RBAC, audit logs, and 40M CDN API requests = $3,800/mo) + Next.js compute/CDN ($350/mo) + 6h/mo (216h) ops ($16,200 CEE / $35,640 US) = $177,600 ($211,440 US).
- Option B: Strapi v5 Enterprise Self-Hosted + Next.js Split Topology (35 seats + SSO + Audit Logs + Workflows):
- Setup (260h @ $75/hr CEE = $19,500; $165/hr US = $42,900): Separate Strapi v5 Koa ECS cluster, Document Service GraphQL/REST SDKs, SSO, and Next.js Draft Mode.
- 36-mo EE License + AWS ($3,350/mo x 36 = $120,600): Strapi CMS Enterprise license, modelled quote (35 seats with SAML SSO, Audit Logs, Review Workflows = $2,450/mo) + dual Koa & Next.js clusters, Multi-AZ RDS, and S3/CloudFront ($900/mo) + 12h/mo (432h) dual-cluster & upgrade SRE ($32,400 CEE / $71,280 US) = $172,500 ($234,780 US).
- Option C: Directus Commercial (above the $5M finances threshold) + AWS Hosting (35 seats + SSO + Next.js):
- Setup (220h @ $75/hr CEE = $16,500; $165/hr US = $36,300): Directus Node cluster, Redis cache, PostgreSQL schema permissions, SSO, and Next.js SDKs.
- 36-mo Commercial License + AWS ($2,650/mo x 36 = $95,400): Directus commercial contract above the $5M finances threshold (modelled quote, $1,850/mo) + Directus/Next.js AWS ECS, Redis, and RDS ($800/mo) + 10h/mo (360h) ops ($27,000 CEE / $59,400 US) = $138,900 ($191,100 US).
- Option D: Owned Payload v3 (MIT) Next.js Monorepo on Postgres 17 FORCE RLS + WorkOS SAML SSO + Audit Log:
- Turnkey Build (560h @ $75/hr CEE = $42,000; $165/hr US = $92,400): Payload v3 Collections/Lexical blocks inside Next.js
/app, Postgres 17 FORCE RLS tenant isolation, anoverrideAccess: falsewrapper, WorkOS SAML strategy, an immutable audit log, and live preview. - 36-mo AWS + WorkOS ($750/mo x 36 = $27,000): Unified Next.js + Payload Graviton4 Fargate ($220/mo), Multi-AZ RDS PG17 ($280/mo), S3 + CloudFront CDN serving 40M
revalidateTagviews ($125/mo), and 1 WorkOS enterprise SSO connection ($125/mo; $0 per-seat CMS fee) + 7h/mo (252h) single-repo ops ($18,900 CEE / $41,580 US) = $87,900 ($160,980 US).
- Turnkey Build (560h @ $75/hr CEE = $42,000; $165/hr US = $92,400): Payload v3 Collections/Lexical blocks inside Next.js
Sensitivity check: Small editorial teams vs enterprise SSO payback math
- Under 10 seats without SAML SSO: Sanity's Free plan covers up to 20 seats (Growth is $15 per seat up to 50), and a Payload v3 MVP costs $12k to $18k to build, so both stay lean. Sanity wins if non-technical editors want Google-Docs-style multiplayer cursors on day one. Git-first engineering blogs can skip the database entirely with MDX (see why we left Webflow).
- Enterprise SSO and multi-brand payback (35+ seats): At $75/hr CEE, Option D's monthly OpEx is $1,275 ($750 AWS/WorkOS plus $525 ops). The extra CapEx pays back in 7.56 months against Strapi v5 Enterprise ($22,500 / $2,975 a month saved), 9.02 months against Sanity Enterprise ($30,000 / $3,325 a month saved), and exactly 12 months against Directus Commercial ($25,500 / $2,125 a month saved).
2026 headless CMS and multi-brand portal build costs (senior CEE vs US)
Our JavaScript development, custom software development, and progressive web app development squads deliver fixed-milestone CMS builds at $75/hr senior CEE rates. The US agency column uses $165/hr, 2.2x higher:
Inside the worked 560-hour ($42,000 CEE vs $92,400 US) Payload v3 multi-brand build
- 1. Collections & Lexical UI (25% / 140h / $10,500 CEE vs $23,100 US): Multi-locale collection schemas, governed Lexical React block components, S3 media upload pipelines, and custom admin views.
- 2. Postgres 17 FORCE RLS & Safe Local API (25% / 140h / $10,500 CEE vs $23,100 US): Drizzle migrations, restrictive tenant policies, the
overrideAccess: falsewrapper, and optimistic locking. - 3. WorkOS SAML SSO & Audit Log (20% / 112h / $8,400 CEE vs $18,480 US): Okta/Entra SAML group-claim role mapping, an append-only audit log table, and release approvals.
- 4. Next.js RSC, Live Preview & ISR (20% / 112h / $8,400 CEE vs $18,480 US): Split-pane live visual preview,
revalidateTagedge invalidation, JSON-LD schema emitters, and WCAG 2.2 AA components. - 5. Content Migration & CWV QA (10% / 56h / $4,200 CEE vs $9,240 US): Legacy Strapi/WordPress ETL scripts, automated redirect verification, Playwright RBAC tests, and sub-1.8s LCP tuning.
Strapi and Sanity look inexpensive at five seats, then triple in TCO the day compliance mandates SAML SSO and immutable audit logs across thirty-five editors. Embedding MIT Payload v3 directly inside Next.js on PostgreSQL 17 RLS cuts ongoing platform run-rate from fifty thousand dollars a year to fifteen thousand without maintaining a second backend server.
Payload vs Strapi decision checklist and shipped Idealogic proof
Idealogic builds content-driven portals and multi-tenant web apps where Core Web Vitals, localization, and strict tenant boundaries all matter at once:
- Multi-jurisdiction investor & asset portals (E-States): Structured offering disclosures, localized compliance workflows, and high-conversion Next.js investor portals across real estate and proptech.
- Creator catalog & campaign workspaces (Ugreator): Multi-role brand and creator publishing dashboards integrated with automated payout state machines across the social and creator economy.
- Regulated document & identity portals (Chaindoc): Tamper-evident audit logs, RBAC workspaces, and eIDAS-qualified workflows engineered for fintech and enterprise legal teams.
Five checks settle Payload vs Strapi vs Directus vs Sanity in 2026:
- Building on Next.js App Router and PostgreSQL, and you want MIT ownership with no SSO license fee? Choose Payload CMS v3, and enforce
overrideAccess: falseplus Postgres FORCE RLS. - Backend is Go, Python, or Java with non-Next.js consumers, and there's budget for Enterprise SSO? Choose Strapi v5 Enterprise.
- Wrapping an existing brownfield SQL database for internal ops, with under $5M in total annual finances or a commercial budget? Choose Directus.
- Editorial team wants zero-ops cloud hosting and real-time multiplayer cursors without VPC Postgres rules? Choose Sanity (Free up to 20 seats, Growth up to 50, Enterprise for SSO).
- Unsure whether you even need DXP personalization modules on top? Check our CMS vs DXP guide first, or contact our team for a 48-hour headless CMS architecture and TCO audit.

Deputy CEO at Idealogic. Overseeing operational execution, product discovery, client partnerships, and engineering squad delivery.
Frequently asked questions
Choose Payload CMS v3+ when your engineering squad builds in TypeScript and Next.js App Router, wants zero-network-hop Local API queries backed by Drizzle ORM on PostgreSQL, and needs permissive MIT ownership without per-seat SAML SSO or audit-log paywalls. Choose Strapi v5 when your backend runs a polyglot Go, Java, or Python stack outside Next.js or when non-technical admins prefer browser-driven schema modeling paired with Strapi Enterprise governance.
No. Payload joined Figma in June 2025, and both companies' announcements say it stays open source and self-hostable. The repository license is still MIT. Production teams deploy Payload inside a standard Next.js app on AWS ECS, Vercel, Cloudflare, or Docker.
In Payload's server-side Local API, calls such as payload.find() and payload.update() default to overrideAccess true even when you pass a user object, so they skip every TypeScript collection access-control function unless you also pass overrideAccess false. In multi-tenant portals, we pair strict TypeScript wrapper functions that enforce overrideAccess false with PostgreSQL FORCE ROW LEVEL SECURITY, so a forgotten flag or a raw Drizzle query fails closed.
Strapi bills Strapi Cloud hosting and Strapi CMS feature licenses separately. The core engine is MIT licensed, but Audit Logs and Review Workflows sit behind a Strapi CMS Enterprise contract, and admin SSO needs Enterprise or a paid add-on. You also run and scale a standalone Koa Node.js server cluster next to your Next.js frontend, which Payload doesn't need.
Directus introspects existing SQL schemas without code generation, but its BSL 1.1 license allows free production use only for organizations under $5 million in total annual finances (revenue, funding, and grants combined), so funded mid-market companies need a commercial license. Sanity pairs a real-time Studio with a hosted Content Lake queried via GROQ ($15 per seat monthly on Growth), but keeps SAML SSO, custom roles, and audit logs on Sanity Enterprise.
Payload's @payloadcms/db-postgres adapter maps TypeScript Collections and Globals directly onto normalized PostgreSQL tables, foreign keys, and versioned SQL migrations powered by Drizzle ORM. Inside React Server Components or Server Actions, Payload's Local API executes typed SQL queries in-process without JSON-over-HTTP REST or GraphQL serialization, and exposes payload.db.drizzle for complex analytical joins inside shared database transactions.
A macro CMS vs DXP evaluation decides whether an organization should buy a six-figure monolithic suite (such as Adobe Experience Manager, Sitecore, or Optimizely) versus assembling a composable stack. Once you reject monolithic DXP shelfware in favor of headless architecture, this guide settles the code-level engineering choice between Payload CMS, Strapi v5, Directus, and Sanity.
Partnering with Idealogic's senior Central and Eastern European TypeScript, Next.js, and PostgreSQL engineers at $75 per hour (versus $165 per hour at US domestic agencies), a single-brand Payload or Sanity MVP runs $12,000 to $18,000, a Strapi-to-Payload migration runs $18,000 to $27,000, and a worked 560-hour multi-brand platform with PostgreSQL 17 FORCE RLS, WorkOS SAML SSO, and immutable audit logging costs $42,000 ($92,400 at US rates).
More from the journal

Paddle vs Lemon Squeezy in 2026: MoR Fees, Stripe & TCO
Paddle vs Lemon Squeezy and Stripe Managed Payments in 2026, compared on fees and 36-month TCO: stacked international and subscription surcharges, merchant of record liability, B2B Net-30 invoicing, PG17 entitlement guards, and the math for a $6M/yr SaaS.

Nylas vs Cronofy in 2026: Calendar Sync, CASA & TCO
Nylas vs Cronofy vs Aurinko and owned Google Workspace + Microsoft Graph sync in 2026: public per-account pricing, which Google scopes need CASA Tier 2, a PostgreSQL KMS token vault, and 36-month TCO math at 6,000 accounts.

Persona vs Sumsub in 2026: KYC Pricing, PAD & 36-Mo TCO
Engineering and 36-month TCO teardown of Persona vs Sumsub vs Veriff vs Entrust Onfido in 2026: retry and per-service billing terms, ISO/IEC 30107-3 liveness, PG17 FORCE RLS KYC waterfalls, and 20k/mo applicant math.