Trusted Tech Leader/ Case studies

Case study · SaaS platforms

Programmatic SEO page generation SaaS

A multi-workspace SaaS that turns CSV data and templates into SEO-checked pages and publishes to WordPress, Shopify, WooCommerce and PrestaShop.

Sector
Marketing agencies and SEO
Category
SaaS platforms
Maturity
In active development; feature-rich app with CI regression workflows

At a glance

Who it's for
SEO agencies, marketers and site owners who need to publish many location- or keyword-based pages.
Core stack
React, TypeScript, Supabase, Stripe
What sets it apart
One pipeline from spreadsheet rows to verified, SEO-scored pages live on the customer's own CMS, across several e-commerce and CMS platforms.
On this page

What it looks like in use

Setting up a campaign from a spreadsheet, then watching the pages move through to the customer's website.

Page generator: Campaigns
Python courses by cityProcessing
Campaign setup
TemplateCourse in a city
Data sourcesKeyword group, locations list
Keyword group fills{course}
Locations fills{city}
Title pattern{course} Training in {city}
Page address/python-training-{city}
WebsiteWordPress connected
Generate pages
Pages from this campaign
Python Training in TorontoPublished
Python Training in OttawaPublished
Python Training in CalgaryPublishing
Python Training in HalifaxGenerating
Four more pagesPending
Page previewPython Training in Toronto
Learn Python in Toronto with expert instructors.SEO checked
Illustrative screen sample data, not client data

Real screens from the working product (sample data)

A screen showing where each gap in the page layout gets its value, such as a keyword group for the course and a locations list for the city, though values can also come from a spreadsheet, above a live preview of the finished page.
Real screen, sample data Showing where each gap in the page layout gets its value (here the course comes from a keyword group and the city from a locations list; values can also come from a spreadsheet), with a live preview of the finished page. Sample data.
A campaign listing eight pages, each with a status.
Real screen, sample data One campaign listing its eight pages with a status for each. Simulated on a test setup with sample data.

From a spreadsheet to live pages

  1. Maya uploads a spreadsheet

    Each row holds the details for one page, such as a course and a city.

  2. She chooses a page layout

    She picks a template and matches each gap in the page, such as the city name, to a spreadsheet column, a keyword group or a locations list.

  3. One page is written for every row

    The system fills in the template for each row, so every page has its own title, text and web address.

  4. Each page gets a quality check

    Titles and page details are checked for search engines before anything is published.

  5. The pages go live on the site she manages

    The finished pages are sent straight to her WordPress or shop site, with the result shown for each one.

The challenge

Publishing hundreds of near-identical landing pages, one per city, product or keyword, is slow and error-prone when done by hand in a CMS.

Copy has to be unique enough to avoid duplicate-content problems, and each page needs correct metadata and structured data. Every target platform, from WordPress to Shopify, also has its own publishing API and page-builder format. The original brief asked for a SaaS where a user uploads a CSV, maps columns to template variables, generates one page per row and pushes them to a connected website. It also called for usage-based subscription plans and an admin panel for the platform owner.

The solution

The platform is a React single-page dashboard backed by Supabase: Postgres, authentication and a large set of edge functions. Users create campaigns that combine a template, a CSV or keyword group and a connected website.

A generation engine fills the variables, optionally enriches copy with AI, scores SEO quality and publishes through platform-specific connectors. The codebase has since grown into a broader suite: a template marketplace, an AI site builder, SEO audit and indexing tools, scheduled runs, workspaces with members and a companion WordPress plugin.

For your technical teamhow it works under the hood

Campaign and page lifecycle

Campaign and page lifecycle in the page generation platformState diagram in two parts. Campaign states: a draft is queued, a queued campaign is started by the scheduled runner and becomes processing, and processing ends as completed or failed. A processing campaign can be paused, and one that is stuck with no active job is reset to draft. Page states, one per CSV row: pending, generating, publishing, then published, with a failed state reachable from publishing.CAMPAIGNPausedDraftQueuedProcessingCompletedFailedqueuestartall pages handledpausegeneration errorstuck with no active job: reset to draftScheduled runner starts due queued campaignsPAGE, ONE PER CSV ROWPendingGeneratingfill, AI, SEO checkPublishingconnector callPublishedlive on the siteFailed
Top: a campaign moves from draft to queued to processing and ends completed or failed, with pause and a reset for stuck runs. Bottom: each CSV row becomes a page that moves through generating and publishing, and can fail without stopping the others.

Key features

What the system does, as built in the codebase.

  • 01

    Campaign wizard with CSV mapping

    Upload data, map columns to template variables, pick a site and generate pages in bulk.

  • 02

    Multi-platform publishing

    Connectors for WordPress, Shopify, WooCommerce and PrestaShop, plus an Elementor page-builder engine and a Gutenberg engine.

  • 03

    AI content and SEO tooling

    AI row generation, content rewriting, translation, SEO optimisation and an analysis engine with analyzers and auto-fixers.

  • 04

    Template marketplace and kits

    Browsable layout templates, with template scanning and validation.

  • 05

    Scheduled, reversible publishing

    Queued campaigns run on a schedule, and published pages can be rolled back.

  • 06

    Indexing and internal linking

    Search-engine ping and indexing submission, plus automatic internal-link building.

  • 07

    Subscriptions, credits and invoicing

    Plan checkout, a customer portal, usage entitlements, AI credits and PDF invoices.

  • 08

    Admin panel and referral programme

    Owner-facing user, payment and website management, with affiliate and referral pages.

How it fits together

From a spreadsheet row to a verified page on the customer's site.

  1. CSV and keyword dataupload, mapping, keyword groups
  2. Template enginevariable fill, section variants, AI enrichment
  3. SEO engineanalyzers, fixers, quality score
  4. Connector layerWordPress, Shopify, WooCommerce, PrestaShop
  5. Target websitepages published and verified
  6. Supabase backendPostgres with RLS, auth, edge functions
Data and keyword groups feed a template engine. An SEO engine scores and fixes the output. A connector layer publishes to the target website, while the Supabase backend holds data, authentication and the edge functions behind every step.
  • A scheduled runner picks up queued campaigns.
  • Billing webhooks update plan entitlements that gate generation.

Implementation notes

Engineering decisions worth knowing about.

  1. Workspace multi-tenancy

    Workspaces have owner, admin and member roles, with row-level security on most tables.

  2. Column-level grant hardening

    Secrets such as site credentials, payment IDs and webhook secrets are readable only by the service role. A regression script runs in CI to guard this.

  3. Encrypted third-party tokens

    Tokens are encrypted at rest with AES-GCM in a shared edge-function module.

  4. Connector abstraction

    Each target platform implements a common interface, with theme adapters and asset import. New platforms plug into one factory.

  5. Page-builder fidelity tests

    Fixtures, publish-verify and regression suites cover the Elementor engine and run in a dedicated workflow.

  6. Internationalisation

    Roughly thirty locale files, build-time validation scripts and an in-app auto-translate provider.

  7. Email queue and webhooks

    A transactional email queue handles suppression and unsubscribes, and Stripe webhooks are processed server-side.

  8. Static prerender and sitemap

    Prerendering and sitemap generation run at build time, and the app deploys as a single-page app with rewrites.

Technology stack

Taken from the project's dependency and config files.

Frontend
  • React 18
  • TypeScript
  • Vite
  • Tailwind CSS
  • shadcn/Radix UI
  • TanStack Query
  • React Router
  • Recharts
  • Framer Motion
Backend
  • Supabase
  • PostgreSQL
  • Supabase Auth
  • Deno edge functions
AI
  • OpenAI SDK
  • Multi-provider AI service module
Payments
  • Stripe
Integrations
  • WordPress REST API
  • Companion WordPress plugin
  • Shopify Admin API
  • WooCommerce
  • PrestaShop
  • Elementor
  • Google indexing
  • IndexNow
Quality
  • Vitest
  • Playwright
  • ESLint with a custom i18n plugin
  • GitHub Actions
Marketing assets
  • Remotion
For your technical teamtechnology layers

Technology layers

Technology stack layers of the page generation platformLayered diagram, top to bottom: frontend (React 18, TypeScript, Vite, Tailwind CSS, shadcn/Radix UI, TanStack Query, React Router, Recharts, Framer Motion); backend (Supabase, PostgreSQL, Supabase Auth, Deno edge functions); AI (OpenAI SDK, multi-provider AI service module); payments (Stripe); integrations with outside platforms (WordPress REST API, companion WordPress plugin, Shopify Admin API, WooCommerce, PrestaShop, Elementor, Google indexing, IndexNow); quality (Vitest, Playwright, ESLint with a custom i18n plugin, GitHub Actions); marketing assets (Remotion).FRONTENDReact 18TypeScriptViteTailwind CSSshadcn/Radix UITanStack QueryReact RouterRechartsFramer MotionBACKENDSupabasePostgreSQLSupabase AuthDeno edge functionsAIOpenAI SDKMulti-provider AI service modulePAYMENTSStripeINTEGRATIONSWordPress REST APICompanion WordPress pluginShopify Admin APIWooCommercePrestaShopElementorGoogle indexingIndexNowQUALITYVitestPlaywrightESLint with a custom i18n pluginGitHub ActionsMARKETING ASSETSRemotion
The app and backend sit in the first two layers. AI, payments and the outside platforms are services it calls. Only the integrations boxes are dashed, marking the outside platforms and services it publishes to.

Business value

What the system enables. No numbers are claimed.

  • Lets a small team publish large batches of location or keyword pages without manual CMS entry.
  • Supports the Elementor and Gutenberg page builders and theme adapters, so generated pages fit the target site.
  • Gives agencies a single workspace to manage several client sites and team members.
  • Turns usage into plan tiers, credits and invoices without manual billing work.

Need a system like this built?

Tell us what it needs to do and we will scope it with you on a call.

Discuss a build