Content engine: See how each piece of content earns pipeline credit.

The problem

Editorial teams can see publishing activity and traffic without knowing which content introduces buyers, assists later conversions or costs more than expected to produce. This system is designed to connect editorial execution with search, consented behavioral events and CRM progression while keeping attribution distinct from proof of causation.

The dashboard

12 views: Today, Editorial summary, Reader replies, Briefing requests, Editorial calendar, Production board, Piece performance, Subscribers, Lead scoring, Sponsors, Search, Reporting.

The story river: six weeks of published pieces placed on their publish dates, each block sized by the signups it drove, with the readers who became briefing requests hanging beneath it.

State lives in the URL (view, filters, date range), records open in drawers with their history, and every action that changes something is written to an activity log and is safe to repeat.

Data model

• articles(id, canonical_url, topic_id, status, published_at)
• editorial_tasks(id, article_id, owner_id, due_at, state)
• content_costs(id, article_id, category, amount_cents, incurred_at)
• search_daily(id, article_id, query, date, clicks, impressions)
• touchpoints(id, visitor_key, article_id, occurred_at, consent_state)
• identity_links(id, visitor_key, crm_contact_id, evidence)
• conversions(id, crm_contact_id, type, occurred_at)
• attribution_edges(id, conversion_id, article_id, role, model_version).

Rules in the backend

1. Proposed first-touch credit goes to the earliest eligible recorded touch within 90 days before conversion.
2. An assisted article must appear before conversion within that window and must not be the credited first-touch article.
3. Assisted counts are non-additive across articles and must not be summed as unique conversions.
4. Cost per article equals recorded direct costs plus approved allocated labor; missing costs remain unknown.
5. Publishing requires approved copy and required metadata; a board-status change alone cannot publish.

Storage

Recommend Supabase Postgres with tenant RLS. Retain raw event payloads for 30 days, consented journey data for 13 months and editorial/audit records for 24 months. Respect consent withdrawal and deletion requests. Mask personal addresses in shared reports.

Reliability

Use incremental imports with overlap windows, idempotent publishing requests and a dead-letter queue. Preserve the last successful dataset during connector failures. Show Search Console delays, incomplete dates, unmatched identities and source-specific freshness.

Security and access

Writers draft; editors approve; publishers release; analysts inspect results. Clients receive read-only reporting. Webflow and CRM tokens stay server-side, with scopes restricted to the necessary collections and properties.

Cost

Production labor, freelancers, Webflow plans, analytics tiers, Segment volume and orchestration can cost money. Report recorded costs rather than assuming organic traffic is free.

Technology stack

• Webflow
• Search Console
• GA4
• Segment
• HubSpot
• Customer.io

Map a workflow like this

Start with the handoffs your team already runs and where they break.

Discuss a workflow →