THE MARKETING HUB

One intake portal that keeps Kanban, calendar and project tracker in sync

Full-Stack + IntegrationsReact / Express / PostgreSQL / Tracker API2026

The Problem

Marketing requests came in by email, Slack and hallway conversation, and nothing tracked them in one place. Each request lived wherever it was submitted: a thread, a sticky note, a forwarded message sitting in someone's inbox. So deadlines took the team by surprise. Duplicates wasted hours, with two people commissioning the same asset and neither knowing the other had asked.

The bigger problem was that nobody could see capacity. When a request came in with a tight deadline, there was no reliable way to tell whether it could be done by then. The answer came from whoever happened to know, not from the data, which meant either over-promising or a last-minute scramble to reprioritize.

The team also kept a spreadsheet tracker as its record of truth, shared with an outside partner team. A new intake tool had to sync with that tracker rather than replace it, so the integration had to be designed in from the start.

Marketing Hub intake flow: web form to Kanban card, calendar event, and dual project-tracker destinations.
SYSTEM DIAGRAM — INTAKE FORM TO DUAL PROJECT-TRACKER ROUTING

The Solution

The Marketing Hub is a production marketing-operations portal, deployed as a private internal tool (). One submission form has 12 fields: task description, requester, priority, deadline, objective, audience, deliverables, services, reference links, destination, assignee and date. Submitting it creates three things at once: a Kanban card on the live board, a deadline entry on the priority-coded calendar and a synced row in the right project tracker. All three come from the same record, so they always show the same state.

One submission makes one validated record and three views that stay in sync.

The Kanban board has drag-and-drop swimlanes, so status updates happen where the work is visible. The calendar codes entries by priority, so the team sees crunch points coming. Nobody re-enters anything after intake. The record flows to both views automatically, and the tracker sync logs its status per row, so failures show up right away.

The portal is built on the organization's existing brand system, so it looks like one of their own products rather than a reskinned off-the-shelf tool.

Craft Details

DUAL-DESTINATION ROUTING

A toggle sends the request to either the internal marketing-ops team or an outside partner team, and the assignee dropdown refills to match. The record syncs to the matching tracker sheet. Each row's sync status is tracked as pending, synced or error, so nothing quietly disappears into a queue.

GRACEFUL DEGRADATION

If the tracker integration isn't configured or the API is down, the form still saves locally and tells the submitter plainly that the sync failed. An integration problem never loses data. The local record is the source of truth, and submitting doesn't depend on the sync.

DUPLICATE DETECTION

Automated near-duplicate detection flags repeat requests before they reach the queue. The submitter sees a duplicate banner while submitting, not in a weekly triage meeting, and can point to the existing request or cancel.

SLA CAPACITY CHECK

On submission, the deadline is checked against per-service SLA definitions and live team-capacity records. So "can this be done by then?" gets answered at intake instead of when the deadline slips. If a request fails the check, the submitter sees a conflict notice with the earliest date that works.

Built to Be Demonstrated

The portal shipped with a written walkthrough guide for a CMO demo, and that shaped the design. Every view had to make sense to a non-technical audience at a glance. The main path had to work on the first try: submit a request, watch it appear on the board and calendar, confirm the tracker row. The walkthrough is there to show the value in under five minutes, not to cover for a complicated interface.

The brand system was critical here too. People ignore a portal that looks foreign to the organization, and they use one that looks like it belongs.

The Build Angle

The routing and project-tracker integration wave was built spec-first. It was 30 commits across May 31 and June 1, 2026, on top of the core product shipped in the first two days. The spec was broken into small, tightly scoped pieces, each piece was implemented and checked against it, and everything was reviewed before deploy. It's the same spec-first approach documented at /work/conductor-pattern: spend the judgment up front and keep the implementation mechanical. The integration wave landed with no regressions to the core product, because the spec was tight enough that nobody had to guess at scope.

Stack

React 18ViteTypeScriptExpressPostgreSQLTracker API integrationCloudflare

Result

12
Database tables
28
API endpoints
12
Intake fields
30
Commits (Feb 8 – Jun 1, 2026)
~1,100 lines
TypeScript (approx.)
~2,800 lines
CSS (approx.)

The build came in two waves. The core product shipped in the first two days: intake form, Kanban board, calendar view, database schema and all 28 API endpoints. The routing and tracker integration wave followed on May 31–June 1 and added dual-destination routing, per-row sync status, duplicate detection and the SLA capacity check. Thirty commits, two phases, and one production deployment as a private internal tool ().

UNMEASURED

Request volume since launch and adoption rate aren't tracked. The live site has no analytics, by design. Tracker sync outcomes are logged per row but haven't been rolled up into a success rate. These gaps are intentional, because the org values data ownership over telemetry. So the honest answer to "how is it performing?" is that the data is in the database, not in a dashboard.

The Intake Flow

The capacity question gets answered at intake, not at the deadline.