All case studies
AI Automation2024

Content Workflow

From idea to everywhere, automatically

Client
Digital Marketing Agency (15 B2B clients, 4-person content team)
Role
Automation Specialist
Duration
5 weeks
Year
2024
60 → 85
Posts Monthly
-73%
Time to Publish
+18%
Avg Engagement
15 → 22
Client Capacity

Overview

A boutique B2B marketing agency was drowning in operational overhead. Managing content for 15 clients across LinkedIn, Twitter, and blogs meant their 4-person team spent 60% of their time on logistics (approvals, scheduling, reformatting) instead of strategy and creation. The founder reached out when they had to turn down 3 new clients because they literally didn't have capacity to manage more content calendars.

!

The Challenge

Each client had content scattered across Google Docs (drafts), email (approvals), Trello (calendar), and individual platform schedulers. To publish one post: writer drafts in Docs → emails client for approval → client replies with edits in email → writer updates doc → manually copies to LinkedIn, reformats for Twitter character limits → schedules in native tools → updates Trello. This took 45 minutes per post. They were publishing 60 posts/month and it was killing them.

The Solution

I built a single source of truth in Notion where all content lives, then automated everything downstream. Writers draft in Notion with client-specific templates. Zapier watches for status changes: 'Ready for Review' → sends Slack message to client with approval link. Client approves → triggers GPT-4 to generate platform-specific variations (LinkedIn long-form, Twitter thread, blog excerpt). Buffer API schedules across all platforms. Notion remains the master calendar showing what's published where.

The Process

1

Content Audit & Pain Point Mapping

Shadowed the team for 3 days, documenting every tool touch and handoff. Discovered 14 steps from draft to publish with 6 context switches. Interviewed clients—most hated email approval threads and wanted simple yes/no buttons. Mapped ideal state: write once in Notion, approve in Slack, publish everywhere automatically.

2

Notion Workspace Design

Built Notion workspace with database for all content. Each row = one piece of content with fields: client, draft, status, platforms, publish dates. Created templates for different content types (thought leadership, product updates, case studies). Set up client-specific dashboards showing their content pipeline.

3

Zapier Automation Chains

Built 12 Zaps handling the full lifecycle: 1) Status change to 'Review' → Slack message to client 2) Client clicks approve → Calls GPT-4 API to generate variations 3) GPT response → Creates Buffer posts 4) Buffer scheduled → Updates Notion with links 5) Post published → Slack notification with analytics link. Added error handlers for API failures.

4

GPT-4 Content Adaptation

Fine-tuned prompts for each platform: LinkedIn = professional tone + longer, Twitter = conversational + thread format, Blog = SEO-optimized + full article. Trained on client brand voices using their best-performing past content. Hit 85% approval rate on AI-generated variations (vs 40% in early testing before prompt refinement).

Tech Stack

ZapierNotion APIBufferGPT-4AirtableSlack
We went from turning away clients to finally having capacity to grow. What used to take 45 minutes now takes about 12. The AI-generated variations save us so much time on reformatting for different platforms.
C
Client feedback
Agency Operations Director

Ready to build something great?

Let's discuss your project and explore how I can help you achieve similar results.

Get in touch