Section 10.8.1: Project 1 — AI-Powered Social Media Strategy
🎯 Learning Objective
Complete a full project involving a social media content strategy using AI for research, creation, planning, and optimization — from A to Z. This project simulates a real digital consulting engagement and trains you to orchestrate multiple AI tools in a professional workflow.
Why This Project?
Social media content creation is one of the most immediate use cases for generative AI in business. According to HubSpot (2025), 82% of marketers already use AI for at least one step in their content production. But most limit themselves to "asking ChatGPT for a post." This project teaches you to go much further: from strategic audit to multi-format creation, through editorial planning and performance measurement.
What you will master:
- →AI-assisted competitive audit (Perplexity + GPT-5 for analysis)
- →Structured editorial planning (4-week calendar, format mix)
- →Multi-tool creation (text, visuals, proofreading, optimization)
- →Methodical documentation (reusable prompt guide for business use)
- →Critical evaluation of AI content quality
The Project Brief
Context: You are a digital transformation consultant and you need to create a 4-week social media strategy for a client (real or fictional).
Expected Deliverables:
- →Audit of the current presence + competitive benchmark (2-3 page document)
- →Editorial strategy with detailed publication calendar
- →12 pieces of content created (LinkedIn posts, articles, associated visuals)
- →Reusable prompt guide (by content type and by tool)
- →Tracking metrics and KPIs with dashboard
Difficulty Level: ⭐⭐⭐ (intermediate — requires skills from chapters 1 to 5)
Choosing Your Client
You can work on a real client (your company, a freelancer, an association) or a fictional one (invent a startup, a restaurant, a consulting firm). A real client is preferred because you can measure the actual impact of created content. If you choose a fictional client, create a detailed persona: industry, size, target audience, tone of voice, business objectives.
Phase 1 — Research and Audit (1h)
The audit is the foundation of any strategy. Without data, you're making assumptions. With AI, an audit that used to take 2 days can be completed in 1 hour.
Step 1a: Audit of the Existing Presence with Perplexity
Perplexity is ideal for research because it cites its sources and accesses the web in real time.
Analyze the social media presence of [company/brand]:
1. What channels do they use? (LinkedIn, Twitter/X, Instagram, TikTok, YouTube)
2. Publishing frequency on each channel (daily, weekly, monthly)
3. Best-performing content type (engagement measured by likes/comments/shares)
4. Average engagement per post (compare to industry average)
5. Current tone of voice and editorial positioning
6. Detailed comparison with 3 direct competitors on the same metrics
Sources: public data, recent posts (last 3 months), free monitoring tools.
Present results in a comparative table.
Step 1b: Opportunity Analysis with ChatGPT (GPT-5)
You are a social media strategy consultant with 10 years of experience.
Based on this audit [paste the Perplexity results below]:
1. Identify the 5 biggest missed opportunities, ranked by potential impact
2. For each opportunity, estimate the effort required (low/medium/high) and expected ROI
3. What type of AI content could be created with the least effort and greatest impact?
4. Propose a differentiating editorial positioning (not generic "industry expert")
5. Identify current content trends in this industry that the client isn't leveraging
6. What underused formats would have the most impact? (carousels, short video, newsletters, threads)
[AUDIT]
[paste Perplexity results here]
Step 1c: SWOT Matrix for Social Media Positioning
Synthesize the audit into a SWOT matrix:
| Positive | Negative | |
|---|---|---|
| Internal | Strengths: what the client does well | Weaknesses: identified gaps |
| External | Opportunities: trends, empty niches | Threats: strong competitors, algorithm changes |
Common Pitfall: The Superficial Audit
Don't just list the channels. A good audit quantifies: average engagement rate, actual vs. desired frequency, followers/engagement ratio, growth over 6 months. This data drives the entire strategy. If Perplexity doesn't find enough data, supplement with free tools like Social Blade (YouTube/Instagram) or direct searches on the profiles.
Phase 2 — Strategy and Editorial Calendar (1h)
Step 2a: Define the Editorial Strategy
Before creating the calendar, define the pillars:
As a social media strategist, define the editorial strategy for [brand] on LinkedIn.
Context:
- Industry: [X]
- Primary target: [persona: role, seniority, industry, interests]
- Business objective: [lead generation / brand awareness / recruitment / thought leadership]
- Desired tone of voice: [professional but approachable / sharp expert / quirky and human]
Define:
1. The 4 content pillars (recurring themes that structure the editorial line)
2. The unique value proposition of the account (why follow this brand rather than another?)
3. The signature format (the type of post that will become their trademark)
4. The optimal frequency and best publishing days/times (with justification)
5. The 3 priority KPIs to track and objectives for 4 weeks
Step 2b: Create the Detailed Calendar
Create a 4-week editorial calendar for [brand] on LinkedIn.
Defined content pillars: [paste the 4 pillars from the previous step]
Constraints:
- 3 publications per week (Monday, Wednesday, Friday)
- Mix: 40% educational value, 30% storytelling, 20% promotion, 10% engagement
- Each entry must contain:
- Date and day
- Content pillar (which of the 4)
- Post type (carousel, text only, image + text, video, poll)
- Specific topic
- Hook (1st sentence that stops the scroll — max 15 words)
- Content body (summary in 2 lines)
- CTA (call-to-action)
- Visual to create (description for AI)
- Recommended hashtags (3-5 relevant ones)
- Integrate current trends in [X] industry
- Week 1 = awareness, Week 2 = education, Week 3 = social proof, Week 4 = conversion
Format: detailed Markdown table
Step 2c: Validate Consistency
Before moving to creation, have your strategy validated by Claude:
You are a demanding communications director. Critically review this editorial strategy and calendar:
[paste the strategy + calendar]
Evaluate on these criteria:
1. Consistency: does the content truly serve the business objective?
2. Differentiation: does it stand out from competitors or is it "generic LinkedIn content"?
3. Feasibility: is 3 posts/week realistic for this client?
4. Progression: is there a real logic across the 4 weeks or is it random?
5. What's missing? What would you change?
Be honest and constructive.
Phase 3 — Content Creation (2h)
This is the heart of the project. You will produce 12 complete pieces of content (text + visual).
Workflow per content piece (repeat 12 times):
Step 3a — Writing (ChatGPT / GPT-5)
For an educational post (carousel):
Write a LinkedIn carousel post about [topic from the calendar].
Context: [brand], target = [persona]
Approved hook: "[hook from the calendar]"
Carousel structure (8-10 slides):
- Slide 1: Hook + provocative question (title only, high impact)
- Slides 2-8: One point per slide, short sentence + one data point or example
- Slide 9: Visual summary (the 3 key takeaways)
- Slide 10: CTA + question for engagement
Tone: [tone defined in the strategy]
Include 2-3 verifiable data points.
Each slide = max 30 words.
For a storytelling post:
Write a LinkedIn storytelling post about [topic].
Hook-Story-Lesson-CTA structure:
- Hook: a striking sentence that stops the scroll (start with a surprising fact or question)
- Story: anecdote or experience in 5-8 short sentences (1-2 line paragraphs)
- Lesson: the actionable takeaway (bulleted list of 3 points)
- CTA: open question to generate comments
Tone: personal but professional, authentic, no corporate-speak.
Length: 150-200 words (optimal format for LinkedIn).
Step 3b — Visual Creation (DALL-E 3 / Midjourney v7)
For each post, create an appropriate visual:
Create [a professional visual for LinkedIn / a carousel cover image]:
Topic: [post theme]
Style: clean, corporate but modern, color palette [brand colors]
Elements: [relevant icons, visualized data, professional character]
Text on image: "[carousel title]"
Format: 1080x1080 (LinkedIn square) or 1080x1350 (portrait carousel)
Do not include: realistic photos of people, unauthorized logos
Step 3c — Proofreading and Optimization (Claude Opus 4.6)
You are a LinkedIn ghostwriter with 50K+ followers. Review this post and visual (description):
[post]
[visual description]
Evaluate:
1. Does the hook really stop the scroll? (score /10)
2. Is the structure optimal for LinkedIn? (spacing, mobile readability)
3. Will the CTA generate comments?
4. Is the tone authentic or "corporate-speak"?
5. LinkedIn SEO: are important keywords naturally present?
Propose an improved version if the score is < 7/10.
Step 3d — Final Personalization (YOU)
This is the most important step. AI created the foundation, but your personal touch makes the difference:
- →Add a personal anecdote or opinion that AI cannot invent
- →Verify that the tone sounds "human" and not "robotic"
- →Add emojis sparingly if it fits the brand style
- →Read aloud: if it doesn't sound natural, simplify
Pro Tip: The 80/20 Ratio
AI generates 80% of the structure and first draft. You bring the 20% that makes all the difference: the human touch, industry expertise, lived anecdotes, internal data. A 100% AI post is detectable. An AI + human post is undetectable and performs better.
Phase 4 — Documentation and Prompt Guide (1h)
Step 4a: Reusable Prompt Guide
Create a professional document you could hand to a client:
| Content Type | Recommended Tool | Prompt Template | Parameters | Average Time |
|---|---|---|---|---|
| Educational post (carousel) | GPT-5 | [full prompt] | Temperature 0.7 | 15 min |
| Storytelling post | GPT-5 | [full prompt] | Temperature 0.8 | 10 min |
| Engagement post (poll) | Claude | [full prompt] | Default | 5 min |
| Visual (simple post) | DALL-E 3 | [full prompt] | 1024x1024 | 5 min |
| Visual (carousel cover) | Midjourney v7 | [full prompt] | --ar 4:5 | 10 min |
| Proofreading/optimization | Claude Opus 4.6 | [full prompt] | Default | 5 min |
Step 4b: Projected Performance Report
Estimate expected metrics for your strategy:
Based on this social media strategy and these 12 pieces of content, project the expected results over 4 weeks:
Content: [paste the list of 12 contents with type + topic]
Industry: [X]
Current account metrics: [followers, avg engagement rate]
Project:
1. Expected engagement rate per content type (carousel vs. text vs. image)
2. Estimated total impressions over 4 weeks
3. Estimated lead generation (if applicable)
4. Comparison with industry benchmarks
5. Realistic recommendations to improve results in month 2
6. ROI of the AI approach: estimated time for 12 manual contents vs. 12 AI-assisted contents
Evaluation Rubric
Your project will be evaluated on these criteria:
| Criterion | Weight | Insufficient (0-4) | Satisfactory (5-7) | Excellent (8-10) |
|---|---|---|---|---|
| Audit quality | 15% | Superficial, no data | Data present, basic analysis | Quantified audit with competitive insights |
| Strategy coherence | 20% | Generic, no real angle | Clear pillars, aligned calendar | Differentiating positioning, documented choices |
| Content quality (12 posts) | 30% | Robotic, generic | Personalized, well-structured | Pro-quality, publishable as-is |
| Prompt documentation | 20% | Vague notes | Structured guide, reusable | Professional guide with parameters, examples, tips |
| Personal initiative | 15% | Strict copy of examples | Adapted to chosen context | Original additions, creative enrichments |
During AI-assisted content creation, what is the most critical step for content quality?