Why Presentations Fail (And How to Fix Them with AI)

TL;DR: Most AI presentation tools focus on formatting rather than generating meaningful business insights, which is why many AI-powered decks fail to persuade.

• 📹 Watch the Complete Video Tutorial
• Why Presentations Fail: The Critical Insight Gap
• The Three-Step AI-Powered Presentation Framework
• Step 1: Deep Research with AI – Generating Real Insights
• How to Set Up Deep Research Correctly
• Real Output: Indonesia Ranks #1
• Step 2: Structuring the Story – From Report to Narrative
• Switching Off Deep Research
• Prompt Highlights for Story Structuring
• AI Output: A Ready-to-Use Slide Outline
• Step 3: Design with Gamma – Polishing the Presentation
• Importing Your Outline into Gamma
• Gamma Settings for Best Results
• Selecting a Professional Theme
• Human Refinement: Why 80% Isn’t Enough
• Two Critical Checks During Review
• Real Editing Examples from the Transcript
• Enhancing Visuals in Gamma: Pro Techniques
• Replacing AI Images with Stock Photos
• Adding Custom Icons
• Fixing AI-Generated Content with “Edit with AI”
• Adding New Slides with AI in Gamma
• Advanced Gamma Features You Should Use
• Embedding Interactive Elements
• Presentation Mode Enhancements
• Gamma’s Limitations and Workarounds
• 1. Export to Google Slides Breaks Formatting
• 2. Free Plan Watermark
• Tool Selection Strategy: Matching AI to Task
• The Wharton Insight on Human-AI Collaboration
• Real-World Case Study: Google Sales Pitch Recreation
• Performance Metrics: What Success Looks Like
• Resource List: Tools and Templates Mentioned
• Troubleshooting Common AI Presentation Pitfalls
• Pitfall: AI Generates Generic or Inaccurate Insights
• Pitfall: Slides Lack Strategic Focus
• Pitfall: Visuals Feel AI-Generated and Impersonal
• Future Considerations: Where AI Presentations Are Headed
• Action Plan: Your Next Steps
• Final Takeaway: Why This Method Works

AI presentation tools promise speed, polish, and automation—but they often miss the mark where it matters most: generating real business insights. In this comprehensive guide, we’ll unpack why most AI-powered presentations fall short, then reveal a battle-tested, three-step framework that combines ChatGPT (or Claude), deep research, and Gamma to build compelling, insight-driven pitch decks that actually win deals.

Based on a real-world example from a former Google salesperson, you’ll learn how to transform raw market data into a persuasive story for executives—while avoiding the common pitfalls that cause AI presentations to fail.

Why Presentations Fail: The Critical Insight Gap

Most AI presentation tools excel at one thing: formatting slides. They can make your content look professional, add animations, and even suggest layouts. But they fail at the most critical part of any business presentation: uncovering the insights that actually matter.

As the speaker explains, there’s a fundamental gap between insight generation and insight presentation. AI tools like Gamma or Beautiful.ai cannot tell you why your client should enter the US market—but once you have that insight, they can present it compellingly.

This is the core reason why presentations fail: they’re built on generic data, not strategic recommendations. Without human-led insight discovery, even the most polished deck will lack persuasive power.

The Three-Step AI-Powered Presentation Framework

To bridge the insight gap, the speaker developed a repeatable, three-phase process used to recreate a real Google sales pitch deck. This method ensures your presentation is both data-driven and story-powered.

Step 1: Deep Research with AI – Generating Real Insights

The foundation of any winning presentation is actionable insight. In this phase, the speaker used ChatGPT’s Deep Research feature (or Claude’s equivalent) to analyze market entry opportunities for a budget skincare brand targeting young adult males (ages 25–35) in Japan, Indonesia, and the United Kingdom.

How to Set Up Deep Research Correctly

Deep Research is not the same as a standard AI prompt. It requires:

• Selection of the latest reasoning model (e.g., GPT-4o or Claude 3.5 Sonnet—versions change rapidly, so always pick the newest)
• Explicit toggling of the “Deep Research” option
• A highly specific, comprehensive prompt

The speaker provides a detailed prompt template (linked in the video description) but highlights these critical components:

1. Clear objective: “Create a market entry report for a budget skincare brand targeting young adult males (25–35) in Japan, Indonesia, and the UK.”
2. Request for clarification: Ask the AI to identify and clarify any missing inputs.
3. Use of Porter’s Five Forces: Apply this consulting framework to analyze the competitive landscape in each market.
4. Deliverable structure: Executive summary first, followed by supporting data and deep-dive analysis.
5. Ranked market scorecard: Most importantly, demand a clear recommendation—which market should we prioritize?

Real Output: Indonesia Ranks #1

After ~7 minutes of processing, ChatGPT delivered a detailed report with:

• A comparison table of all three markets with actual metrics
• Deep-dive analysis per country
• A ranked market scorecard placing Indonesia first

Key actionable insights for Indonesia included:

• Formulate products for tropical climates and oily skin
• Obtain halal certification to build consumer trust
• Prioritize digital channels like Shopee and Lazada

Step 2: Structuring the Story – From Report to Narrative

A research report is not a presentation. The goal now is to package facts into a compelling story that convinces the CMO to act.

Switching Off Deep Research

In the same ChatGPT chat thread, the speaker deselected Deep Research and used a new prompt to generate a presentation outline. This leverages the standard language model’s strength in narrative construction—not data retrieval.

Prompt Highlights for Story Structuring

The prompt instructed ChatGPT to:

• Prepare a presentation for the CMO of a budget skincare company
• Recommend Indonesia as the top market (based on Step 1 findings)
• Reduce decision fatigue by making the recommendation clear and upfront
• Dedicate one slide to explain why Japan and the UK are deprioritized
• Use the SEQA framework (Situation, Evaluation, Question, Answer) or similar consulting storytelling structure
• Limit each slide to three supporting bullet points or insights

AI Output: A Ready-to-Use Slide Outline

The result was an 8-slide outline with:

• Bottom-line-up-front (BLUF) on Slide 1: “Enter the Indonesia market because it offers the highest ROI.”
• Supporting data: Indonesia’s skincare market = $9 billion, growing at 8% annually vs. Japan (4%) and UK (2%)
• Tactical insights: e.g., “Search volume for ‘acne wash pria’ (men’s acne wash) is up 54% year-over-year”
• A dedicated slide explaining why Japan and the UK are less attractive

This phase transforms dense analysis into a clear, persuasive narrative—exactly what executives need to make fast decisions.

Step 3: Design with Gamma – Polishing the Presentation

Now it’s time to turn the outline into a visually compelling deck using Gamma—a modern AI-powered presentation tool.

Importing Your Outline into Gamma

Follow these exact steps:

1. Copy the full outline from ChatGPT
2. In Gamma, click “Create new with AI”
3. Select “Paste in text”
4. Paste your outline and choose “Presentation” as the format

Gamma Settings for Best Results

During setup, configure these options carefully:

Selecting a Professional Theme

The speaker chose the “Icebreaker” theme with a light blue palette for a clean, professional look. Gamma also supports:

• Custom themes (upload your brand guidelines)
• Manual color and font adjustments

Human Refinement: Why 80% Isn’t Enough

Gamma generates a beautiful deck in ~60 seconds—but it’s only 80% complete. Human refinement is essential to ensure accuracy, clarity, and impact.

Two Critical Checks During Review

1. Does the presentation flow smoothly? Is the story logical and persuasive?
2. Was any critical information lost? Cross-check against the original research report.

Real Editing Examples from the Transcript

• Slide 1 correction: Changed “market” to “skincare market” and clarified “domestic e-commerce” based on original ChatGPT report
• Title improvement: Used Gamma’s “Edit with AI” to generate an insight-led title: “Capturing Indonesia’s 54% Year-on-Year Growth: Our Three-Step Strategy”
• Removed fluff: Deleted unnecessary sentences to keep slides concise

Enhancing Visuals in Gamma: Pro Techniques

The speaker demonstrates advanced Gamma editing to boost visual appeal:

Replacing AI Images with Stock Photos

1. Select a slide block
2. Change layout to “Square Images”
3. Click “Generate Image” → choose Unsplash
4. Search for relevant terms: e.g., “Indonesia,” “online shopping,” “skincare men”
5. Adjust focus point by dragging to center key elements (e.g., the Indonesian flag)

Adding Custom Icons

1. Go to Images → Icons
2. Search for relevant icons: “people,” “digital,” “credit card,” “shopping cart”
3. Resize and drag into place—Gamma’s layout engine handles formatting automatically
4. Copy/paste icons across columns for consistency

Fixing AI-Generated Content with “Edit with AI”

When Gamma’s auto-generated bullet point was off-target:

1. Clicked “Edit with AI” on the card
2. Instructed: “Replace the third card with a more relevant market insight”
3. After regeneration, manually restored the icon layout (AI doesn’t preserve design)

Adding New Slides with AI in Gamma

Need to insert a new data slide? Here’s how:

1. Go back to ChatGPT and copy key numbers (e.g., market sizes)
2. In Gamma, click “Add card with AI”
3. Prompt: “Create a slide comparing the absolute size of the skincare market for Japan, Indonesia, and the UK in descending order. Assert that Indonesia is the best market to enter. Use these numbers: [paste data].”
4. Gamma generates a clean, formatted table—saving hours of manual work

Advanced Gamma Features You Should Use

Embedding Interactive Elements

Like Notion, Gamma supports the forward slash (/) command:

• Type /button to add a call-to-action
• Edit button text to “Access Full Report”
• Link to your original research document
• Center-align for polish

Presentation Mode Enhancements

Enable Spotlight Mode in presentation view: it blurs the entire slide until you click, focusing audience attention—a feature the speaker “absolutely loves.”

Gamma’s Limitations and Workarounds

Despite its power, Gamma has two notable drawbacks:

1. Export to Google Slides Breaks Formatting

• Spacing issues appear
• Charts (e.g., bar graphs) don’t resize properly to fit slides
• Workaround: Treat Gamma as the final delivery format—or manually adjust in Google Slides

2. Free Plan Watermark

• A “Gamma” watermark appears in the bottom-right corner
• Workaround: Go to Slide Menu → Edit Theme and manually delete the watermark

Tool Selection Strategy: Matching AI to Task

The speaker emphasizes: not all AI tools are equal for every task. Success comes from using the right tool at the right time:

The Wharton Insight on Human-AI Collaboration

The speaker closes with a powerful quote from Professor Ethan Mollick of Wharton:

“The greatest value comes not from surrendering control entirely to AI or clinging to entirely human workflows, but from finding the right points of collaboration for each specific task.”

This philosophy underpins the entire three-step method: AI handles scale and speed; humans provide judgment, strategy, and refinement.

Real-World Case Study: Google Sales Pitch Recreation

This entire workflow was tested by recreating a pitch deck the speaker used while at Google. The original client was evaluating market entry into Japan, Indonesia, and the UK for a budget men’s skincare line.

Using the three-step AI method, the speaker was able to:

• Replicate (and improve upon) the original strategic recommendation
• Produce a polished, executive-ready deck in under 30 minutes
• Surface new insights (e.g., halal certification, Shopee/Lazada focus) that might have been missed manually

Performance Metrics: What Success Looks Like

While the video doesn’t share closed-won stats, it highlights qualitative wins:

• Time saved: 7-minute research phase vs. days of manual analysis
• Clarity gained: Clear market ranking removes ambiguity
• Actionability improved: Tactical recommendations (e.g., “get halal certified”) drive real next steps

Resource List: Tools and Templates Mentioned

• ChatGPT (with Deep Research enabled) – for insight generation
• Claude (Geminis) – alternative for deep research
• Custom GPT – generates deep research prompts from one sentence (linked in video description)
• Gamma – AI presentation builder (free plan available)
• Unsplash – for high-quality stock photos inside Gamma
• Porter’s Five Forces – strategic framework for market analysis

SEQA Framework – for structuring executive narratives

Troubleshooting Common AI Presentation Pitfalls

Pitfall: AI Generates Generic or Inaccurate Insights

Solution: Use Deep Research with highly specific prompts. Always cross-check outputs against known data.

Pitfall: Slides Lack Strategic Focus

Solution: Enforce a “bottom-line-up-front” structure. Dedicate slides to explaining why alternatives were rejected.

Pitfall: Visuals Feel AI-Generated and Impersonal

Solution: Replace AI images with curated Unsplash photos. Add custom icons. Manually adjust focus points.

Future Considerations: Where AI Presentations Are Headed

As AI evolves, expect:

• Better export fidelity to PowerPoint/Google Slides
• Integration of real-time data directly into slides
• More sophisticated storytelling frameworks built into tools

But one thing won’t change: the need for human judgment in defining the problem, interpreting results, and crafting the narrative.

Action Plan: Your Next Steps

1. Identify your next presentation (e.g., market entry, product launch, investment pitch)
2. Run Deep Research using the prompt template
3. Generate a story outline with clear recommendations
4. Build in Gamma and refine with human edits
5. Present with confidence—backed by real insights, not just pretty slides

Final Takeaway: Why This Method Works

Presentations fail when they prioritize form over substance. This three-step method flips the script: insights first, visuals second. By leveraging AI for what it does best—and applying human expertise where it matters most—you create decks that don’t just look good, but drive decisions.

As the speaker puts it: “Where was this when I was a salesperson?” Now, it’s here—and it’s ready for you to use.

Why Presentations Fail (And How to Fix Them with AI)

Buy this item