Generative AI, Demos, and Data – Turning Research into Action

Design Phase – Part 2 – Presentation Structure

Series co-authors: Loren Sylvan & Dan Axman

Transitioning from data element design to presentation flow

You’ve done the research—now it’s time to design your presentation/demo. How do you go from insight to impact?

In Part 1 of the Design Phase, Loren and I focused on defining data elements to create structured, synthetic datasets tailored to the needs of our 3PL prospect. In Part 2, we turn insights into action by showing how we worked with ChatGPT to help craft a cohesive presentation and demonstrate how SAS and SAS Viya address 3PL challenges and deliver strategic business insights.

Creating a presentation that resonates with a prospect or customer involves more than assembling content, screenshots, and software demos; it requires an engaging, choreographed journey that tells a story that resonates with the audience. Often, the full capabilities of a system are not leveraged due to time constraints or the breadth of functionality. What features should we emphasize? What visual components are most effective? These difficult questions are often overlooked.

In our collaboration with ChatGPT, we outlined the key components and structure of our presentation and began developing visual screen mockups to accompany our demo—bringing our data and insights to life.

Structuring the Demo Flow for Impact

A cohesive and relevant demo requires a structured journey. With help from ChatGPT, we broke it down into sections: starting with an overview of the current business and industry landscape, transitioning into SAS Viya’s capabilities, and closing with a "real-life" demo using industry-relevant data. ChatGPT helped craft a compelling narrative that was both customer-focused and solution-oriented.

• Key Prompt: How can we create a logical presentation flow that tells a compelling story for our prospect?
• Best Practice: Create a narrative that starts with a problem statement, demonstrates value through capability highlights, and ends with practical solutions.

We mapped out the journey of the demo, focusing on:

• Establishing Context: Begin with a clear overview that frames the client's challenges and industry trends.
• Building Complexity Gradually: Move from high-level insights to deeper predictive analytics.
• Ensuring Clarity: Each page or screen must have a clear purpose that moves the story forward without overwhelming the viewer.

Presentation Flow Example:

Based on insights from the Research phase and our work on data element design, ChatGPT suggested a presentation structure tailored to the prospect’s needs and time constraints. Here’s what was generated as the recommended sequence for a one-hour presentation:

Introduction and Voice of the Customer (VOC)

• Objective: Set the context by aligning with the Company’s goals and challenges.
• Time: 5 minutes
• Key Points: Summarize the Company’s 3PL challenges, positioning the relevance of SAS Viya.

Relevant SAS Use Cases in 3PL & Related Industries

• Objective: Showcase practical examples of how SAS has solved similar challenges for other 3PL or related companies.
• Time: 5 minutes
• Key Points:

Story #1: A customer analytics use case demonstrating improved customer retention.

Story #2: A logistics optimization case, showcasing how real-time tracking enhanced operational efficiency.

Story #3: A forecasting example for inventory or demand planning.

Overview of SAS Viya for 3PL

• Objective: Present SAS Viya as the backbone for real-time and predictive insights.
• Time: 5 minutes
• Key Points: Position SAS Viya as the complete analytics platform for 3PL, tailored to solve challenges like those in the use cases.

Operational Insights: Real-Time Data & Predictive Maintenance

• Objective: Highlight how Viya enhances operational efficiency.
• Time: 10 minutes
• Key Points: Show dashboards for real-time tracking and predictive maintenance.
• Deliverable: Mock dashboard with fleet and maintenance analytics.

Customer Analytics: Satisfaction, Retention & Churn Prediction

• Objective: Illustrate customer-focused insights and churn reduction.
• Time: 10 minutes
• Key Points: Present metrics on customer satisfaction and predictive churn.
• Deliverable: Visualization of customer satisfaction and churn insights.

Strategic Benchmarking and Financial Analysis

• Objective: Demonstrate competitive benchmarking to inform strategic decisions.
• Time: 10 minutes
• Key Points: Show the Company’s position relative to competitors in cost and efficiency.
• Deliverable: Benchmarking dashboard with relevant 3PL metrics.

Forecasting and Scenario Planning with Predictive Models

• Objective: Emphasize predictive analytics for future planning and financial impact.
• Time: 10 minutes
• Key Points: Showcase demand forecasting and scenario planning for 3PL. Deliverable: Forecasting and model prediction screens for demand fluctuations.

Data Integration and Automation

• Objective: Highlight Viya’s data automation and integration capabilities.
• Time: 3 minutes
• Key Points: Show the ease of integrating multiple data sources and real-time reporting.
• Deliverable: Screens summarizing automation and data integration in Viya.

Closing Remarks: Key Points to Reinforce

• Objective: Wrap up with a recap and strong call to action.
• Time: 2 minutes
• Key Points: Summarize the value of SAS Viya in addressing the Company’s needs, financial benefits, increased profitability, and ROI.
• Deliverable: Summary slide with next steps and tailored SAS Viya benefits.

Designing the Screens for Visual Impact

Visuals should tell a clear story at every stage of the demo. Our goal was to avoid overloading the screens with too much data or analysis while still making the insights clear. We worked with ChatGPT to create a set of screen mockups, focusing on key metrics, with easily understandable charts and a minimalistic design.

• Key Prompt: What insights must the screen communicate, and how can we best visualize them for clarity and impact?
• Best Practice: Keep visualizations simple and focused on the primary insight to avoid overwhelming the audience.

Storytelling with Screenshots

ChatGPT generated report and dashboard concepts for our presentation slides that illustrated key metrics, like on-time delivery rates, transportation costs, and inventory levels, helping us visualize the story we wanted to discuss with the client. These visuals provided a starting point that we collaboratively refined.

• Collaborative Data Refinement: Our discussions highlighted how asking the right questions could make all the difference in the data design process. ChatGPT suggested variables, visual elements, and relevant metrics, which we iterated upon multiple times.
• New Insight: The demo flow and screen design process was outstanding. Instructing ChatGPT to build the images with the look, feel, and capabilities of SAS Viya was spot-on, and the iteration process where we asked for simplification in cases when a report was too busy worked to perfection! ChatGPT also proactively suggested new reports and analytics based on data not in the integrated dataset, adding incredible value to our final output.

Examples from Interactive Design Iterations:

• Simplifying the Overview Dashboard: Initially, we created a complex reports with multiple pages, KPIs, dropdowns, sliders, and filters. After reviewing it, ChatGPT simplified the interactive elements, retaining only the most essential components—such as a time slider for trend charts and a dropdown for segment-specific analysis. The goal here is not to have ChatGPT build the demo. The goal is to create a compelling flow, content, and visual layout. ChatGPT's iterative design of actual screens, based on our feedback on complexity and readability, was one of the surprising areas where ChatGPT excelled.
• Adding Competitive Intelligence to the Demo: It became clear that adding a competitive intelligence element tailored to our 3PL prospect was key. A prompt like "Can you design an industry-relevant competitor intelligence screen in the style of SAS VA?" again led ChatGPT to suggest impactful layouts, including KPI tiles, competitor comparisons, and risk analysis heatmaps. We iteratively simplified these designs to focus on clarity and relevance.
• Competitive Intelligence Dashboard – Before and After. The first iteration was busy and did not adhere to the SAS color scheme. The second iteration was simpler and illustrated exactly what the report would look like in SAS VA.

Another ChatGPT accomplishment was identifying which predictive analytics would be most impactful for a 3PL company, what the SAS Model Studio flow would look like, what variables to include, and how the predictive insights would benefit the prospect. Although it didn't design the screens, it did show how the demo would look in SAS Model Studio. This also illustrates how an initial representation of model comparison might be represented in SAS Model Studio.

Key takeaway: On the topic of using ChatGPT for dashboard/report prototype images, we found it’s a good prototyping tool due to the inherent limitations of image creation that include text within the images. Keep in mind that when starting with Dall-E 3 outside the context of a thread within ChatGPT 4.0, Dall-E 3 can create images that in many cases, may be better (in our opinions).

Also, we found that ChatGPT has difficulty with readable text and spelling in the context of images, but that wasn’t a showstopping issue since we ultimately must build the actual demo. The images are simply designs to guide us. Overall, we wanted to illustrate that relevant design can be achieved with the use of LLM’s.

Iterating on Feedback

An iterative approach to presentation design was crucial. Each run allowed us to refine the narrative. ChatGPT was invaluable in applying the foundation model to all interactions, and our own observations helped enhance each aspect, from screen flow to data presentation.

• Best Practice: Treat every demo iteration as a learning opportunity to improve the overall experience.
• Iterative Improvement of Screens and Flow: We designed multiple versions of the demo screens to reduce redundancy and focus each page on unique content. ChatGPT's suggestions for visual layout and flow were particularly valuable, as they allowed us to consider various approaches to structuring information and user interaction.

Lessons Learned & Key Takeaways:

The Power of a Clear Narrative: ChatGPT continued to help us refine our story by aligning the presentation flow with the insights gathered in earlier phases. By using targeted prompts like "What should the presentation highlight for a 3PL company focusing on scalability and efficiency?" it surfaced clear narrative arcs that ensured every section of the presentation added value.

Customization Through Collaboration: Creating a presentation that resonates with multiple stakeholder personas—executives, technical staff, and business leads—requires collaboration and customization.

Streamlined Visual Planning: ChatGPT generated structured outlines and design ideas for visuals, offering suggestions like including trend analysis and competitor benchmarking graphs. It also helped ensure that visuals corresponded to the prospect's pain points and supported a cohesive narrative.

Iterative Refinement: ChatGPT allowed us to revise presentation elements based on feedback quickly. For example, when initial visuals didn’t strongly link to financial outcomes, ChatGPT refined the suggestions to emphasize ROI-focused metrics and predictive modeling.

Focus on Engagement: ChatGPT suggested using anecdotes and real-world use cases, which made the presentation more relatable. For example, it provided storylines around predictive maintenance and customer churn models that directly addressed 3PL industry challenges.

Summary:

In Part 2 of the Design Phase, Loren and I collaborated with ChatGPT to design a presentation structure that aligned SAS Viya's analytical and modeling capabilities with the needs of our 3PL company prospect. By layering impactful visuals with a clear narrative, we created a presentation structure that captures attention and demonstrates SAS Viya's transformative potential.

Looking Ahead

With the Design phase complete, we’re ready to move on to the next step: Creating the software Demo. In the next article, we’ll expand upon how we used ChatGPT for synthetic data creation, report building, predictive models, and finalizing the demo flow.

Have you used Generative AI to enhance your demo creation process? Share your stories and best practices in the comments!

Articles in this Series:

• Embracing Generative AI: Transforming Software Demos from Good to Exceptional (6/24/24)
• Generative AI, Demos and Data: Going from Cool to Useful? (9/14/24)
• You’re tasked with delivering an exceptional software demo. Now what? Research! (9/30/24)
• Generative AI, Demos, and Data – Turning Research into Action - Design Phase - Part 1 – Data Elements & Design (11/6/24)