# S&P 500 AI Impact Report - Data Dictionary ## Dataset Overview This dataset contains analysis of 6 quarters of S&P 500 earnings transcripts (Q1 2024 - Q2 2025), examining how companies discuss and implement artificial intelligence initiatives. The data was compiled by analyzing 503 corporate entities listed on the S&P 500 in December 2024. **Total Records:** 3,019 company-quarter observations **Time Period:** Q1 2024 through Q2 2025 **Companies Covered:** 503 S&P 500 entities **Data Source:** FinancialModelingPrep API earnings transcripts **Analysis Method:** Claude Sonnet 3.5 LLM analysis with structured categorization --- ## Column Definitions ### Company Identification **Symbol** - **Type:** String - **Description:** Stock ticker symbol for the company - **Example:** AAPL, MSFT, GOOGL **Security** - **Type:** String - **Description:** Full legal name of the company - **Example:** Apple Inc., Microsoft Corporation **GICS Sector** - **Type:** String (Categorical) - **Description:** Global Industry Classification Standard (GICS) primary sector - **Values:** Information Technology, Health Care, Financials, Consumer Discretionary, Industrials, Materials, Energy, Utilities, Real Estate, Communication Services, Consumer Staples - **Purpose:** High-level industry grouping for sector analysis **GICS Sub-Industry** - **Type:** String (Categorical) - **Description:** Detailed GICS sub-industry classification - **Example:** Application Software, Biotechnology, Property & Casualty Insurance - **Purpose:** Granular industry analysis and comparison --- ### Time Period **Quarter** - **Type:** String - **Description:** Calendar quarter of the earnings call - **Format:** Q[1-4] YYYY - **Range:** Q1 2024 to Q2 2025 - **Example:** Q4 2024 **Mapping** - **Type:** String - **Description:** Unique identifier combining ticker symbol and quarter - **Format:** [SYMBOL]Q[1-4] YYYY - **Example:** AAPLQ4 2024 - **Purpose:** Unique key for each company-quarter observation --- ### AI Implementation Classification **Mapping** (AI Use Category) - **Type:** String (Categorical) - **Description:** Primary classification of how the company engages with AI - **Values:** - **"Using AI Technology"**: Company deploying AI tools for internal projects or operations - **"Creating AI Technology"**: Company building large-scale AI services to sell (e.g., Google, Microsoft, OpenAI) - **"Supporting AI Infrastructure"**: Company providing hardware/infrastructure for AI (e.g., NVIDIA, cloud providers) - **"Benefiting from AI Market Changes"**: Company experiencing indirect benefits from AI growth (e.g., utilities, data centers) - **"No AI Discussion"**: No mention of AI in earnings transcript - **Empty/Null**: Quarter predates analysis period or data not available - **Note:** This report primarily focuses on "Using AI Technology" category **AI Use Category** - **Type:** String (Categorical) - **Description:** Refined categorization of AI usage type (duplicate/confirmation of Mapping column in some cases) - **Same values as Mapping column above** --- ### Implementation Details **Implementation Status Category** - **Type:** String (Categorical) - **Description:** Stage of AI initiative deployment - **Values:** - **"Planning"**: AI project discussed but not yet started - **"Early Implementation"**: Initial rollout or pilot phase - **"Launched"**: Fully deployed and operational - **Empty/Null**: Status not specified or no AI discussion **App Type** - **Type:** String (Free text) - **Description:** Specific description of the AI application or tool being used - **Example:** "AI assistant in Adobe Acrobat", "GenAI for client projects", "AI-powered customer support" - **Note:** Extracted directly from earnings transcript analysis **Business Unit** - **Type:** String (Categorical) - **Description:** Department or function deploying the AI initiative - **Values:** R&D, Operations, Marketing, Sales, IT, Customer Service, Engineering, Human Resources - **Purpose:** Identify which departments are leading AI adoption --- ### Application Characteristics **App Type** (Second column) - **Type:** String (Categorical) - **Description:** Standardized category of AI application type - **Values:** Generative AI, Customer Service, Product Enhancement, Data Analytics, Enterprise Services, Business Process, etc. - **Note:** Appears to be a duplicate column with standardized categorization **User Focus** - **Type:** String (Categorical) - **Description:** Whether AI tool faces customers or internal operations - **Values:** - **"Front-End Consumer Facing"**: AI tools directly used by customers/end-users - **"Back Office"**: AI tools for internal operations and efficiency - **Empty/Null**: Not specified --- ### Return on Investment (ROI) **AI Return Primary** - **Type:** String (Categorical) - **Description:** Type of measurable impact or return mentioned - **Values:** - **"Operational Efficiency & Throughput"**: Productivity improvements, faster processing - **"Cost Reduction & Spend Optimization"**: Direct cost savings - **Empty/Null**: No ROI mentioned (scored as 0 for analysis) - **Scoring Method:** Companies mentioning ROI receive score of 1; no mention = null - **Purpose:** Calculate percentage of companies reporting measurable AI impact **Return Summary** - **Type:** String (Free text) - **Description:** Specific ROI metrics or impact statements extracted from transcripts - **Example:** "60% productivity increase and 50% cost reduction", "$100 million+ in annual OpEx reduction" - **Note:** Qualitative or quantitative results as reported by companies --- ### Strategic Classification **Strategy Category** - **Type:** String (Categorical) - **Description:** High-level strategic approach or technology area - **Values:** Document Automation, Voice AI, Generative AI Services, Data Analytics, etc. - **Purpose:** Group similar AI strategies across companies --- ## Data Quality Notes ### Methodology 1. **Source Reliability**: Data extracted from official SEC-filed earnings transcripts 2. **Analysis Method**: Claude Sonnet 3.5 LLM with structured prompts 3. **Validation**: Spot-checked for inconsistencies and hallucinations 4. **Constraint**: LLM instructed to analyze only transcript content, no external data ### Limitations 1. **Self-Reported Data**: Based on what companies chose to disclose in earnings calls 2. **ROI Completeness**: Only ~minority of companies provide quantitative ROI metrics 3. **Categorization**: Some subjective judgment in classification despite systematic approach 4. **Coverage Gaps**: Empty cells indicate either no AI discussion or data not captured 5. **Temporal Bias**: More recent quarters may show higher AI adoption due to increased awareness ### Missing Data Patterns - **Empty cells in Q1-Q3 2024**: Many early quarters have sparse data as companies weren't yet discussing AI extensively - **Q4 2024 - Q2 2025**: More complete data as AI discussion became more common - **ROI fields**: Majority are null/empty - reflects reality that most companies don't publicly share specific metrics --- ## Usage Recommendations ### For Researchers - Filter to **"Using AI Technology"** for internal adoption analysis - Use **ROI metric** as boolean indicator (mentioned vs. not mentioned) rather than continuous variable - Sector comparisons should account for sample size differences - Consider temporal trends when analyzing adoption patterns ### For Business Analysis - **Implementation Status** provides pipeline of AI maturity - **Business Unit** shows which departments lead adoption - **User Focus** distinguishes customer-facing vs. operational AI - **Strategy Category** enables benchmarking against similar approaches ### For Reproducibility - All categorizations traceable to original earnings transcripts - FinancialModelingPrep API provides source data - Claude Sonnet 3.5 prompts available in methodology documentation - Contact researcher for clarification on specific categorizations --- ## Citation If using this dataset, please cite: **Title:** S&P 500 AI Impact Report **Author:** Ben Magnuson **Period Covered:** Q1 2024 - Q2 2025 **Publication Date:** 2025 **URL:** https://www.bmagnuson.com --- ## Contact & Support **Researcher:** Ben Magnuson **Website:** bmagnuson.com **Questions:** questions@bmagnuson.com For methodology details, see: Methodology --- ## Version History **Version 1.0** (2025) - Initial release covering Q1 2024 - Q2 2025 - 503 S&P 500 companies - 3,019 company-quarter observations --- ## License This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). You are free to: - Share — copy and redistribute the material - Adapt — remix, transform, and build upon the material Under the following terms: - Attribution — You must give appropriate credit to Ben Magnuson, provide a link to the license, and indicate if changes were made. Full license: https://creativecommons.org/licenses/by/4.0/