Earnings and Presentation Transcript Analysis
Extract performance and guidance from transcripts and track language sentiment and term frequency use across periods
Last updated: 29 October 2025 (v4)
Objective:
This prompt provides analysis of earnings or other presentation transcripts to extract all key information from management and Q&A, isolate guidance and forward looking statements and compare how they change over time including performing sentiment and term frequency analysis.
Explanation:
An investor’s time is best spent on signals that separate insight from noise. Reading long transcripts, let along multiple of them to understand how the story of a company is evolving is very time consuming and necessarily covers a lot of information that is only marginally relevant and possibly distracting. This prompt equips investors to extract and focus on the signal by structuring a transcript-only analysis of a company’s last earnings calls.
It extracts the hard numbers, revenue, margins, free cash flow, EPS, and growth rates, while also tracing how management’s guidance and long-term goals shift from quarter to quarter. Beyond the figures, it scores sentiment in both prepared remarks and Q&A, surfacing whether management’s tone grows more confident or cautious. Finally, it distills recurring themes such as growth drivers, risks, product and strategy, so an investor can see how the story evolves across periods. The result is a compact, comparable view of performance, outlook, sentiment, and risk that helps investors judge whether management’s narrative is strengthening, weakening, or diverging from the numbers.
As always, be aware that models can make mistakes. At each step, examine the response and challenge information or conclusions that appear erroneous before proceeding to any subsequent steps. If in doubt use a second model with the same prompt to verify the information and generate challenge questions and answers (CoVe process) to correct interpretations of data.
Link to blog post explanation:
Preferred Model(s):
ChatGPT-5+ (preferred) or Gemini 2.5 FLASH (Note that Gemini PRO often crashes when analyzing 3 transcripts. The lower parameter Flash model handles the request and produces good quality results)
Important Execution Notes:
Insert ticker, exchange and company name where indicated under inputs.
Attached last X transcripts (default is three).
Tip: When you attach the transcripts, name each file with the period (e.g., Company_Q1FY25_transcript.pdf), so the tables can reference them cleanly.The default is set to the last 3 transcripts (this provides information on the evolution of sentiment and guidance). However, the user can vary this as desired.
Sample Output:
Copy/Paste Prompt Set:
Important note: Subscribers can use this prompt set for their own analysis. However, the prompt is copyrighted by The Inferential Investor, paywalled, and must not be shared without permission.