The AI Earnings-Call Analysis Prompt
The exact prompt buy-side analysts paste to turn a 60-minute transcript into a one-page teardown — guidance, tone, and consensus divergence. Model-agnostic (Claude, ChatGPT, Copilot, Gemini).
For: Buy-side / sell-side analysts · PMs · IRTo analyze an earnings call with AI, paste the full public transcript into your model and ask for a structured teardown, not a summary. A working prompt instructs the model to: (1) list forward-guidance changes versus the prior quarter (raised / cut / reaffirmed) with the exact quote; (2) flag capex and R&D spend signals; (3) identify management tone shifts with evidence; (4) surface anything repeated or dodged; (5) score divergence from current sell-side consensus 1–10; and (6) flag language that historically precedes a guidance cut. The critical instructions are “quote the transcript for every claim” and “if a section has no signal, say ‘no change.’” Always paste only public transcripts — never material non-public information — and cross-check every figure against the original filing before it informs a decision. The same prompt works identically in ChatGPT, Claude, Copilot, or Gemini.
A quarter generates 500+ transcripts in a six-week window. Most analysts read 20–30 closely and skim the rest. An LLM reads all of them in the time it takes to get coffee — but only if you prompt it like an analyst, not like a search box. The difference between a useful teardown and a generic recap is the structure you impose and the discipline you force (quote everything, flag the absence of signal). Here is the exact workflow.
The workflow, step by step
- Get the public transcriptPull the transcript from the company IR site, EDGAR, or a transcript service. Public only — this is the compliance line that matters.
- Start a fresh chat and paste itA clean context window avoids cross-contamination from earlier chats. Paste the full transcript, then the prompt below in the same message.
- Run the structured teardown promptThe prompt forces guidance deltas, tone, capex/R&D signals, dodged questions, and a consensus-divergence score — each tied to a quote.
- Pressure-test the outputAsk a follow-up: “Which of these would a bear on this name attack first, and how would management defend it?” This surfaces the variant view.
- Verify before it leaves your deskCross-check every quoted figure against the transcript and the filing. AI can misattribute a number — you own what circulates.
The exact prompt (copy-paste)
Model-agnostic. Swap the [BRACKETS] and paste your source material into the same chat.
You are a buy-side equity analyst. Here is the earnings-call transcript for [TICKER], quarter [Q/FY]. Produce a one-page teardown: 1. Forward guidance changes vs the prior quarter (raised / cut / reaffirmed) with the exact quote. 2. Capex & R&D spend signals and any change in tone about them. 3. Management tone shift vs last call (defensive / confident / evasive) with evidence. 4. Anything management repeated or dodged. 5. Score divergence from current sell-side consensus 1-10 and say why. 6. Flag any language that historically precedes a guidance cut. Output as a bulleted brief. Quote the transcript for each claim. If a section has no signal, say "no change."
Get all 7 workflows in one pack
This is one of seven paste-ready workflows in The AI Finance Workflows Pack — the exact prompts, pro moves, and compliance guardrails finance pros run at their desk. Free.
Instant access on the next page + the free weekly brief. Unsubscribe anytime.
Prefer to browse first? Open the Workflows Pack → · Finance Prompt Library →
Frequently asked
What is the best AI prompt to analyze an earnings call?
The most useful prompt asks for a structured teardown rather than a summary: guidance changes vs prior quarter (with quotes), capex/R&D signals, management tone shifts, dodged questions, a 1-10 consensus-divergence score, and any language that historically precedes a guidance cut — with the instruction to quote the transcript for every claim. Paste the full public transcript in the same message. This content is educational and not financial advice.
Can ChatGPT or Claude summarize an earnings call accurately?
They can, provided you paste the primary-source transcript directly rather than asking the model to retrieve it. Models rely on training data with a cutoff and can hallucinate figures, so always cross-check every number against the transcript and the filing before use. Long transcripts may exceed the context window and need to be split into segments. Do your own research (DYOR).
Is it safe to paste earnings data into AI tools?
Public earnings transcripts and filings are fine to paste. Never paste material non-public information, draft internal numbers, or confidential deal data into consumer AI tools — that can breach both securities law and your firm’s policy. When in doubt, sanitize or omit.