Earnings call summaries: what a good one contains
A good earnings call summary compresses an hour into a minute without losing what the hour was about. A bad one compresses it into vibes. Since AI now writes most earnings call summaries, the difference between the two has become a question of method, and it is checkable from the outside. Here is what to demand from any summary, human or machine.
The five elements
- The headline figures: revenue and earnings, with the year-over-year change, as stated on the call.
- The guidance: raised, held, or cut, and in which words. This is the forward-looking half, and often the price-moving half.
- The dominant Q&A topic: what analysts kept returning to, because that is the quarter's open question.
- One or two verbatim quotes: the sentences that carry the quarter, in management's own words.
- What was not answered: the deflections and the topics conspicuously absent.
How AI summaries fail
The classic failure is smoothing: an AI summary that reads well but averages away the tension, the hedge in the guidance, the question asked twice. The more dangerous failure is invention, a figure or claim that appears in no transcript. Both failures share a root: summaries generated from a model's general fluency rather than from the specific transcript, checked line by line.
The test is simple: does every claim in the summary trace to a quote or a queried figure? If a summary cannot show its sources, it is asking to be trusted rather than verified.
Summaries at watchlist scale
One summary is a convenience; ten per evening is a workflow. In earnings season, the question is not whether a single summary is good but whether the method holds across every name you cover, in the same structure, comparable week over week.
This is where a chat interface beats a feed: "Summarize today's call for X" produces the same five elements for every name, and any summary can be drilled into with follow-ups when something flags. earnings.chat answers each one with quotes, sources, and a coverage line, so the trust question answers itself.
From summary to comparison
The step most summaries never take: putting quarters side by side. "How does this quarter's summary compare to the last two?" turns three summaries into a trajectory, and a trajectory is where decisions live. A summary tells you what happened; the comparison tells you what is changing.
252,000+ transcripts, answers with verbatim quotes and sources, new calls within minutes.
Open earnings.chat