Most investors listen to earnings calls the wrong way. They track headline EPS and revenue against consensus, note whether guidance was raised or lowered, and move on. That approach leaves most of the signal on the table.
Academic research and buy-side practice converge on the same finding: the language management uses during an earnings call contains information that the market takes weeks, sometimes quarters, to fully price in. A 2024 study in the Journal of Accounting and Economics found that conference call tone has highly significant explanatory power for abnormal returns not just in the days after a call, but across the following 60 trading days. The numbers are the news. The language is the edge.
Here are five things worth tracking on every call.
1. Guidance specificity — and how it has changed
There is a meaningful difference between "we expect revenue of $4.2 billion to $4.4 billion in Q3" and "we remain cautiously optimistic about the second half." The first is a commitment. The second is a hedge with a positive framing.
The pattern that matters most is the change in specificity over consecutive calls. A company that guided to "gross margin expansion of 80 to 120 basis points" last quarter and this quarter says "we expect margins to remain broadly stable" has communicated something meaningful — even if no analyst flags the language shift. Research from the University of Chicago Booth School of Business found that greater use of euphemistic language in earnings calls is associated with lower stock prices in the following quarter, even after controlling for the disclosed financial results.
Listen for: numerical ranges shrinking or disappearing entirely. Modal verbs (hope, believe, expect) replacing factual statements. Guidance framed as aspiration rather than commitment.
2. Tone distance between the CEO and CFO
This one is underappreciated. A 2025 paper published in the Financial Review introduced the concept of "tone distance" — the measurable divergence in sentiment between the CEO and CFO during the same earnings call. The finding: greater tone divergence is negatively associated with stock returns around the earnings announcement, and positively predicts volatility and operational risk in subsequent quarters.
In practice, this means listening to whether the CEO and CFO are telling the same story. A CEO opening with confident language about growth while a CFO hedges on cost outlook and working capital is a divergence worth noting. So is the inverse: a CFO who sounds more positive than the CEO about near-term financial results while the CEO remains cautious about the market environment.
You don't need a sentiment model to catch this. Just pay attention to the register each executive uses when they discuss the same topics in their respective prepared sections.
3. What disappears between quarters
Removal is often more informative than addition. Management teams don't accidentally forget to mention a metric they've been highlighting every quarter. When a KPI that was prominently cited for three consecutive calls suddenly disappears from prepared remarks, that absence is data.
Compare these three versions of the same metric: "Gross margin expanded 120 basis points year-over-year." vs. "Gross margin was resilient given the input cost environment." vs. no mention of gross margin at all. Each step represents a further withdrawal from a positive narrative the company was previously building.
The same applies to forward-looking language. If a company repeatedly discussed its "path to profitability" and that phrase is absent this quarter, it's worth asking why. Harvard Business School research on managerial tone changes found that when managers shift their language on a metric, analysts and investors who catch it update their expectations significantly — but most don't catch it without specifically looking.
4. Response quality in the Q&A session
The prepared remarks are controlled. The Q&A is where you actually learn something.
Genuine confidence produces short, quantitative answers. A CFO who knows their margin trajectory answers in two sentences and adds a data point that wasn't in prepared remarks. A CEO who is confident about competitive positioning answers competitive questions without pivoting to strategy or referencing a press release.
Research published in Management Science specifically studied evasive responses in earnings calls and found that executives regularly use linguistic deflection to avoid difficult questions without technically lying — and that investors who recognize this update their assessments more accurately than those who don't. Deflection patterns are consistent across companies: restating the question before answering it, pivoting to a different metric, citing prepared remarks as if that constitutes an answer, or providing a long qualitative response to a quantitative question.
The signal isn't that management deflects — they always will on sensitive topics. The signal is deflection on a question they answered directly last quarter.
5. Linguistic complexity as a signal
Research from the American Accounting Association found that a manager's linguistic complexity during a conference call carries information about their private knowledge of future earnings. Critically, the type of complexity matters: complexity that adds genuine nuance and precision (informative complexity) is a positive signal, while complexity that obscures straightforward information (obfuscatory complexity) is a negative one.
In practice, this means paying attention when an executive gives an unusually long, dense answer to a simple question. "What is your Q3 revenue guidance?" should have a one-sentence answer. When it doesn't — when the response is three paragraphs of context-setting before a number finally appears, or doesn't appear at all — that complexity is itself informative.
The University of Chicago Booth also found that vocal delivery quality correlates with market reaction: an increase of one standard deviation in vocal delivery quality across speakers led to a nearly 6% increase in abnormal trading volume. Hesitation, mumbling, or unusual pausing on specific topics are worth noting — not as personality judgments, but as indicators of where confidence is and isn't present.
None of this requires a finance degree to apply. It requires listening to the call with a specific framework, comparing what you hear to what was said last quarter, and treating language as data — not just context for the numbers. That's the edge buy-side analysts have always had in earnings season. It's learnable.