Coming into last February’s earnings call, Walmart executives wanted to convey a carefully balanced narrative. In the face of an uncertain economic environment, they projected strength and resilience, highlighting strong e-commerce revenue growth and robust investment in AI. Then the Q&A with investors began.
One investor asked about unanticipated costs in the coming year, citing the unpredictability of tariffs. Two others pressed management on whether consumer spending weakness among lower-income shoppers was becoming more severe than the executives had implied in their prepared remarks.
The executives did their best to stay on message, but the questions forced them to discuss consumer stress in more granular terms, including shifts in purchasing behavior, discretionary pullbacks, and margin pressures. Investors appeared to interpret the tone of those exchanges as more cautious than the headline earnings numbers suggested. Immediately after the call, Walmart’s stock prices wavered, and several analysts revised expectations and price targets.
This outcome is an example of how unanticipated questions can create moments where executives leak signals—through pauses, evasions, or improvised elaboration—that investors interpret as meaningful. And a new research paper from Columbia Business School Assistant Professor Wei Cai, Columbia University PhD candidate Nandil Bhatia, and University of California Berkeley Professor Sameer B. Srivastava shows that these “curveball questions” are measurably associated with larger stock-price movements and analyst-rating changes.
For leaders, the findings offer a reminder that audiences may pay as much attention to how they answer questions as to the presentation itself. And for audiences, the research highlights the strategic power of asking better questions. “The right curveball doesn’t just interrupt the narrative,” Cai says. “It can reshape it.”
Putting curveballs under the microscope
Cai and her team started by defining just what a curveball question is. Arguing that questions vary along two dimensions, predictability and topicality, they defined curveball questions as those that are topical but unpredictable. Because they are relevant to the topic at hand, they can’t be easily ignored, but because they are difficult to anticipate, they often force spontaneous responses that may reveal new information.
The authors analyzed more than 126,000 quarterly earnings-call transcripts from publicly traded U.S. companies between 2011 and 2021. To identify curveball questions, they used an AI language model trained specifically on financial texts that allowed them to measure how relevant a question was to the discussion already underway, as well as how difficult it would have been for management to anticipate it.
They then created a call-level curveball measure based on the highest-scoring curveball question among analysts’ opening questions, reflecting the idea that a single, especially disruptive question can shape how investors interpret the call. The authors compared the score with key financial market outcomes the same day after the call, finding that calls with a higher maximum curveball-question score were associated with larger same-day absolute stock returns, larger absolute abnormal returns, and a higher likelihood of changes in average analyst ratings.
To validate their findings, the researchers checked whether high-curveball questions actually rattled executives. They did: executives gave longer answers, dodged questions more often, and were more likely to have the CEO step in.
“People want to know whether these kinds of questions matter,” Cai says. “And the answer is yes; investors do react to how executives handle curveballs.”
The broader implications of curveballs
The implications of the study extend far beyond quarterly earnings calls. Curveball questions can emerge in almost any evaluative situation, including job interviews, startup pitches, political press conferences, or academic talks. In all those settings, one side is trying to project confidence and control while the other is probing for information that may not appear in the prepared script.
That helps explain why curveballs can feel so destabilizing. A polished presentation demonstrates preparation. But an unexpected, on-topic question tests adaptability and poise. How someone responds in those unscripted moments can reveal gaps in knowledge—or genuine expertise.
The study also suggests that transparency is not just a matter of disclosure rules or carefully managed messaging, but can depend on the quality of the questioning itself. The researchers found that “star” analysts and highly connected analysts were especially likely to ask curveball questions, while newer analysts sometimes surfaced novel lines of inquiry precisely because they weren’t locked into conventional assumptions.
By their nature, curveball questions can’t be predicted, but presenters can try to reduce their chances of getting one by being as thorough as possible in the information they share both before and during their presentations. Once a curveball has come their way, though, presenters would do well to remain calm.
“If there's a long pause while you prepare to answer, this could cause you to panic, which would have a negative consequence for how your presentation is received,” Cai says. “Instead, think of pauses in response to curveball questions as a regular thing. Then you have more capacity to think on the spot.”
