If you want to uncover key insights, you might as well try this set of questioning methods.

We have all learned how to ask sharp questions. Interviewers are trained to use open-ended questions to break through the rehearsed responses of the interviewee. When employees give vague and general statements, managers will press them for details. Facing respondents who avoid core issues, journalists and negotiators will follow up with continuous questions. All these practices share the same goal: to dig out information that the other party is reluctant to disclose.

However, questions that sound sharp often fail to elicit candid answers. Experienced respondents can anticipate such questions, prepare a set of polished remarks, and steer the conversation to the direction they want to talk about. They can talk at length while revealing almost no substantive information. Even a highly challenging question may not dig out the truth.

This study, recently published in Strategic Management Journal, proposes an alternative approach: Ask questions that respondents cannot anticipate but are capable of answering, which we call curveball questions. Such questions break the respondents' preset response scripts, making it difficult for them to apply ready-made remarks, and their answers can better reflect their real thinking logic.

This study analyzes curveball questions in the scenario of earnings calls, where corporate executives first make prepared statements and then answer questions from financial analysts. However, the logic behind curveball questions can be fully applied to more scenarios: board meetings, venture capital roadshows, job interviews, and corporate strategic reviews.

The Value of Curveball Questions

We analyzed a total of 126,910 earnings call transcripts from 5,609 listed U.S. companies between 2011 and 2021. Executives give prepared presentations during the meetings and then respond to questions in real time. For each question raised by analysts, we measured two indicators: the first is unpredictability, which measures the difficulty of anticipating the question from the management's prepared speech draft; the second is topic relevance, which measures how closely the question is related to the content of the aforementioned speech. Questions that score high on both indicators are curveball questions.

The study shows that the higher the curveball score of a question, the more likely it is to make executives flustered. Executives will respond with longer remarks and are more prone to giving irrelevant answers; such questions often prompt the CEO to answer in person instead of delegating the response to the head of the corresponding business line. From the perspective of the entire meeting, curveball questions are often accompanied by greater absolute fluctuations in stock prices and changes in excess returns, and analysts are more likely to adjust their average ratings. This pattern indicates that curveball questions can dig out new information and push analysts and the market to re-evaluate the development prospects of the enterprise.

We also examined scenarios where curveball questions are more likely to arise. Star analysts and analysts in core positions on professional social networks are better at asking such questions than their peers. This shows that curveball questions are a commonly used tool for senior and experienced researchers. Curveball questions are also more likely to be raised when analysts have divergent views on the enterprise's prospects, the enterprise's business is difficult to categorize simply, or the analyst has just started covering the company.

How to Design High-Quality Curveball Questions

Before introducing the experience that managers can draw on, it is necessary to explain the limitations of this study first. The scenario of this study is rather special: the questioners (financial analysts) are highly professional and experienced, and the managers' responses will be closely scrutinized, which will directly impact financial performance. However, we verified that this conclusion also holds in another scenario — the Federal Reserve press conference, where the questioners vary slightly in professionalism but the stakes are equally high.

Therefore, although we do not yet have direct evidence from evaluation scenarios such as interviews and board meetings, this framework has a wider scope of application. It works best when the following conditions are met: respondents will anticipate and prepare for the questions from the evaluating party; the questioner has sufficient information to design questions that are difficult to anticipate; and both parties can improvise follow-up questions during the interaction.

Based on the above conclusions, we offer seven suggestions to help questioners use curveball questions in interviews:

1. Start from the other party's narrative logic

The essence of curveball questions is based on the preceding conversation: relative to the previous statements, this question is unexpected. In other words, the same question may be unpredictable in one conversation, but very easy to anticipate in another. Therefore, the question should start from the respondent's own statements: What assertions has the other party put forward? What are the underlying assumptions that support these assertions?

For example, in a strategic review meeting, the management is most likely prepared to answer questions related to demand forecasting, but may not be ready to explain how this forecast relies on one specific distribution partner. The breakthrough lies in digging into the untested logical links in this narrative, rather than simply asking questions in a more aggressive tone.

2. Follow up on logical connections instead of just asking for facts

Many follow-up questions only ask for more detailed figures, time nodes, or examples to support the other party's views. Although such questions are necessary, they are very easy to anticipate in advance. A more valuable approach is to ask the respondent to connect the logic between different viewpoints.

For example: How will the new pricing model affect the customer retention goals mentioned earlier? Where will standardization construction conflict with local independent authority? What do the recruitment plans mean for the cost targets? Such questions are closely linked to the respondent's own statements, while requiring them to deduce a logical chain that has not been sorted out in advance.

3. Modify a key assumption

Another idea for curveball questions: Examine the core assumptions on which the argument is based, and test whether the original conclusion still holds if that assumption is no longer valid.

For example: What will you do if the market demand growth rate halves? If the regulatory approval is delayed, which parts of the plan are still feasible? What evidence would make you overturn the current decision? Of course, such counterfactual assumptions must be reasonable and of great importance. Assumptions that are too unrealistic, while unexpected, will be easily dismissed as irrelevant by the other party and do not count as curveball questions.

4. Combine professional knowledge with external perspectives

The research conclusions present a thought-provoking contradiction: analysts who can ask curveball questions fall into two categories: those who are deeply rooted in the industry with profound professional accumulation, and those who are new analysts who have just started covering the enterprise and bring a fresh perspective. Managers can integrate the advantages of both. In a job interview, you can arrange an interviewer who understands the position but did not participate in the design of the interview outline; during a project review, you can invite a familiar colleague from an adjacent team to examine the plan.

5. Create a sense of surprise prudently, and do not deliberately show aggression

Curveball questions do not need to be confrontational, and in many cases they can help deepen mutual understanding. In management meetings, such questions can dig out hidden dependencies and avoid risks in project implementation; in coaching conversations, they can help employees re-examine their assumptions and resolve misalignment in communication with colleagues. In many cases, you can directly state the purpose of the question.

Example: "This decision is crucial. I would like to ask, if the competitive landscape changes significantly from now to the implementation stage, will your plan still hold up?"

When phrased properly, the question has depth while leaving room for the other party to think, without causing excessive pressure. On the contrary, if the only purpose of the question is to embarrass the other party, it will be difficult to achieve cognitive gains.

6. Grasp the timing of questions

Timing is critical to forming a curveball question: It must closely follow what the other party has just said, stay closely related to the theme, so that the other party cannot simply evade or change the subject. In contrast, interviewees who are skilled in media communication and public response can easily skillfully avoid a brilliant but off-topic high-difficulty question.

7. Continuously update your questioning ideas

Once a question becomes commonplace, respondents will prepare response plans in advance. This does not mean that we need to fix a set of "top curveball question lists". In fact, if you rigidly follow the above methods, respondents will also anticipate that you are going to use curveball questions and prepare responses in advance. An excellent questioner must continuously iterate on their ideas to design questions that are difficult to anticipate.

The Era of Curveball Questions

AI amplifies the shortcomings of traditional "aggressive questioning", because AI greatly reduces the response preparation cost for respondents. Respondents can use generative AI to anticipate questions that are most likely to be asked and rehearse their responses repeatedly. Traditional aggressive questions will become easier and easier to prepare for in advance. When response preparation becomes simple, the value of questions generated based on specific scenarios will continue to rise.

The question that is best at digging out the truth is not necessarily the most aggressive or the most bizarre. It is closely related to the topic, forcing the respondent to respond directly, while being difficult to anticipate, so that the ready-made answers they have prepared are not sufficient to cope with it. By digging into assumptions, connecting logic, putting forward reasonable alternative scenarios, listening flexibly and continuously iterating questioning ideas, managers can obtain more information through curveball questions, while keeping the conversation from falling into confrontation.

Nandil Bhatia, Wei Cai, Sameer B. Srivastava | Article

Nandil Bhatia is a PhD candidate in strategic management at Columbia Business School. Wei Cai is an associate professor of business at Columbia Business School. Sameer B. Srivastava is an associate professor at the Haas School of Business, University of California, Berkeley, and the Harold Furst Chair in Management Philosophy and Values.

This article is from the WeChat Official Account "Harvard Business Review" (ID: hbrchinese), author: HBR-China, published with authorization from 36Kr.