How to anticipate analyst questions before an earnings call

Artificial Societies ran three preregistered simulations and compared each with the call transcript, as Dr Linda Wei and Edoardo Chidichimo report in Testing Simulations Against Reality (September 2026). NVIDIA was descriptive: three societies, five runs per question, and each analyst’s predictions graded against the real question as a close match, a partial overlap or no match. Marvell and Credo were controlled comparisons with hypotheses fixed in advance. Each set a detailed persona against a basic profile that held only the analyst’s name, title, employer and coverage, with the same model, instructions and briefing, so that the comparison isolated what the extra detail about the person contributed.

Artificial Societies’ samples were small, the authors note. Marvell had 21 analysts on the roster, of whom 9 could be evaluated, and 525 generated questions; Credo had 18 on the roster, 13 of whom asked questions on the call, and 840 generated questions. Each analyst had five candidate questions in each arm, or version, of the study. The post highlights three findings:

1. 1.

   Detailed personas came closer to the real questions. For Marvell and Credo, their questions resembled the real ones more closely in substance and phrasing than the basic profiles did. For Credo, they did not improve the prediction of broad topics; the gain was in what analysts asked about those topics and how they put it.

2. 2.

   Detailed personas produced questions distinctive to the analyst. With a Credo briefing that gave company facts and named no debates, and with names, firms and introductions removed, an evaluator picked the right analyst from a line-up of five more often with detailed personas than with basic profiles, and more often than the 20% expected by chance.

3. 3.

   The briefing strongly influenced what came back. For Marvell, almost every briefed response focused on gross margins, a controversy the briefing had named.

BNY’s practice note on the earnings conference call (opens in a new tab) (2025) describes an investor relations team building the Q&A book over five weeks:

• •

  five weeks before the call, confirm dates and deadlines;

• •

  at four weeks, determine the key themes;

• •

  at two weeks, draft and review the press release, the script and the Q&A;

• •

  at one week, finalise them and read them through;

• •

  the day before, hold a team rehearsal with a final read-through and a mock Q&A.

The note advises teams to “run a simulated Q&A exercise” that asks the difficult questions, and to re-review the key questions from competitors’ calls in the final days. Its inputs are the business environment, previous and current financials, past earnings calls and competitors’ calls.

BNY and Sharon Merrill Advisors both organise the book by topic. David Calusdian of Sharon Merrill recommends Q&A messaging built around major topic areas, in bullets rather than prose, and rehearsing the hard questions out loud (July 2026 (opens in a new tab)). What comes out is a list of likely topics, each with an agreed answer and an executive chosen to give it.

A topic-based Q&A book tells management what will come up, and it says less about who will ask and what answer each analyst wants. For a team preparing an executive, the broad agenda may already be clear; the harder work is anticipating which part of the story needs explaining, which assumptions might be challenged, and where two analysts may want very different answers, as the authors of the earnings calls post write.

The topic list can also be wrong. In Artificial Societies’ hyperscaler earnings-call case study, the company expected a margin reset to dominate its call. Simulated analysts largely looked past it and pressed on the durability of demand and the scale of the company’s financing commitments, and on the live call the questions ran the same way. A mock Q&A run by colleagues draws its questions from the same room that wrote the topic list.

Simulation adds an entry per analyst, written in that analyst’s manner, to a book organised by topic. Before Credo’s first-quarter fiscal 2027 call on 1 September 2026, Artificial Societies built a simulated audience of 18 analysts from public records only, including their previous questions, and asked each for the single question they most wanted management to answer. Thirteen of those analysts asked questions on the call.

One analyst’s simulated question came close to the question he later asked. Both opened with congratulations on the record quarter and asked the same executive, by name, to go a level deeper. Both went straight to the optical business, hung the question on the $600 million optical target management had set, and asked about the mix of components inside that number. The match was not exact: the simulated question asked about gross margins and the quarterly ramp, while the real analyst picked up a topic management had raised on the call itself and extended the question into fiscal 2028. The earnings calls post shows the two questions side by side.

We would run simulation for a Q&A book in four steps:

1. 1.

   Build a persona for every analyst likely to ask a question. In Artificial Societies’ hyperscaler case study, that was every analyst who had asked a question on the company’s last eight calls, 14 personas in all.

2. 2.

   Ask each persona for the question it most wants answered, and generate several candidates per analyst; the Marvell and Credo studies used five.

3. 3.

   Add the candidates to the Q&A book under each topic, tagged with the analyst, the angle and any number the question anchors on.

4. 4.

   Rehearse the executive against the analyst’s version of the question, not only the topic heading.

Selective-disclosure rules can keep unreleased numbers and draft remarks away from real investors, as they did for the hyperscaler’s team. In that engagement, the team tested unreleased figures and scripts on simulated audiences, and no draft, number or script left Artificial Societies’ private, secured environment. Our page on investor communications testing covers testing the remarks and the guidance as well as the questions, and testing corporate strategy announcements covers the wider audiences around a strategic change or an acquisition.

Brief simulated analysts with the company’s facts and leave out the debates you expect, because in Artificial Societies’ Marvell simulation a named controversy pulled almost every briefed response towards it. That briefing named gross margins. For Credo, Artificial Societies ran every briefed arm twice: once with a briefing that named the live debates (protocol A) and once with the same facts only (protocol B). Removing named debates cut the share of primary persona questions about gross margins from 97% to 20%, but did not produce an improvement in topic overlap that met the preregistered threshold, Artificial Societies reports (September 2026).

The facts-only briefing is also where the identity result appeared: with company facts and no suggested topics for debate, the evaluator matched real questions to the right analyst more often from detailed personas than from basic profiles. The authors add that this sensitivity to framing is not unlike human behaviour. From that design we would recommend four practices to a preparation team; they are our reading of the study, not results it reports:

• •

  Brief the personas with the results, the guidance and the public record, and keep your list of expected issues out of the main briefing.

• •

  Run a named-debates version beside it as a diagnostic, and treat the difference between the two as a measure of how much the briefing is doing.

• •

  Keep an unbriefed run, which the study included as a separate arm, as a check on what the personas raise unprompted.

• •

  If the simulated analysts agree with the topic you named, count that as your assumption echoed back, not as confirmation.

Check predictions against the transcript only if you wrote them down before the call, together with the rules for scoring them. The Center for Open Science defines preregistration as specifying a research plan in advance and submitting it to a registry (Center for Open Science (opens in a new tab)). Artificial Societies registered its predicted questions and evaluation plans on the Open Science Framework (OSF), a free platform whose registrations create “time stamped, read-only versions of a project” (OSF (opens in a new tab)). After a call, it is easy to pick an impressive match from a large set of simulated questions, which says little about how the method performed overall, the earnings calls post notes.

Artificial Societies’ study scored each analyst’s five candidates against their real questions on four measures, each answering a different question a preparation team would ask:

Source: Dr Linda Wei and Edoardo Chidichimo, Artificial Societies, Testing Simulations Against Reality, September 2026.

A team without that apparatus can still run the NVIDIA version: mark each predicted question as a close match, a partial overlap or no match once the transcript is out. BNY’s note already suggests evaluating the quality and depth of the questions posed after each call. Scoring them against a list written two weeks earlier turns that review into a test of the preparation.

Simulated analysts can miss what an analyst learns during the call itself. On the Credo call, the real analyst built his question on a topic management raised during the call, which no simulation run beforehand could hear. Artificial Societies’ study also measured only the questions asked in the room. It does not establish how accurately a simulation measures second- and third-order effects, or how information spreads afterwards, and its 9 and 13 evaluable analysts are what the authors call a small number (September 2026).

Simpler inputs remained competitive in places. In Artificial Societies’ study, analysts’ past questions remained a strong baseline for wording and identity. Adding analysts’ published research notes did not establish an additional benefit in the small group for which notes were available, and for Credo the detailed personas did not improve the prediction of broad topics. The question a pre-call simulation still cannot write is the one an analyst builds from what management says on the call.