Your interview answer was perfect until the follow-up

Interviewers now dig into polished answers with follow-up questions. What they ask, why rehearsed answers crack, and how to prepare the layer underneath.

You have told the story of the delayed launch a dozen times. It has a clear problem, a decision you are proud of and a result with a number in it, and in the interview you tell it without a stumble. The interviewer nods and asks which part of the work was yours. A moment later she wants to know what else the team considered and who argued against the plan. Your rehearsed answer has nothing more to give, and the silence starts to stretch.

That second question is becoming the center of many interviews. A polished first answer is easier to produce than ever, interviewers know it, and more of their attention now goes to what comes after it.

Why interviewers rely on the second question now

In August 2025 Axios, citing the Wall Street Journal, reported that Google, Cisco and McKinsey had brought face-to-face interviews back into part of their hiring. Google’s chief executive Sundar Pichai said the company would add at least one in-person round to make sure the fundamentals are there. A recruiting firm in Dallas estimated that in-person requests from its clients rose from 5% to 30% in a year (Axios).

Surveys show what employers are worried about. In a Gartner survey of 3,290 job candidates at the end of 2024, 39% said they had used AI in their application, and 29% of those had used it to write answers to assessment questions. In another Gartner survey of 3,000 candidates in 2025, 6% admitted to interview fraud, posing as someone else or having someone pose as them. Gartner advises recruiters to learn to recognize evasive behavior (Gartner).

The interview practice platform interviewing.io measured how far a borrowed answer gets. Engineers with at least four years of experience were told to rely on ChatGPT in 32 mock interviews, and none of the interviewers mentioned any suspicion. On coding questions copied straight from LeetCode they passed 73% of the time. On questions the interviewers had written themselves the pass rate fell to 25%, below the 53% that candidates on the platform usually reach with no help at all. One interviewer described the habit that exposes the difference, adding one or two unique twists to a familiar problem (interviewing.io). The study was small and ran on audio only. Its lesson reaches well beyond coding, because interviewers in many fields are moving toward questions that cannot be prepared word for word.

What follow-up questions ask

Structured interviews plan their follow-up questions in advance. The training material of the U.S. Office of Personnel Management calls them probes and tells interviewers to narrow in on the candidate’s own role and actions. Whenever an answer turns vague or slides into what the group did, the interviewer is told to ask for clarification (OPM).

For a story about the past, OPM’s example probes cover three things. About the situation, interviewers ask who was involved and what led up to it. About your action, they ask how you responded and what your role was. About the result, they ask what the outcome was and what you would do differently. A hypothetical question gets its own probes, about what else you could do, what would change your course and how your action would be received.

Amazon tells its candidates the same thing in plainer words. It does not expect perfect answers or rehearsed scripts, and it counts drawing out the details as the interviewer’s job. It asks for one situation told in detail, with I in the places where most people say we (Amazon).

Why a rehearsed answer cracks

A rehearsed answer is written at one level of detail. It carries the headline of the story, the problem, the action and the result, because that is what fits into two minutes. A follow-up question almost always goes one level down, to the number behind the result, the option you turned down, the colleague who pushed back or the week when things went wrong.

When you lived the story, that level is in your memory and surfaces after a moment’s thought. When the answer came from a template, a coach or a chatbot, the level below was never there, and fluent delivery on top hides the gap for only a minute or two. The interviewing.io study recorded exactly this. One candidate gave a correct solution and could not explain the time complexity that came with it.

The most impressive answer carries the most risk. A big claim invites a closer look, so the line you are proudest of needs the most preparation underneath it.

Prepare the layer underneath

Go through each story you plan to tell and write a few words under every part of it, the ones a follow-up question would ask for. In the story of the delayed launch, that means the two people who disagreed about the date and why the customer cared, the call you made yourself and the option you turned down, and how the drop in support tickets was measured and what it was before. Add one honest sentence about what you would do differently, because that is one of the standard questions about results.

Write with the original record open. The project document, the commit history, the email that settled the decision and the dashboard that showed the result bring back details you have forgotten, and they keep you from rounding a number up without noticing.

Practice out loud with a friend who interrupts. Ask them to answer each of your sentences with why or how, three times in a row. Most stories run out at the third why, and your kitchen table is a far better place to discover that than the interview room.

When you cannot answer a follow-up question

It will still happen. Suppose the interviewer asks for the exact drop in tickets and you only remember that it was about a third. Say that it was roughly a third and offer to explain how the team measured it. An honest gap costs much less than a precise figure the interviewer can check later.

When a question is unclear, ask what the interviewer means, which Amazon openly invites. When the follow-up asks about an experience you have not had, describe the nearest one you have and say how it differs.

Where Prepped fits

Prepped prepares from the material you give it, such as your resume, the job description and your project folders, and writes answers to the questions the interview is likely to bring. A second reading checks every sentence about your own experience against that material, and whatever the material cannot support stays in [square brackets] for you to decide on. Those brackets make a ready list for the exercise above, since each one marks a place where a follow-up question would find nothing underneath.

Before your next interview

Pick the three stories you are most likely to tell. For each one, write down what lies under every part of it, check one number against the original record, and tell it to someone who keeps asking why. A good first answer earns you the follow-up question, and the follow-up is where you show what is really yours.

Common questions

Why do interviewers ask so many follow-up questions?

Follow-up questions show what a rehearsed answer hides, such as your own role, your reasoning and how a result was measured. Structured interviews plan them in advance, and now that AI makes a polished first answer easy, interviewers rely on them more.

How do I answer a follow-up question I did not expect?

Take a moment, say what you know and name the part you do not know. Offer the closest detail you have, and ask what the interviewer means when the question is unclear. An honest gap costs less than a guess they can check.

Can interviewers tell if an answer was written by AI?

Rarely from the first answer. In a small experiment by interviewing.io, no interviewer suspected the candidates who used ChatGPT. When the interviewers asked questions they had written themselves, those candidates passed only 25% of the time, because a borrowed answer has nothing underneath it.

Sources

  1. Companies embrace in-person interviews to dodge the chatbots, Axios. Read November 5, 2026
  2. Gartner Survey Shows Just 26% of Job Applicants Trust AI Will Fairly Evaluate Them, Gartner. Read November 5, 2026
  3. How hard is it to cheat in technical interviews with ChatGPT? We ran an experiment., interviewing.io. Read November 5, 2026
  4. Structured Interviews (training presentation), U.S. Office of Personnel Management. Read November 5, 2026
  5. Interview loop, Amazon. Read November 5, 2026