Interview Question and Follow-Up Tactics for Testing Resume Claims in Key-Hire Interviews
I’m working with founders to turn job descriptions into interview questions, and then to turn the answers they get into a hiring decision. We’ve been focusing on how to write questions that will help you test the claims of a candidate who has polished their resume.
To start, we’re looking at the role and figuring out what problems the new hire must solve. Then, we’re considering what the candidate’s resume says about them, identifying which claims are most important to the role and which seem overpackaged.
Then we look at each bullet point on the resume and consider how to test it. One example is a candidate who wrote: “Led a growth process rebuild.” A good way to follow up would be: “What was the baseline? What were your decisions? What changed after you did this work?”
The idea behind these sorts of questions is to invite the candidate to tell us a story. That’s where we can hear if they really did what they said they did or not. But once they start telling a story, we have to ask clarifying questions only when details are missing.
Mentoring research notes that open-ended rather than leading questions were used to reduce bias, and that extra probing was added only where clarification was needed.
Because interviews are so short, it helps to break them into parts. For example, the Urban Institute researchers conducting telephone interviews about community policing split their interviews into two parts. In part I, they asked about day-to-day operations; in part II, they asked about strategy, partnerships, and other aspects of the larger picture. The researchers completed 59 interviews and five follow-up interviews, usually lasting about an hour. They took detailed written notes during interviews and sent thank-you emails afterward to make sure any unanswered questions were resolved.
Asking candidates to explain what they did and how they got results is a great way to find out whether they actually did what they claim. But hiring managers need to remember that they are listening for what is verifiable—not just what sounds impressive. RoleProbe’s system for scoring interviews gives a list of things to listen for: business context, personal ownership, team support, timeline, tradeoffs, and material that can be reviewed after the call. If the answer doesn’t cover all those things, it might be a risky one.
After the interview, the hiring manager should do a debriefing. This isn’t just a memory pass—it’s an evidence pass. If there are unresolved points, they should be noted, and the hiring manager should think through how to check them. At the Urban Institute, they took detailed notes and sent thank-you emails after interviews to resolve unanswered questions.
Impressive outcomes should be tested against definitions, inputs, timing, and matching evidence. When MCE evaluated its own evaluation, it pointed out that claimed parameter values likely were incorrect and that some savings could not be credibly claimed until a year of post-installation usage data existed.
In contexts where one miss is expensive, hiring teams should go beyond the conversation itself to review samples of the candidate’s work, ask for artifacts, or dig into edge cases that might matter to the role. As DBHIDS found, 85% of supervisors reported that agencies hired CPSs with criminal backgrounds, but many still wanted clearer standards for which histories mattered. Hiring teams should focus on job-relevant risk factors rather than blanket assumptions.
Finally, hiring managers should track which claims they tested, what evidence they heard, and which hires proved strong or weak. DBHIDS found that 61% of upper management didn’t track CPS careers or attrition rates, suggesting that without tracking, hiring managers cannot improve over time.
So here’s my plan for every polished resume: define the problem, flag the suspect claims, ask open-ended questions to get stories, then ask specific questions to fill in gaps, document what you hear, and finally, verify the points you don’t yet know.