internshipinterview.org

Free tool

Résumé claim checker

Enter one claim from your résumé. You get a version you can defend and the questions an interviewer will ask. Fixed rules, no AI.

Starts on the worked question’s invented case: an intern’s win-back email at Larkhaven, as an AI tool polished the bullet.

The change
Before50
After70

+20 · up 40%

A claim you can defend

One of 2 people on it: built the send list, wrote two subject lines and set up the holdout and tracking sheet. Lapsed customers who ordered again: 70, against 50 in a different group or period (+20, up 40%; within 30 days); not a controlled comparison.

What an interviewer will probe 3 of 5 flagged

  • Compared with what? Flagged. 50 came from a different group or period, so it can’t show what your work caused.
  • Which part was yours? Flagged. 2 people worked on it, but “drove” credits you with the whole result.
  • Is the base big enough? Flagged. 40% is 20 on a base of 50: a few either way moves it a lot.
  • Over what period? Clear. Time window given: within 30 days.
  • Does the number matter? Clear. An outcome measure, not just attention or activity.

Prepare for these two

  1. What would have happened without your work, and how do you know?
  2. Which part of this was yours, and what did others do?

How the claim checker works

  1. The change. Absolute change = after − before. Relative change = (after − before) ÷ before, cut toward zero: whole percent from 10% up, one decimal below, so it never rounds up. A rate (%) changes in points, not percent. A base of zero has no percentage.
  2. Compared with what? Flagged unless “before” is a group that didn’t get the change, or the same thing before you changed it. A different group or period can’t show what your work caused.
  3. Which part was yours? Flagged when more than one person worked on it and your part is blank, or your claim credits you with the whole result (“drove”, “increased”, “grew”, “led”, “saved” and similar).
  4. Is the base big enough? Flagged when a count starts below 100, or at zero: a few more makes a big percentage.
  5. Over what period? Flagged when the time window is blank.
  6. Does the number matter? Flagged when the metric counts attention or activity: views, impressions, likes, followers, clicks, opens, reach, engagement, downloads, visits, posts, emails sent, lines of code.
  7. The rewrite. Your part first, then the numbers. Without a group that didn’t get the change, it states the numbers and says the comparison isn’t controlled, instead of claiming you caused them.
  8. The two questions. The first two flagged probes, in the order above. With fewer flags: “How did you measure it?” and “What would you do differently next time?”

On the default case, 50 becomes 70: +20, up 40%. But 50 came from a different group last year, two people worked on it and the bullet says “drove”, and 20 on a base of 50 is small. In the worked question, a random holdout shows the email added about 10 orders out of 2,000.

Practice defending your own résumé aloud with an AI interviewer on Socratify.

Last reviewed Nov 9, 2026