The 7 Words That Predict Bad Hires
We analyzed 20,572 real interview answers and had hiring managers grade every one. The answers that predicted a poor hire were built out of measurably different words than the strong ones — and once you know the seven to listen for, you can't un-hear them.
Why the Words Matter
Candidates can rehearse stories about their accomplishments. What is harder to control are the pronouns, verb tenses, and qualifiers they use while telling those stories. Those small language choices can reveal meaningful differences between high and low performers.
No single word predicts a bad hire. But when several patterns appear together, they give interviewers a more objective way to evaluate an answer instead of relying on gut instinct. This guide shows you what to listen for.
The seven signals, at a glance
To find these signals, we graded all 20,572 answers as strong, weak, or in-between, then ran a linguistic and contextual analysis comparing the two groups word by word — the pronouns they chose, the tenses they used, the qualifiers they leaned on. (The full method is at the end of this report.) What surfaced was a consistent, measurable pattern, and here is the whole of it in a single view. Bars to the right mark language that showed up more in low-performer answers — the words that should make you lean in and listen harder. Bars to the left mark language that showed up more in high-performer answers.
Read top to bottom, the pattern is coherent: the weaker answers keep the candidate at a distance from their own experience. They talk about what "you" or "they" should do, in the present or future, wrapped in absolutes and negatives. The stronger answers, as we'll see, do close to the opposite.
Listen for "I" and "we": the language of ownership
High-performer answers contained 21% more "I" language — I, me, my — than low-performer answers. When a candidate refers to themselves, they are taking ownership of the situation they're describing, and that ownership does double duty: it signals accountability, and it quietly signals truthfulness. "I saw the problem and I called the customer" tells you this person was close to the events, actually took the action, and is willing to stand behind the result.
High-performer answers also contained 65% more "we" language — we, us, our. Like "I" language, "we" points to ownership, but it adds something you often want alongside it: a willingness to share credit and to work well with other people. A candidate who naturally says "we" while still making their own role clear is showing you both accountability and collaboration in the same breath.
When you hear a candidate narrate their own experience in the first person, with specifics attached, treat it as a point in their favor — and as a sign the story is probably real.
Watch for "you" and "they": the language of distance
This is the single largest gap in the entire study. Low-performer answers contained 392% more "you" language — you, your, you'll — than high-performer answers. Second-person language is a way of not taking ownership. It's evidence of what we'd call psychological disassociation: the candidate steps out of their own experience and starts describing the world in the abstract.
The contrast is easiest to hear side by side. Notice that only one of these tells you the person actually did anything.
The second answer never claims the candidate called anyone. It only asserts what the best practice would be. That's not a report of experience; it's a recitation of a rule, and it should make you wonder whether the experience exists at all.
Low-performer answers also contained 90% more "they" language — they, them, themselves. Third-person pronouns work much like second-person ones, putting distance between the candidate and the events. "Before someone calls the customer, they should check the call log" has the same problem as the "you" example: it describes a generic procedure rather than a lived moment. When an answer to a "tell me about a time" question is populated mostly by "you" and "they," you're likely hearing a candidate who doesn't have the specific experience you asked for.
The tense tells you whether it happened
A behavioral question asks about the past for a reason: you want a real event, not a philosophy. The verb tense a candidate reaches for is one of the cleanest signals of whether they're giving you one.
High-performer answers contained 38% more past-tense verbs. People who have genuinely done the thing tend to tell a simple, focused story and lay out the facts as they happened — "I saw there was a problem and called the customer." The past tense falls out naturally because they're recounting something real.
Low-performer answers contained 104% more present-tense verbs. A candidate who lacks the experience often avoids the past tense, because there's no specific past to describe. Instead they drift into a general statement of how things work — "When there is a problem, it is best to call the customer." And low-performer answers contained 71% more future-tense verbs, which is the same evasion pointed forward: "I will call the customer when there's a problem." Both are ways of answering a question about the past without ever visiting it.
The tense chart below shows the split clearly: high performers concentrate in the past, low performers spread into the present and future.
Which emotions they name — and how often
High-performer answers contained 28% more positive-emotion words — happy, thrilled, excited. A strong candidate is comfortable saying something like "I was thrilled to help the customer." But the more telling difference isn't how much high performers reach for positive words; it's how rarely they reach for negative ones.
Low-performer answers contained 92% more negative-emotion words — angry, aggravated, afraid, pessimistic, unhappy. When a candidate volunteers negative emotion in an interview, it raises a fair question: why couldn't they find a more constructive resolution to whatever they're describing? An answer like "I'm aggravated about that part of my job, and honestly I'm pretty pessimistic about whether it'll ever change" isn't describing a proactive, coachable, go-getter — and those are precisely the qualities most interviewers are trying to find.
"Can't," "never," "always": negation and black-and-white thinking
Low-performer answers contained 123% more negation — no, can't, couldn't, didn't. It has long been a cliché that hiring managers don't want to hear the word "can't," and the data suggests the cliché is earned. Tension, low emotional intelligence, negativity, and pessimism all tend to surface as negation: "No, that didn't happen at my last job, because you couldn't make those kinds of decisions there."
Low-performer answers also contained 103% more absolutes — always, absolutely, unquestionably. Absolutes such as "I always call the customer" or "I am unquestionably the best person on the team" can come from insecurity, a need to show off, or black-and-white thinking that leaves little room for intellectual flexibility. It's worth asking yourself: when was the last time you encountered a real workplace situation that was always one way? The candidate who lives in absolutes may be telling you how rigidly they see the world.
None of these markers is a disqualifier by itself. A strong candidate will say "always" now and then, and a weak one will occasionally say "I." The signal is in the pattern and the pile-up — an answer built largely out of second person, present tense, absolutes, and negation, with no first-person past-tense specifics anywhere in it. Treat a single marker as a cue to probe deeper, not as a verdict to act on.
Adverbs and the length of the answer
Low-performer answers contained 40% more adverbs — very, really, quickly. Insecurity, thin experience, or a wish to look better than the facts warrant can all trigger a need to embellish, and adverbs are how people "amp up" a claim they can't support with detail. Instead of describing the specific occasion when they had a genuinely good idea, a weaker candidate will say something like "I was quickly coming up with great ideas." The adverb is doing the work the missing example should be doing.
Finally, high-performer answers were 23% longer than low-performer answers. This one is almost mechanical: because high performers have more and better experience with the attitudes and skills you're probing for, they simply have more to say. They give longer, more detailed answers because they can. A consistently thin answer to a question that invites a real story is itself a small piece of evidence.
Putting it to work in your next interview
The practical value here isn't a checklist you run against a transcript after the fact. It's a change in what you're listening for while the candidate is still talking. Once you know that ownership lives in first-person past-tense specifics, and that distance lives in second- and third-person generalities, you start hearing the difference in real time.
Two answers can describe the very same accomplishment and land completely differently. Consider a candidate asked about working through a difficult customer issue.
The first answer is first person, past tense, specific, and it names a real outcome. The second is second person, present tense, absolute, and negated — and it never once tells you the candidate did any of it. Same topic; opposite evidence.
So as you head into your next round of interviews, listen past the story. When a candidate leans on "I" and "we," anchors their answer in the past tense, keeps their emotions constructive, and gives you enough specific detail to picture the scene, you're likely sitting across from a high performer. When the answer drifts into "you" and "they," into the present and future tense, into absolutes, negation, and adverbs standing in for detail, you may have a low performer — or, at minimum, someone who doesn't have the experience they're implying, and whom it's worth slowing down and probing much more carefully.
How we ran the study
We asked 1,427 professionals to write answers to 15 open-ended interview questions as though they were applying for a job. Every answer was validated to confirm it was of sufficient length and contained appropriate content, which left us with 20,572 usable answers.
A panel of hiring managers and HR executives then graded those answers, sorting each one by whether the applicant would likely be a great hire (a high performer), a poor hire (a low performer), or somewhere in between. Overall, 34% of the answers were graded as low-performer answers and 29% as high-performer answers. We then subjected those two groups to a textual analysis, using linguistic and contextual methods to identify exactly where the language of a strong answer and a weak answer diverge — comparing pronoun use, verb tense, emotional valence, negation, absolutes, adverb frequency, and answer length across the two pools.
The percentages reported throughout this document express the difference in how frequently each feature appeared in low-performer versus high-performer answers. They describe patterns across more than twenty thousand responses; any single interview answer can run against the grain, which is why the signals are best used to guide where you probe, not to settle a decision on their own.
