UmAh guides

Filler words in job interviews

A few fillers in an interview are normal and interviewers mostly tune them out. What hurts is a pattern, especially hedges like "I think" and "kind of" wrapped around answers that should be confident. The fix is to prepare the first sentence of your common answers, pause instead of filling, and record one practice run to see where the fillers land.

What interviewers actually notice

There is no good study showing that saying um costs people job offers, and you should distrust any statistic that claims one. What the research does show is narrower and more useful. Listeners who are focused on content barely register ums; listeners focused on style notice them a lot¹. Interviewers are mostly listening for content, which works in your favor. The same experiment found that fillers still shaped the overall impression of the speaker even when listeners could not report hearing them, so a heavy pattern does leak through¹. Separately, filled pauses before an answer signal to listeners that the speaker is less sure of it². In an interview, that is the real cost: not the sound, but the doubt it attaches to your answer.

The hedges matter more than the ums

Um and uh are planning noises. "I think," "probably," "kind of," "sort of," and "maybe" are different: they soften a claim. Softening is fine when you are uncertain and bad when you are not. Compare "I kind of led the migration, I think it went pretty well" with "I led the migration. It shipped on time." Same facts, very different speaker. Go through your prepared answers and strike every hedge that is not doing honest work. Keep the ones that are: "I'd estimate three months, but I'd want to check the data model first" is a hedge that shows judgment.

Where fillers land in interviews

Fillers cluster in three places. The first second after a question, while you decide which story to tell. The pivot in the middle of a story, when you move from what happened to what you did. And the ending, when you are not sure whether you have said enough. Disfluencies in general rise when the speaker is carrying the planning load in a conversation³, and in an interview that load is all on you. Knowing the three hot spots lets you prepare for them specifically rather than trying to be fluent everywhere.

Prepare openings, not scripts

Memorized answers sound memorized and fall apart under a follow-up. Instead, prepare the first sentence of each answer you expect to give: your background, why this role, the hardest problem you solved, a time you disagreed with someone. The first sentence is where the biggest filler cluster lives, and a ready sentence removes it. After that, speak from memory of the actual event. For endings, pick one closing move and use it every time: a one-line result. "That cut onboarding time in half." Then stop. The silence after a clean result sounds like confidence.

Buy time out loud

When a question needs thought, say so. "Let me think about that for a second" is not a filler; it is a sentence, and it buys you two or three seconds of legitimate silence. Interviewers read a short, deliberate pause as composure. It is the stretched um that reads as flailing. Microsoft's coaching guidance for presenters makes the same point: pause if you need a moment, and write down transitional phrases for the moments you tend to fill⁴.

One practice run, recorded

Do one full mock interview out loud and record it. Then count: fillers per minute, and where they landed. Most people are surprised twice, once by how many there are and once by how predictable the locations are. If you want a count without doing the tally yourself, UmAh gives a post-meeting report with fillers per minute and where each one landed. Practice the three hot spots, record again, compare. The guide on practicing a presentation alone has a drill structure that transfers directly.

Pace and talk share

Nervous candidates speed up, and speed produces fillers because the mouth outruns the plan. A pace somewhere in the range most coaching tools recommend, roughly 100 to 165 words per minute⁴, leaves room to think. The other interview trap is talking too long. A good answer to a behavioral question runs one to two minutes; if you pass three, you are usually filling because you have run out of story. More on both in speaking pace and how much to talk.

Keep perspective

Spontaneous speech is disfluent by nature, and listeners are built to handle it⁵. An interviewer who rejects a strong candidate over a handful of ums is making a mistake, and most do not. Your goal is to remove the pattern, so that what they remember is the answer.