Rambling in interviews: shorten a drifted answer

What counts as rambling in an interview answer?

Rambling is an answer where the words go to the wrong parts. Most of it is background, the interviewer waits for what you did, and the result arrives late with no number. Length is part of it, though a two-minute answer can ramble too. The test is how much of the answer an interviewer could actually score.

That's why “I talked too much” is a weak note to take home. It tells you the answer felt long without telling you which part to cut. A useful note names the part: “I spent the first minute on the company and the last 20 seconds on what I did.”

How long should an interview answer be?

For a behavioral question, aim for about two minutes and stop at three. Amazon's own interview guide tells candidates to practice so answers are “concise yet detailed, typically lasting 2-3 minutes each.” In words, that's roughly 300 to 500, depending on how fast you talk.

The word range comes from conversation research. A 2006 study of English telephone conversations by Yuan, Liberman and Cieri found average speaking rates of about 152 to 170 words per minute, depending on the topic. Interview answers vary, so treat the range as a rough guide.

That gives you a way to check a drifted answer tonight. Write it out as close to word for word as you can, count the words, and divide by about 160. If it comes to more than three minutes, it ran long. If it came in under three and still felt like rambling, the problem is where the words went.

Why does a drifted answer score badly even when it's short?

Interviewers score the parts of an answer that match the rubric, and background rarely matches. Highly structured interviews are “scored according to benchmarks of proficiency,” in the words of the US Office of Personnel Management's guidance. Benchmarks describe what a candidate did and what came of it, so setup earns little on its own.

Google's guide to structured interviewing says interviewers should take detailed notes so answers can be reviewed later, and score them against a standard rubric. If two of your three minutes were context, the notes hold mostly context. The person who reads them later sees a short action and a result with nothing to weigh.

The same Amazon guide asks for “I” statements and “precise metrics and data that quantify your impact.” Those are the two things a rambling answer tends to lose first. The setup feels necessary while you're speaking, so it gets the time, and the action and the figure get squeezed into the last 20 seconds.

What does a drifted answer look like next to a short one?

Below is the same story told twice, in answer to “Tell me about a time you dealt with a difficult stakeholder.” Both answers are labelled examples written for this page, and the company, people and figures are made up. The first runs 311 words, about two minutes when spoken. The second runs 143 words, under one minute.

Before: 311 words (example)

So, yeah, good question. I think the one that comes to mind is from my last job at a mid-sized logistics company. I was on the operations team. Well, technically I was on operations, but our analyst left, so I picked up a lot of the reporting work as well. Anyway, there was a sales director who was known for being hard to work with. Everyone had a story about him. He'd send emails late at night and copy in half the company. So we had this project to change how we quoted delivery times, because the old process ran on a spreadsheet that someone built years ago and nobody fully understood. It had macros, and some of them didn't work anymore. And the sales team relied on it, so any change was a big deal. I was asked to lead the rollout, which was a bit of a stretch for me at the time, but I was happy to do it. The sales director didn't like the new quotes because they were longer than what his team was used to giving. He said we'd lose deals. We had a few meetings about it, and some of them got pretty tense. At one point he went to my manager's manager. I think part of it was that he hadn't been included early on, which in hindsight was fair. Different teams have different priorities, and I learned a lot from that. So in the end we worked it out. We did a trial with a couple of his reps, and the numbers looked good, and he came around. Actually, I think the quotes were more accurate after that, and there were fewer complaints from customers about late deliveries. So yeah, it went well, and we're on good terms now. I think that's the main thing I took from it, involving people early.

After: 143 words (example)

At my last job I led the rollout of a new way to quote delivery times. The sales director pushed back hard. He said longer quotes would lose deals, and he took it to my manager's manager. I asked him for one meeting and started by saying he should have been involved earlier, which was true. Then I offered a two-week trial with two of his reps, so he could judge it on his own team's numbers. I tracked every quote they sent against when the order actually arrived. Over the trial, on-time quotes went from about 70 percent to 90 percent, and his reps couldn't point to a lost deal. He signed off, and the rest of sales moved over the next month. What I'd do differently is bring the person most affected in at the start, before the design is fixed.

The first answer is inside Amazon's two-to-three-minute range, and it still rambles. The table counts the words in each part of both answers. The action, the part an interviewer most needs, went from 19 words (6 percent of the answer) to 52 words (36 percent).

Two things carry most of the change. The background that set the scene (the side role, the macros, the late-night emails) was true and none of it was scorable. And “we did a trial” hid three decisions the speaker made alone: asking for the meeting, owning the late invite and choosing to test on the director's own team.

How do you shorten an answer that drifted?

Write the answer out, label every sentence by what it does, then rebuild it in order: situation in two sentences, your task in one, the action in three, the result with a figure, and one line of lesson. Cut everything that has no label. Give each answer about 20 minutes.

1. Write the answer as you gave it, as close to word for word as you can, and count the words.
2. Label each sentence: situation, task, action, result, lesson, or none.
3. Delete every sentence labelled none, then cut the situation to two sentences.
4. Rewrite every “we” in the action as what you did, in at least three sentences.
5. Replace any vague result with one figure you can defend, or say plainly that you don't have one.
6. Count again and read it aloud with a timer. Stop when it runs under two minutes.

Step 5 matters most. If you don't know the real figure, find it or leave the number out. A number you can't explain if asked is worse than none. The STAR method guide covers the four parts in more depth, with examples of where each one breaks under a follow-up.

What is the “say instead” pattern?

The “say instead” pattern is a pair: the line you actually said, quoted, and the line you'll say next time, written out in full. It works because it fixes one specific moment, which you can rehearse, instead of a general goal like “be more concise,” which you can't. Write one pair for each drifted section.

Here are pairs from the example above. Each quoted line is the before answer, and each line after the arrow is what to say instead.

• “Well, technically I was on operations, but our analyst left…” becomes nothing. Cut it.
• “So we had this project to change how we quoted delivery times, because the old process…” becomes “At my last job I led the rollout of a new way to quote delivery times.”
• “We had a few meetings about it, and some of them got pretty tense.” becomes “I asked him for one meeting and started by saying he should have been involved earlier.”
• “The numbers looked good, and he came around.” becomes “On-time quotes went from about 70 percent to 90 percent.”
• “I think that's the main thing I took from it, involving people early.” becomes “What I'd do differently is bring the person most affected in at the start.”

Put the pairs in your notes from the round. The 10-minute debrief template has a section for the answer you'd redo, and the guide to the post interview notes template explains how to fill it while the wording is still fresh.

How do you find which answer drifted if you can't remember?

Look for the moment the interviewer reacted. A cut-in, a long pause before “okay,” or a follow-up asking what you did yourself often follows an answer that drifted. Mark those moments in your notes, then rebuild the answer that came just before each one. Start with the one or two answers you'd most like to redo.

The guide on how to tell you lost the room lists the reactions that most often mark that point. If the interviewer said nothing at all after a long answer, the guide to what it means when the interviewer went quiet covers the different kinds of silence.

How do you check the new answer before the next round?

Say it aloud three times with a timer, then check every fact against your resume and what you said in earlier rounds. A shorter answer often drops a detail you gave before, and a dropped detail is fine. A changed figure is a different matter, because later interviewers may have read the earlier notes.

The check on whether your story changed between rounds shows how to catch a figure that moved. If you have another round with the same company, the guide on preparing for round two covers what to rehearse first.

If you recorded the round with everyone's consent, Prept transcribes it and reads each answer for structure, evidence and numbers, then quotes the moment you lost the room and gives you words to say instead. It's analysis after the call, never during it, and your first interview hour is debriefed free with no card.