You ramble past 90 seconds or stop short and lose them. Ten minutes with a rubric turns a vague intro into one you’ve actually timed and tested.
| Target length | 90 seconds, about 180 to 210 words |
|---|---|
| Default speaking rate | 140 words per minute (conversational range is about 120 to 150 wpm per NCVS data) |
| Typical first draft overage | 30 to 60 seconds over target |
| Revision rounds needed | Usually 2 to 3 rounds to converge |
1. Write the rough draft
Talk through your intro out loud or type it messy: who you are, one thing you’ve done, what you want next. Don’t trim anything yet. A fresh graduate might ramble through every college project instead of picking one.
2. Give AI the rubric
Paste your draft into any AI chat with this: ‘Score this interview intro against: word count and estimated time at 140 words per minute, whether it opens with a hook instead of my name is, one clear throughline from a past result to what I want next, one specific number or result, and a close that says why this role. Score each out of 5 and give the exact sentence to cut first.’ This turns ‘make it better’ into five concrete checks.
3. Cut what it flags
AI will point to filler like ‘I also’ or ‘I think it would be a good fit’ and say how many seconds over 90 you are. Cut only what it names, never your one achievement sentence. Most first drafts run 30 to 60 seconds over.
4. Re-score until it lands
Paste the trimmed version back in and ask it to score again. Aim for 180 to 210 words and five out of five on the hook and the close. This usually takes two or three rounds, not ten.
5. Time yourself out loud
Say the final version aloud with a stopwatch. Nerves and your natural pace change the math. If you run long, that’s your real words-per-minute, tell the AI to use that number next time instead of the default 140.
A worked example
A fresh graduate’s first draft ran three minutes with no clear ask.
Before
Hi, my name is Priya Sharma and I recently graduated from Delhi University with a degree in computer science. During my time there I did a lot of projects, I was part of the coding club, I also did an internship at a small startup where I learned Python and SQL, and I also worked on a group project for my final year where we built a small app for tracking expenses, and I really enjoyed that experience because it taught me a lot about teamwork, and now I am looking for a job as a data analyst because I think it would be a good fit for my skills, and I’m a hard worker and fast learner, thank you.
After
I’m Priya, a data analyst who turned a slow expense-tracking spreadsheet into an app three hundred students still use today. At Delhi University I taught myself Python and SQL to build it, then interned at a startup where I automated a report that used to take a day down to twenty minutes. I’m looking for a data analyst role where I can do that at a bigger scale, and this role’s focus on customer analytics is exactly the kind of problem I want to solve next.
Three ways to bend it
Career switcher
Use the same rubric but add a sixth check: does the intro name the skill that carries over, not the job title that doesn’t. A switcher from teaching to sales might cut ‘I was a teacher for six years’ down to ‘I’ve spent six years convincing skeptical rooms to say yes.’
Laid-off tech worker
Add a check for how the gap is handled: one sentence, no apology, no over-explaining. Ask AI to flag any sentence that spends more than one line justifying the time off.
Returning parent
Add a check that frames the break as a choice, not something to apologise for. Tell AI to flag any sentence starting with ‘sorry’ or ‘I know it’s been a while’ and suggest a direct replacement.
When it goes wrong
AI says 90 seconds but saying it out loud takes way longer
You speak slower than the model’s default assumption of 140 words a minute. Time yourself once, then tell the AI your real pace and ask it to rescore against that number.
It cuts so much the intro feels bare and robotic
Tell it explicitly to protect your one achievement sentence and only cut connective filler like ‘and then’ or ‘so basically’. Rubrics are meant to flag, not auto-edit.
When not to use this
This only works once you already know the one story you want to tell. If you’re still deciding between three different achievements to lead with, figure that out first on paper. Asking AI to time and score a draft that doesn’t know its own point yet just gives you a shorter, faster version of the same confusion.
What you can now do
You walk in with a pitch you’ve actually timed, not one you’re hoping lands in ninety seconds.
Part of Making Life Easier With AI, a weekly walk-through of one small thing AI can take off your plate.
Take this further
Build the outline for my interview presentation: a 90-day plan for this role. Structure: slide 1, what I understood about the team's goal, in one line. Slides 2 to 4, the first 30, 60 and 90 days, each with two priorities and what done looks like. Slide 5, the risks I would watch and how I would report progress. Slide 6, the three questions I would ask the team. One idea per slide, seven slides maximum, no filler text. Keep every claim something I can defend out loud in the room. The role: [paste job description]. My context: [paste five bullet points about you].
Takeaways
- Write the messy first draft out loud or typed, covering who you are, one accomplishment, and what you want next, without trimming anything yet.
- Paste the draft into an AI chat with the five point rubric (word count and time, hook opening, one throughline, one number, and a why this role close) and ask for a score out of 5 on each.
- Cut only the exact sentences the AI flags as filler, then resubmit and re-score two to three times until you land at 180 to 210 words with top scores on the hook and the close.
- Time the final version aloud with a stopwatch, and if you run long, tell the AI your real words per minute instead of the 140 default before the next rescoring.
- If you are a career switcher, laid off, or a returning parent, add the matching extra rubric check before you start cutting rather than after.
Questions people actually ask
How long should my interview introduction actually be?
Aim for about 90 seconds, which lands around 180 to 210 words if you speak at a typical conversational pace of roughly 140 words per minute. That length is enough to cover one clear achievement and what you want next, but short enough that an interviewer stays with you instead of waiting for you to wrap up.
How many words is a 90 second intro?
At a common conversational pace of about 140 words per minute, 90 seconds works out to roughly 180 to 210 words. Your own pace may run faster or slower than that default, which is why timing yourself out loud once and adjusting the number is more reliable than trusting the estimate alone.
How do I use AI to practice my elevator pitch?
Paste a rough, untrimmed draft into any AI chat and ask it to score the draft against a short rubric covering word count and time, whether it opens with a hook, one clear throughline, a specific number, and a close tied to the role. Cut only what it flags, then resubmit the trimmed version two or three times until it scores well across the board.
Why does my elevator pitch always run way longer than planned?
Most first drafts run 30 to 60 seconds over because people try to list every accomplishment instead of choosing one. Nerves can also slow your real speaking pace below the 140 words per minute default, so timing yourself aloud once and feeding your actual pace back into the rubric closes that gap.
How should I pitch myself after a career change or a layoff?
Use the same rubric but add a check that matches your situation: name the transferable skill instead of the old job title if you switched fields, cover an employment gap in one sentence without over explaining, or frame a career break as a choice rather than something to apologize for before you start trimming.
Sources
- How to Craft an Effective Elevator PitchBloomTech
- Average Speaking Rate and Words per MinuteVirtualSpeech
- Identifying the Skills That Can Help You Change CareersHarvard Business Review
What happens next
Expect interview prep tools to start building a timed scoring rubric like this directly into their product instead of leaving candidates to assemble one by hand in a chat window. Watch whether those tools begin asking for a person’s real speaking pace up front rather than defaulting to a generic words per minute figure, since that is the step most people currently skip.
