Behavioural Interviews Module 2 – Structuring answers that score

Results, numbers and what you learned

Turn a vague ending into a result with evidence, give honest numbers or comparisons, separate your part from the team's, and end with what you learned.

Beginner
15 minutes
Examples run with Node.js 24.21.0 and quickjs 0.32.0
By MySmartCoPilot

Free: A result is a change, with its evidence · Your part, the team's part and luck · End with what you learned, and show that it changed you. With a pass: 2 in-depth parts and 3 interview answers, about 9 minutes.

What you will learn

Express results honestly as numbers, comparisons or observable change
Separate your contribution from the team's and from luck
Protect confidential figures with ranges and relative change
Write a closing reflection that shows changed behaviour

Before you start

The STAR method and its variants, used well

What you’ll learn · Before you start

What you will learn

Express results honestly as numbers, comparisons or observable change
Separate your contribution from the team's and from luck
Protect confidential figures with ranges and relative change
Write a closing reflection that shows changed behaviour

Before you start

The STAR method and its variants, used well

On this page

An ending such as “and it went really well” stops just where the interviewer needed the most. “The nightly job had failed three times a week; in the eight weeks after my change it did not fail once” is a result they can write down and weigh.

Below are what makes results evidence, what to say when you have no numbers, how to separate your part from the team’s and from luck, and how to end with what you learned so that it shows a change in you. A checker you can run flags the endings that are still vague.

A result is a change, with its evidence

Results count when they can be checked. Google’s interview tips suggest a formula, “Accomplished [X], as measured by [Y], by doing [Z]”, and Meta’s guide asks for tangible or quantifiable results. Both ask for what changed, how you know and what you did to cause it. Step through one vague ending as four questions turn it into evidence:

Interactive · stage by stage Four questions that turn an ending into evidence Step through one vague ending as it is rewritten. Each question adds one thing the interviewer can write down.

The drawing shows the whole scene; its 6 steps are listed below.

Every step as text

The vague ending. Meera, an invented frontend engineer, ends a story with: 'and the page got a lot faster.' It may be true, but all an interviewer can write down is that she says so.
What changed?. The product page, for shoppers on phones. Naming the thing that changed, and for whom, is the first piece of evidence.
By how much?. Its slowest loads, the 95th percentile, fell from about 3.2 seconds to 1.1. A before and an after turn "faster" into a size.
How do you know?. From the site's monitoring, over the two weeks before the change and the two weeks after it, the same weekdays in each. Now the number has a source.
Because of what you did?. She moved the recommendations carousel out of the first screen's load. One clause, and the result is hers rather than the team's or the weather's.
The rewritten ending. 'After I moved the recommendations carousel out of the first screen's load, the product page's slowest loads on phones fell from about 3.2 to 1.1 seconds, over two weeks of monitoring before and after.' Four answers, one sentence.

The evidence takes one of three forms: numbers with a before and an after, a comparison such as “half the time it used to take”, or a change someone could see, such as a team adopting your tool. For engineers the useful numbers are usually users affected, response times, error rates, cost, time saved, delivery dates and adoption; for students, real users of a project say more than marks.

When there are no numbers, there are still honest results: a before and after in words, a proxy measure: A number that moves with the result you care about when the result itself was not measured, such as support tickets standing in for how often users were stuck. such as support tickets, a practice another team took up, or the plain fact that you did not measure it, with what you would measure now. Never invent a figure; a number invites follow-up questions, and an invented one cannot survive them. When a figure is confidential, a range, a relative change: A change given as a share of the starting value, such as 'about a fifth more' or '30 % lower', instead of as an amount. or an order of magnitude keeps the evidence without the amount.

The checker below reads four endings and says what each is still missing. Press Edit, put the endings of your own stories in ENDINGS, and run it again:

Four endings, checked for evidence and for your part JavaScript · result-checker/result_checker.mjs

// A result checker: for each ending of a story, what kind of evidence it gives, whether it says what was yours, and
// what an interviewer would still have to ask. Replace ENDINGS with the endings of your own stories and run it again.
// The four endings below are invented.

const ENDINGS = [
  'It went really well and everyone was happy.',
  'The page got a lot faster after the change.',
  'After my query fix, p95 latency fell from about 2 seconds to about 400 ms.',
  'We did not measure it, but the support team stopped forwarding these complaints to us.',
];

// A number (digits or a number word) or a percentage.
const NUMBER = /\d|%|\b(?:one|two|three|four|five|six|seven|eight|nine|ten|half|twice|double|halved|tripled)\b/i;
// A change someone could see, without a number.
const VISIBLE = /\b(?:stopped|started|adopted|replaced|removed|no longer|now uses|switched to|signed up|renewed)\b/i;
// Words that say something changed but not by how much.
const VAGUE = /\b(?:went (?:really )?well|a lot|much better|big success|significant(?:ly)?|improved|faster|slower|happy)\b/i;
// The speaker's own part.
const MINE = /\b(?:I|[Mm]y|[Mm]e|[Mm]ine)\b/;
const GAP = /\b(?:did not|didn't|never) measure/i;

ENDINGS.forEach((ending, i) => {
  const evidence = NUMBER.test(ending) ? 'a number' : VISIBLE.test(ending) ? 'a change someone could see' : 'none';
  const notes = [];
  if (evidence === 'none') {
    const vague = VAGUE.exec(ending);
    notes.push(vague ? `"${vague[0]}" says it changed, not by how much or how you know` : 'no evidence that anything changed');
  }
  if (!MINE.test(ending)) notes.push('your part is not in it: say what changed because of what you did');
  if (GAP.test(ending)) notes.push('an honest gap: add what you would measure next time');
  console.log(`${i + 1}. ${ending}`);
  console.log(`   evidence: ${evidence}`);
  for (const note of notes) console.log(`   ask: ${note}`);
  if (!notes.length) console.log('   ask: nothing; this ending can be written down as it is');
});

Output

1. It went really well and everyone was happy.
   evidence: none
   ask: "went really well" says it changed, not by how much or how you know
   ask: your part is not in it: say what changed because of what you did
2. The page got a lot faster after the change.
   evidence: none
   ask: "a lot" says it changed, not by how much or how you know
   ask: your part is not in it: say what changed because of what you did
3. After my query fix, p95 latency fell from about 2 seconds to about 400 ms.
   evidence: a number
   ask: nothing; this ending can be written down as it is
4. We did not measure it, but the support team stopped forwarding these complaints to us.
   evidence: a change someone could see
   ask: your part is not in it: say what changed because of what you did
   ask: an honest gap: add what you would measure next time

Recorded with Node.js 24.21.0 on macOS 26 arm64. To run it yourself: mise exec node@24.21.0 -- node result_checker.mjs

Only the third ending passes as it stands. The first two say that something changed but not by how much, and the fourth, an honest gap, still lacks its owner: whose work made the complaints stop?

Checkpoint

Check: A result with evidence

One question Which endings give the interviewer evidence? Choose every one.
Choose every answer that is right.

"It was a huge improvement." "My manager said it was great work." "Releases took about half as long as before." "The design team now uses my component library for every new screen." "Build time fell from 14 to 6 minutes."
Show the answer: One question
Answer:
- "Releases took about half as long as before."
- "The design team now uses my component library for every new screen."
- "Build time fell from 14 to 6 minutes."
Evidence comes as numbers with a before and an after, a comparison, or a change someone could see. Adjectives and praise tell the interviewer how people felt, not what changed.

Your part, the team’s part and luck

An interviewer scores what you did, so a result needs one more thing: how much of it was yours. Say what changed because of you and what others or circumstances added: “My query fix cut the The response time that 95 % of requests beat: the slowest 5 % take longer, so it shows how slow the slow cases are. latency from about 2 seconds to about 400 ms; a cache a colleague added later took it lower.” That is stronger than claiming the whole improvement, because it can be checked and it survives the question “what else changed?”. Step through how that improvement is shared out:

Interactive · stage by stage Splitting one result between you, others and luck Step through one improvement as it is credited. Each step takes one cause and says how much of the change it explains.

The drawing shows the whole scene; its 5 steps are listed below.

Every step as text

The whole change. Over a month, an endpoint's p95 latency fell from about 2 seconds to about 300 ms. Claiming all of it is tempting, and the first follow-up about other changes would undo it.
Your part. The speaker's query fix came first and took it from about 2 seconds to about 400 ms, measured on the same weekdays before and after the fix.
A colleague's part. A cache a colleague added two weeks later took it from about 400 ms to about 300 ms. Saying so costs nothing and makes the speaker's own number believable.
Circumstances. Traffic was lower in one holiday week, which would have lowered latency on its own. Comparing the same ordinary weekdays before and after keeps that week out of both numbers.
The honest sentence. 'My query fix cut the p95 from about 2 seconds to about 400 ms; a cache a colleague added later took it lower.' Each clause names a cause and a size, so each can be checked.

Luck needs the same honesty. If traffic fell in a holiday week, errors would have fallen without you, so compare like with like: the same weekdays, the same season, the same load. Expect a number to be followed by questions about how it was measured, against what baseline: The value of a measure before a change, taken in comparable conditions, against which the result after the change is compared. and what else changed at the time.

Checkpoint

Check: Your part, the team's and luck

One question Your fix cut a page's load time from 5 to 3 seconds, and a teammate's caching later cut it to 1 second. Which ending is the strongest?
Choose one answer.

"I cut the load time from 5 seconds to 1." "My fix cut it from 5 to 3 seconds; a colleague's cache later took it to 1." "We cut the load time from 5 seconds to 1." "The page became much faster."

Show the answer: One question

Answer: "My fix cut it from 5 to 3 seconds; a colleague's cache later took it to 1."

Say what changed because of you and what others added. A shared result told honestly is stronger than a whole result claimed, because every part of it can be checked.

End with what you learned, and show that it changed you

Microsoft’s version of STAR ends with a fifth part, Reflection, and Meta’s guide asks candidates to share what they learned from a past situation, adding that being open about a project that was not a success won’t be counted against you. The reflection is the last thing the interviewer hears, so give it a sentence or two that show what you learned. Endings come in three strengths:

Ending	What it shows
“It was a great learning experience.”	A verdict, with nothing learned in it
“I learned to test with real volumes.”	A lesson
“Since then I replay our busiest day before any change to a batch job, and two releases later it caught a memory problem.”	A changed habit, used again

The third is the strongest because it can be checked: it says what you do now and when it mattered again.

Key idea

End with what you learned as a habit you changed and have used since: something the interviewer can see, not a verdict on how it went.

Checkpoint

Check: A reflection that shows change

One question Which reflection would an interviewer rate highest?
Choose one answer.

"Since then I write the rollback steps before every release, and they saved us an hour during an outage last quarter." "I would never make that mistake again." "It was a great learning experience." "I learned that releases are risky."

Show the answer: One question

Answer: "Since then I write the rollback steps before every release, and they saved us an hour during an outage last quarter."

A reflection is strongest as a habit you changed and have used since, with a time it mattered again. Lessons without a change, and verdicts, show far less.

Six vague results, rewritten with the evidence behind them Pro: opens with a Pro, Premium or Ultimate pass

Six endings of the kind interviewers hear every day, each rewritten with the evidence it was missing. The rules of the result checker above run on all twelve and show what changed.

In this part:

Six endings, before and after
What each rewrite added
When the number does not exist

1 runnable example (JavaScript) · about 5 minutes

Opening this part with your pass…

This part opens with any MySmartCoPilot pass.

Pro pass · 1 week₹79$2.99
30 days₹179$5.99
1 year₹1,699$45

One payment; a pass never renews. The price shown is the total you pay: no fees or charges are added at checkout.

Choose a pass I have a licence key

It also opens every other in-depth part in Learn and every Pro tool. Sales are final, except where the law gives you a right to withdraw, for example in the EU and the UK (refund policy).

Confidential figures, and the questions your numbers invite Pro: opens with a Pro, Premium or Ultimate pass

Some of your best numbers belong to your employer or a client. This part shows how to keep the evidence while hiding the figure, then how to answer the questions that follow any number you give.

In this part:

From an exact figure to a relative change, with the arithmetic
Five questions a number invites, answered
A number that changes between interviewers

1 runnable example (JavaScript) · about 6 minutes

Opening this part with your pass…Pro part. It opens with the same pass as the part above.

Practise: rewrite your endings

Put the endings of your three best stories into the checker in this lesson’s first section and fix what it asks for. Then test yourself on the ideas of the first three sections:

Interview questions

The first is a warm-up, answered here. Each of the others needs a story of your own with numbers in it: answer aloud before you open its first step.

Fresher to mid · Warm-up

What can you say about a result when you have no numbers?

Say what changed in a way someone could check, even without a number. A before and after in words works: "releases used to need a weekend; now they go out on a weekday afternoon". So does a proxy, a number that moves with the result, such as support tickets about the problem, and so does a change someone could see, such as another team adopting your tool. If nothing was measured, say so, and say what you would measure now: "We did not log it then; I would count pages per week from the start." What you should never do is invent a figure. A number invites follow-up questions about how it was measured, and an invented one does not survive them.
Mid to senior

Tell me about a time you improved something measurably. How do you know your work caused the improvement?

Model answer, Pro. Model answers open with any pass; prices are in the notice on this page. First step and follow-ups
Mid to senior

What was the business impact of the most important project you have worked on?

Model answer, Pro. First step and follow-ups
Fresher

Your final-year project got a good grade. What did it achieve beyond the grade?

Model answer, Pro. First step and follow-ups

Key takeaways

A result is a change with its evidence: numbers with a before and an after, a comparison, or a change someone could see.
When there are no numbers, use a proxy or a visible change, or say honestly that you did not measure it; never invent a figure.
Say what changed because of you and what others or luck added, and compare like with like.
End with what you learned as a habit you changed and have used since.

Terms in this lesson

baseline: The value of a measure before a change, taken in comparable conditions, against which the result after the change is compared.
proxy measure: A number that moves with the result you care about when the result itself was not measured, such as support tickets standing in for how often users were stuck.
relative change: A change given as a share of the starting value, such as 'about a fifth more' or '30 % lower', instead of as an amount.

References

Interviewing at Google: interview tips (Google Careers)
Software Engineer Full Loop Interview Guide (linked from the onsite preparation page) (Meta Careers)
Interview tips (Microsoft Careers)
Atlassian engineering interview guide (Atlassian)

Report a problem with this lesson

Your country

Results, numbers and what you learned

What you will learn

Before you start

What you will learn

Before you start

A result is a change, with its evidence

Your part, the team’s part and luck

End with what you learned, and show that it changed you

Six vague results, rewritten with the evidence behind them Pro: opens with a Pro, Premium or Ultimate pass

Confidential figures, and the questions your numbers invite Pro: opens with a Pro, Premium or Ultimate pass

Practise: rewrite your endings

Interview questions

What can you say about a result when you have no numbers?

Tell me about a time you improved something measurably. How do you know your work caused the improvement?

Your final-year project got a good grade. What did it achieve beyond the grade?

Key takeaways

References

Related tools

What you will learn

Before you start

What you will learn

Before you start

A result is a change, with its evidence

Your part, the team’s part and luck

End with what you learned, and show that it changed you

Six vague results, rewritten with the evidence behind them Pro: opens with a Pro, Premium or Ultimate pass

Confidential figures, and the questions your numbers invite Pro: opens with a Pro, Premium or Ultimate pass

Practise: rewrite your endings

Interview questions

What can you say about a result when you have no numbers?

Tell me about a time you improved something measurably. How do you know your work caused the improvement?

What was the business impact of the most important project you have worked on?

Your final-year project got a good grade. What did it achieve beyond the grade?

Key takeaways

References

Related tools