# Tell me about a time you used data to make a decision: examples

Tell a short story in five steps: the question you needed to answer, the data you gathered and why you trusted it, what the data showed, the decision you made, and the result. Include one thing the data couldn't tell you and how you handled it. You don't need to be an analyst: a spreadsheet, a customer log or a simple count can be the data.

By the Speaking Coach team at Sodhera Intelligence. Updated 2026-10-22. Source: https://www.orecci.com/speaking-coach/interview-questions/tell-me-about-a-time-you-used-data-to-make-a-decision/

## What the interviewer is checking

This is a behavioural question about judgement. The interviewer wants to see that you:

- **Start from a real question,** not from data you happened to have.
- **Know where the data came from** and whether it's trustworthy.
- **Read it sensibly,** without overclaiming.
- **Actually decided something** and acted on it.
- **Know the limits** of what the data told you.

It's asked far beyond analyst roles: in retail, operations, nursing, teaching, marketing and sales. For those roles, a simple count you kept yourself is perfectly good data.

## A five-step structure

The [STAR method](/speaking-coach/guides/star-method/) works, but for this question it helps to split the middle so the data part is clear.

1. **The question.** What decision needed making, and why it mattered. "We had to decide which two of our five ad channels to keep after a budget cut."
2. **The data.** What you used and why you trusted it. "Three months of cost and sign-up data from each channel, which I cross-checked against our sign-up form's 'how did you hear about us' answers."
3. **What it showed.** The finding in one or two sentences. "Two channels brought in nearly all of the paying customers."
4. **The decision.** What you decided or recommended, and anything that pushed against it.
5. **The result.** What happened, ideally with a number. Plus one limit or lesson.

Aim for 90 seconds to two minutes.

## Weak versus strong elements

| Element | Weak | Strong |
| --- | --- | --- |
| Question | "I was looking at some data and noticed..." | "We needed to decide whether to add a night shift." |
| Data | "I used our analytics." | "Six weeks of hourly order counts from the till system." |
| Trust | Not mentioned | "I removed two weeks affected by a system outage." |
| Finding | "The data was really interesting." | "Over 40 percent of weekday orders came after 8 p.m." |
| Decision | "We made some changes." | "I recommended a short evening shift instead of a full night shift." |
| Limits | Not mentioned | "The data didn't tell us about customers who left because we were closed, so we tested it for a month first." |

## Example answers

These use made-up details. Use your own story and real numbers.

### Retail store supervisor

> "Our store manager wanted to know if we should extend weekday opening hours. I pulled six weeks of hourly sales from the till system and counted how many customers we turned away at the door at closing, which I'd started noting on a sheet. Sales in the last opening hour were much higher than earlier in the evening, and we turned away around 15 people a night. I recommended trying an extra hour on Thursdays and Fridays first, rather than every day. Over the two-month trial, those two evenings covered the extra staff cost, so we kept it. The one thing the data couldn't tell us was whether those were new customers or just people shifting their visit, so we watched overall weekly sales too."

### Marketing coordinator

> "After a budget cut we had to drop two of five ad channels. I put together three months of spend and sign-ups per channel, then checked them against the 'how did you hear about us' answers on our sign-up form, because our tracking had a known gap with one channel. Two channels brought in most of the paying customers. But one cheap channel looked bad on sign-ups while customers from it stayed much longer. I recommended keeping that one plus the strongest performer. Six months later, cost per paying customer was lower than before the cut."

### Customer support team lead

> "Our team was falling behind on tickets, and the instinct was to hire. Before asking for headcount, I tagged a month of tickets by type. A third were password reset requests that customers could do themselves if the reset link was easier to find. I worked with the product team to move the link, and those tickets dropped by more than half within six weeks. We still hired one person, but one rather than the three we'd planned."

### Graduate

> "In my final-year group project, we had to choose between two designs for a campus recycling bin layout. Instead of voting, I suggested we count how much waste went in the wrong bin at two test sites for a week. The count showed the colour-coded design cut wrong-bin waste noticeably. We chose it and the university facilities team adopted it for one building."

## Mistakes to avoid

| Mistake | Why it hurts | Try instead |
| --- | --- | --- |
| Starting with the data, not the question | Suggests you look at numbers without purpose | Lead with the decision that needed making |
| Too much technical detail on methods | Loses non-technical interviewers | One sentence on the method, more on the finding and decision |
| Overclaiming certainty | Interviewers know data has limits | Name one limit and how you handled it |
| No decision | It becomes a story about analysis, not judgement | Say clearly what you decided |
| Invented or exaggerated numbers | Follow-ups expose them | Use real figures or honest rough sizes |
| "The data told us what to do" | Sounds like you didn't use judgement | Show where you weighed the data against other factors |

## Expect a challenge on your data

A good interviewer will push: "How did you know the data was accurate?", "What else could explain that?", "What if the data had said the opposite?" Prepare one sentence for each. Practise answering those out loud, because a calm response to a challenge is often what the interviewer is really testing. Speaking Coach's interviewer follows up on what you actually said, so it can press on the weakest part of your data story.

For analyst roles, see [data analyst interview questions](/speaking-coach/interview-questions-by-role/data-analyst/) for the technical side.

## Questions people ask

### What if I'm not in a data role?

Use any decision where you counted or measured something first: customer complaints, sales by hour, time spent on tasks, survey answers. Simple data, used well, is a strong answer.

### What if the data-driven decision turned out wrong?

That can still be a good story. Say what the data missed, what you learned, and what you check now.

### How technical should I be?

Match the role. For analyst roles, mention your method briefly and expect follow-ups. For other roles, focus on the question, the finding and the decision.

### Should I mention the tools I used?

Briefly, if they matter to the job: "I pulled it into a spreadsheet and built a simple pivot table." Don't let tools take over the story.

### What if my manager made the final decision?

Say what you recommended and why, then what was decided. Recommending well, based on data, is the skill they're looking for.

---
Speaking Coach on the App Store: https://apps.apple.com/app/id6767124039
Speaking Coach on Google Play: https://play.google.com/store/apps/details?id=com.sodhera.speakingcoach