Learn

Your story says it got faster, and the next question is how you know

6 November 2026 · Behavioral: your part of the story

"Tell me about the project you are proudest of." You have a good answer ready, with a number in it, because every guide told you to quantify your impact. You say it. The interviewer writes nothing down and asks: "How did you measure that?" The next minute decides whether the number helped you or hurt you.

The situation

Behavioral rounds reward specific results, so candidates learn to end stories with a figure. The trouble is that a number is not the end of a story. It is an invitation. An experienced interviewer will pull on it, because the follow-up shows more than the figure itself: whether you know where it came from, what else could explain it, and who actually felt the difference.

A post on DEV this week made the same point about CVs. Its test was an interviewer pointing at a line and asking, "Can you show me where you did that?" A spoken story gets the same test, just without the repository open.

What goes wrong

Here is an invented story of the common kind:

Our nightly reporting job was slow. I rewrote the slowest step to load data in batches instead of row by row, and made it five times faster. The business was really happy.

Now put three follow-ups to it.

• "Five times faster than what, measured how?" If the honest answer is "it felt about that much", the number is now a liability.
• "Did anything else change that month?" If the database was upgraded in the same week, the interviewer has just found a second explanation, and you did not mention it.
• "Who was really happy, and why?" "The business" is nobody. A result that no named role noticed sounds like a result nobody needed.

None of these questions is hostile. Each one simply asks for the part of the story that the number skipped.

Here is the same incident, told so that it survives them:

The nightly report took about three hours, and on bad nights it finished after the finance team started work, so they waited for it. I took two weeks of run times from the scheduler logs as the baseline. The slowest step loaded rows one at a time, so I changed it to load in batches. Over the next two weeks the job ran in roughly forty minutes. A database upgrade went out in the same window, so I ran the old step once on the new database to separate the two: most of the gain was the batching. The finance team stopped waiting for the report.

There is no multiplier in it, and it is stronger. Every claim has a source, the obvious alternative is ruled out, and the effect lands on a person.

What to say out loud

The shape is: the baseline and where it came from, the change you made, the result and how you checked it, and who noticed.

The sentence that carries the most weight is the one that names a competing explanation and deals with it: "A database upgrade went out the same week, so I ran the old code on the new database to separate the two." It shows you think about evidence, and it answers the hardest follow-up before it is asked.

If you did not measure, say so, and offer what you do have: "We did not track run times then. What I can tell you is that the finance team used to wait for the report most mornings, and after the change they stopped asking about it." An honest qualitative result beats a precise number that falls apart on the second question.

Round numbers only when the data was rough, and say it was rough. "About forty minutes" is defensible. "4.6 times faster" invites a question about the decimal you never had.

The exercise

1. Write down the impact story you would tell today. Underline every number and every word like "faster", "fewer" or "happier".
2. Next to each one, write where it came from: a log, a dashboard, a ticket count, or memory. Anything marked "memory" gets rounded or replaced with what was observed.
3. For the biggest claim, write one other thing that changed at the same time. Then write one sentence on how you know it was not the cause, or admit that you do not.
4. Replace "the business" or "the team" with the role that noticed, and what they stopped or started doing.
5. Tell the story aloud in under two minutes. Then ask yourself the three follow-ups above and answer each one in a single sentence. Repeat in three days without notes.

Where this fits in Gadfly

Gadfly has four rooms: algorithms, system design, behavioral and code review. The behavioral room keeps a bank of your stories, separates the competencies a story claims from the ones it actually earns, and asks you to retell a story later, as in the last step above. Its report is a set of observations made by rules over what you said, not a hiring prediction.