Tell me about a time you got something adopted when you had no authority to mandate it.

Why this question sits at the senior boundary

Below a certain level you can be effective doing excellent work inside a scope someone gave you. Above it you cannot, because the things worth doing cross boundaries you do not control — another team's endpoint, a platform group's roadmap, a product manager's quarter. So this question is the most reliable single discriminator between a strong senior engineer and someone with a senior title, and it is weighted accordingly at staff and lead level.

It is also where the difference between influence and insistence shows up fastest. Insistence is repeating a correct argument more loudly until someone gives way or you are overruled. Influence is changing what someone wants. The tell is whether anything about your own position moved during the story.

What separates the strong answers

Four things, and a strong answer usually has at least three.

The opposition was reasonable. If your story is that you wanted tests and your colleagues did not care about quality, you have not described influence, you have described being right in a room of straw men. The hard cases in real organisations are the ones where the other side has a legitimate concern: the platform team was committed for the quarter, the other lead had been burned by exactly this migration two years earlier, the PM was protecting a date promised to a customer. Naming their reason accurately is the strongest single signal here, and it costs you nothing.

You produced evidence rather than argument. Opinions cancel out; a measurement, a spike, a small prototype, a cost figure or a two-week trial on one service does not. The move that most often unlocks these situations is shrinking the ask until it is cheap enough to say yes to — one service, one sprint, reversible — and letting the result do the arguing.

You sequenced deliberately. You did not convince eight people at once; you convinced one person the others respected, or found whose objection was blocking everyone else's and dealt with that first. "I raised it in the team meeting and eventually people agreed" reads as luck.

You gave something up. Almost every genuine adoption story contains a concession: narrower scope, a longer timeline, a compatibility shim you did not want to maintain, an escape hatch for the most reluctant team. Candidates omit it because it feels like a weakness. It is the opposite — it is the evidence that a negotiation happened rather than a campaign.

Proving adoption, not approval

A decision meeting is not adoption. What counts is other people's behaviour after your attention moved elsewhere: teams using the pattern without being asked, the standard surviving in the codebase a year later, the runbook being edited by someone else, the practice appearing in a review you were not in. Interviewers probe this because unfinished influence is so common — approval won, rollout stalled, everyone quietly back to what they were doing.

Where the answer is honestly mixed, say so. "Two teams adopted it, the third never did because they were mid-rewrite and it would have been the wrong call for them" is a stronger ending than a claim of universal adoption, because it shows you can tell the difference between being right in general and right for everyone.

The overcorrection that quietly fails

The failure mode most candidates worry about is sounding arrogant, and so they overcorrect into an answer where nothing they did mattered — the data spoke for itself, everyone got on board, it was a team effort. That is the version that fails, because there is no influence in it. The rubric line here is did this person move an organisation, and a story in which the organisation moved on its own scores zero against it.

The harder failure is subtler: a story where you won because someone senior backed you. If the resolution is "I escalated to my director and she told them to do it", you have described borrowing authority, which is the thing the question excludes. Escalation is legitimate in real life and worth mentioning as a step you took carefully, but it cannot be the mechanism of the win. If it is, choose a different story.

Sample phrasing for the hardest sentence

The sentence candidates struggle with is the one where they describe someone else's objection without disparaging them. These constructions work, and are worth having ready:

Instead of	Say
"They were resistant to change"	"They had been through a failed migration two years earlier and wanted a rollback plan I had not thought about"
"The PM did not understand the tech debt"	"The PM was protecting a date she had committed to externally, which I had not factored in"
"I convinced them eventually"	"I dropped the parts that were expensive for them and got agreement on one service first"

A pattern to adapt, not to deliver verbatim

"We had four services writing to the same Postgres tables directly, and I wanted to put a single owning service in front of them. I had no authority over three of those four teams.

The objection was real and I had it wrong at first. I assumed the resistance was inertia; in fact the team with the heaviest usage had a nightly reporting job that would have gone from a single join to about a dozen calls, and they had a batch window they could not miss. Their position was defensible and my original proposal would have broken them.

So I did two things. I shrank the ask — instead of migrating everyone, I asked for one team, the smallest, for one sprint, with the direct table access left in place so it was reversible. And I changed the design to give the reporting team a read replica view rather than forcing them through the API, which is a concession I did not like because it kept a coupling I wanted gone, but it was the difference between a no and a yes.

I also went in a particular order. I spoke to the reporting team's tech lead first, one to one, before raising anything in a wider forum, because she was the person the other leads would look at. Once she was neutral rather than opposed, the wider conversation was about sequencing rather than about whether.

Where it got to: three of the four are behind the service now, about seven months on, and the constraint I care about — nobody has added a fifth direct writer — has held, including twice when I was not in the review that caught it. The fourth team never migrated. They were being deprecated anyway and it would have been wasted work, and I would make the same call again."

The features to copy are the admission that he misread the resistance, the concession he disliked, the named first conversation, and the honest incompleteness at the end.

Before you walk in

Have two of these ready, one technical and one not, because the follow-up is often "and a non-technical example?". For each, write down the other side's best argument in their words, the thing you conceded, the order you approached people in and why, and one piece of evidence that the change outlived your involvement. If you cannot state the other side's argument in a form they would agree with, you do not yet have an influence story — you have a story about being right.