You can read a hundred engineering blog posts about a company’s “values” and still walk into your first week genuinely surprised by what the place is actually like. That gap between the story a company tells and the way it actually behaves is where most career disappointments quietly come from. This piece is about closing that gap — not with a checklist you tick off in a single interview, but with a way of paying attention that keeps working long after you have said yes.

Culture is a strange thing to evaluate, because it rarely announces itself directly. Nobody schedules a meeting called “here is how we treat each other when things go wrong.” You mostly find out by living inside it — by watching how a disagreement in a pull request gets resolved, by noticing whether the on-call engineer sounds exhausted or steady on a Tuesday standup, by clocking who gets credit when a project ships and who quietly absorbs the blame when it does not.

This is exactly why so many smart, careful people still end up somewhere that does not fit them. Not because they did not ask questions in the interview, but because the questions they asked were answered with the company’s rehearsed best self, and rehearsed best selves are, understandably, not where the truth tends to live. A healthy culture and an unhealthy one can sound almost identical when a hiring manager is trying to make a good impression. The difference shows up in behaviour, not in language — and behaviour is exactly what an interview process is designed to hide, gently, on both sides.

None of this means interviews are useless, or that companies are lying to you. Most people describing their team’s culture genuinely believe what they are saying; they are just describing it from the inside, where blind spots are hardest to see. Your job, as someone evaluating a culture from the outside, is to look past the description and toward the evidence sitting quietly underneath it.

It also helps to remember that culture is not one single thing spread evenly across an entire company. A large organisation can hold several genuinely different cultures inside it at once — one team that is thoughtful and calm, another two floors away that is chaotic and anxious, both technically operating under the same set of company values printed on the same wall. When you evaluate “a company’s culture,” you are really evaluating the specific team, the specific manager, and the specific slice of the organisation you would actually be joining. Broad reputation matters less than most people assume; the immediate, local reality matters far more.

The single biggest shift you can make in how you evaluate a company’s culture is small but powerful: spend more of your energy watching, and less of it asking. A direct question — “is your culture healthy?” — will almost always get you a direct, positive answer, because it is an easy question to answer well regardless of what is actually true. What you want instead are situations where the answer reveals itself without anyone needing to perform it.

During interviews, this means paying attention to things most candidates barely notice. How does your interviewer talk about a teammate who is not in the room — with warmth and specificity, or with vague, careful diplomacy that feels rehearsed? When you ask about a recent project that did not go well, does the answer include what was learned, or does it quietly deflect blame onto “the business” or “leadership” in a way that suggests a habit, not a one-off? Does anyone laugh, genuinely, at any point in the conversation? Small, ordinary human warmth is surprisingly hard to fake convincingly across an entire interview loop.

If you get the chance to sit in on any real working moment — a team standup, a pairing session, even a few minutes shadowing a code review — take it seriously, even if it feels like a small, informal part of the process. Five minutes of watching a team actually work together often tells you more than forty-five minutes of a curated conversation about how they work together.

Remote interviews add their own small wrinkle worth watching for. Without the hallway chatter, the shared lunch, or the incidental five minutes waiting for a meeting room, you lose a lot of the ordinary texture that used to reveal culture almost by accident. It is worth deliberately asking for the remote equivalent — a short informal call with a would-be teammate, a look at how the team communicates asynchronously, even a sample of a real internal document if the company is comfortable sharing one. The goal is the same either way: find any window, however small, into how the team behaves when nobody is specifically performing for a candidate.

There is also something quietly useful in what you do not see or hear. If nobody mentions a customer by name across an entire interview loop, that absence is a data point. If nobody references a real, specific colleague on another team, that is a data point too. Healthy cultures tend to leak specifics almost without meaning to — the person one desk over gets mentioned, the last outage gets remembered, the customer who wrote the funny bug report gets brought up. Interview loops that stay entirely abstract are often abstract for a reason, whether or not anyone in the room realises it.

A handful of concrete, observable signals hold far more weight than any stated value, because they are expensive to fake and cheap to notice if you know to look. None of these alone tells the whole story, but together they paint a picture that is very hard for a company to dress up convincingly.

How incidents and mistakes get discussed

Ask, plainly, what happens after something breaks in production. A healthy answer usually involves a blameless review focused on the system and the process, not the person who happened to be holding the pager. An unhealthy one often includes a story, told half-jokingly, about someone who “should have known better” — a small tell that mistakes are quietly tracked against individuals, even when nobody says so directly.

The tone inside code reviews

If you can see any real pull request history — sometimes shared during interviews, sometimes visible on an open-source project the company maintains — read a few comment threads closely. Healthy review culture tends to sound curious: questions, suggestions, the occasional compliment on a clean solution. Unhealthy review culture tends to sound territorial or clipped, with terse corrections and very little acknowledgement of effort.

Who does the talking in group settings

In any interview involving more than one person from the company, notice the balance of voices. Does one senior person answer every question while everyone else nods along quietly, or does the group naturally pass the conversation between them, each adding something the others did not? The second pattern, more often than not, reflects a team where opinions genuinely get heard day to day, not just in front of a candidate.

How on-call and after-hours work is described

Ask specifically what a bad on-call week looks like, not just what the rotation schedule is. A team with a healthy relationship to on-call will usually be able to describe concrete steps they have taken to reduce alert fatigue — better runbooks, saner thresholds, a genuine effort to fix root causes instead of just silencing the noise. A team without that story often reveals, without meaning to, that on-call pain is simply accepted as the cost of the job.

Tenure, and the honesty around it

Short average tenure is not automatically a red flag — some fields and career stages naturally involve more movement — but it is worth asking about directly, and watching closely for how comfortable the person answering seems with the question. A team that can talk plainly about turnover, including the reasons behind it, tends to be healthier than one that visibly bristles or deflects when the topic comes up.

Read the table sideways rather than top to bottom. Any single row can go either way for perfectly innocent reasons — someone was tired, someone was new, someone was uncomfortable on camera. What matters is whether the column on the left keeps showing up across several different conversations, or whether the column on the right keeps quietly reappearing in different rooms with different people. A pattern is much harder to fake, in either direction, than any single sentence.

The way you phrase a question shapes how honest the answer can be. A question that sounds like an exam — “how would you describe your culture?” — invites a rehearsed, polished response. A question that sounds like genuine curiosity about a real moment tends to invite a real story instead, because it is harder to have a rehearsed answer ready for it.

None of these questions are traps, and they should not be asked like one. Asked with genuine warmth and curiosity, they usually get warmth and honesty back — and even the discomfort of an evasive answer tells you something real and useful about what you would be walking into.

The trickiest warning signs are the ones that arrive wrapped in genuinely pleasant-sounding language. They are not lies; they are just phrases that can mean either something healthy or something rougher underneath, and the difference is worth learning to hear.

Just as some warning signs hide inside pleasant-sounding phrases, some of the healthiest signals are easy to miss because they do not announce themselves at all — they show up as small, understated moments that a distracted candidate might not even register.

None of these green flags is proof on its own; the point is to notice how many of them show up naturally, without you having to steer the conversation to reach them. One or two across a long interview loop is fairly ordinary. Four or five, arriving without prompting, is a small quiet chorus worth listening to carefully.

Even the most careful evaluation before joining is still an evaluation from the outside, and the first few weeks on the inside will teach you things no interview could. It is worth paying attention on purpose, rather than simply letting the onboarding blur happen to you.

Notice how your first pull request gets reviewed. Is feedback specific and kind, or sparse and cold? Notice whether anyone checks in on you without being asked to, and whether that check-in feels genuine or like a box being ticked on an onboarding checklist. Notice how the team talks about customers, about leadership, about each other, when they think the conversation is just ordinary and unremarkable — those unremarkable moments are usually the most honest data you will ever get.

Give yourself permission to trust what you observe during this period, even if it contradicts the story you were told during hiring. It is uncomfortable to admit a mismatch that early, especially after all the effort of getting the offer, but the discomfort of noticing early is almost always smaller than the cost of ignoring it for a year.

It is also worth keeping a small, private note of your own early impressions — not to build a case against anyone, just to protect your own memory from the natural pull toward rationalisation. A few months into any job, it is remarkably easy to talk yourself into believing a rough start was normal, or that you simply had not adjusted yet. Sometimes that is true. But a short list written honestly in week two is a much more reliable witness than a memory reshaped, gently and understandably, by the desire for the decision to have been the right one.

A single small moment inside an otherwise unremarkable interview can, sometimes, tell you more than the entire rest of the process combined. Here is one such moment, from a real engineer weighing two nearly identical offers.

A software engineer I know — call her Meera — once interviewed at two companies in the same month, for nearly identical roles. Both interview loops were polished, both teams described themselves as collaborative and low-ego, and both offers landed within a few thousand dollars of each other. On paper, the decision looked like a coin flip.

What tipped it, in the end, was not anything either company said directly. In her final round at the first company, she asked about a recent project that had slipped its deadline. The engineer answering paused, genuinely thought about it, and described exactly what had gone wrong — an underestimated integration, a dependency they had not accounted for — and what they had since changed about how they scoped similar work. It felt like a real memory, not a rehearsed one.

At the second company, the same question got a smoother, more polished answer: “we are pretty good about hitting our deadlines, and when we do not, we just communicate early and reset expectations.” Technically a fine answer. But it had no texture to it — no specific project, no specific lesson, nothing that felt like it had actually happened to a real team of real people.

Meera took the first offer. Eight months in, she told me the honesty in that one small answer had turned out to be a genuinely accurate preview of the team — a place where mistakes got named plainly and fixed, rather than smoothed over with reassuring language. She still does not know what the second team was actually like day to day. But she is fairly confident that the difference in how they answered one ordinary interview question told her something true, long before she could have proven it any other way.

What stayed with her afterward was not the specific answer itself, but the realisation that she had almost missed it — that in the moment, both answers had sounded perfectly reasonable, and it was only in comparing the texture of one against the other that the real difference became visible. She has since made a habit of asking the same kind of question in every interview she takes, not because it guarantees a right answer, but because it reliably tells her something the polished parts of the conversation never quite manage to.

It is easy to let a genuinely nice benefits package quietly stand in for a real answer about culture, especially when you are tired of interviewing and the offer looks good on paper. Free lunches, a generous learning budget, a beautiful office — none of these are bad things, and none of them tell you much at all about how people actually treat each other under pressure.

Perks are visible, easy to list on a careers page, and relatively cheap for a company to offer compared to the harder, slower work of building genuine trust and psychological safety across a team. A company can have wonderful perks and a genuinely difficult culture underneath them; the perks, if anything, can sometimes function as a kind of glossy distraction from that gap, whether or not that is anyone’s intention.

This is not a reason to dismiss perks entirely — they are a real, legitimate part of compensation and daily quality of life. It is simply a reminder to evaluate them as their own separate category, not as evidence about culture. A team can be wonderful with a modest benefits package, and a team can be unhealthy with an extraordinary one. Keep the two questions distinct in your own mind, even when a company’s marketing tries hard to blur them together.

The clearest way to keep them separate in your own head is to answer them as two independent questions, in writing, before comparing offers. What is the compensation and benefits picture on its own terms? And — entirely separately — what have you seen about how the team actually behaves? Blending the two saves time and reliably regretted decisions in roughly equal measure.

If you can find a way to talk with a current employee outside the official interview process, or a former one who is willing to be candid, take that conversation seriously — it is often the single richest source of honest information available to you. People tend to speak far more freely once they are not being evaluated, or once they no longer have anything at stake in how the company is perceived.

Approach these conversations with genuine curiosity rather than a hunt for gossip. Ask what a normal Tuesday looks like, rather than asking directly “is the culture good?” Ask what they would tell a friend who was considering an offer there. Ask, gently, what they wish they had known before joining. People are usually generous with honest answers when a question feels like real curiosity rather than an attempt to extract a verdict.

Weigh what you hear with a little care, the same way you would any single data point. One person’s difficult experience might reflect a genuine, widespread pattern, or it might reflect a particular team, a particular manager, or a particular season the company was going through. Where possible, look for the pattern across a few different conversations rather than resting your whole decision on any single account — including a glowing one.

It is also worth being generous back. If someone takes twenty minutes of their evening to talk honestly with you about a company they used to work at, remember that generosity later — introduce them to someone useful, answer their questions the same way if the roles ever reverse, thank them specifically rather than generically. The small ecosystem of people willing to speak plainly about their workplaces runs on that kind of quiet reciprocity, and it is worth participating in with real care.

Everything so far has been written from the vantage point of someone evaluating a culture from outside it — a candidate, a curious observer, someone deciding whether to say yes. But if you lead a team, the same lenses are worth turning inward, on a regular basis, before an outsider ever has to figure them out for themselves through careful detective work.

The uncomfortable truth is that most leaders have a distorted view of their own team’s culture, for a very ordinary reason: people tend to soften what they say to the person who holds power over their role, their raise, and their next promotion. If you only ever hear about your team’s culture from your team’s culture, you are hearing an edited version, filtered through exactly the same instinct to present a polished best self that shows up in interviews. This is not dishonesty on anyone’s part — it is just how hierarchy quietly shapes what feels safe to say out loud.

Ask the same questions of yourself that a candidate would ask of you

Sit down, honestly, with the questions from earlier in this piece. What would your team actually say happened after your last production incident — a calm, blameless review, or a story about who “should have known better”? Could every person on your team describe a real moment where the group changed its mind about something, based on evidence rather than seniority? If you are not confident in the answer, that uncertainty is itself worth taking seriously.

Create moments where honesty costs less

Anonymous, genuinely acted-upon feedback channels; skip-level conversations that are not secretly reported back; exit interviews conducted by someone outside the direct reporting line — all of these exist to lower the cost of honesty, because honesty that costs something will always be rationed. The specific mechanism matters less than the underlying intention: making it genuinely safer for the truth to reach you, even when the truth is uncomfortable to hear.

Watch what you reward, not just what you say you value

A team learns its real values by watching who gets promoted, who gets praised in public, and whose work quietly goes unrecognised — far more than by reading a values document, however well written. If a document says “we value sustainable pace” while the person who burns themselves out delivering a rushed feature gets celebrated in the next all-hands, the document will lose every single time. Culture is built from what gets rewarded in practice, not from what gets written down with good intentions.

Evaluating a culture honestly, whether you are joining it or building it, takes the same basic ingredient either way: a willingness to look past the comfortable, polished version of the story and toward the smaller, quieter evidence sitting underneath it. That willingness is rarer than it should be, and it is worth practising deliberately, in both directions.

Short, direct answers to the questions that come up most often when engineers start taking culture evaluation seriously as a discipline of its own — on both sides of the interview table.

What if I cannot get an honest read at all during the interview process?

This happens, especially with very polished, high-volume hiring processes where every interviewer is trained to say the same reassuring things. If nothing you ask produces a specific, textured answer anywhere in the loop, treat that flatness itself as information. A healthy culture usually leaves at least a few small cracks of honesty somewhere in the process, even by accident.

Is a small startup automatically riskier on culture than an established company?

Not automatically, in either direction. Small teams can have unusually tight, honest, low-ego cultures precisely because there is nowhere for dysfunction to hide — or they can have deeply unhealthy ones built entirely around one founder’s temperament, with no HR structure to soften it. Size changes the shape of the risk, not whether the risk exists. The same careful observation applies either way.

How much weight should online reviews carry in this evaluation?

Treat them as one more data point, not a verdict. Anonymous review platforms tend to attract disproportionately strong feelings, both glowing and furious, and rarely reflect the quiet, ordinary middle where most people’s actual daily experience lives. Look for recurring, specific themes across many reviews rather than trusting any single dramatic one.

What if the culture seems healthy but the technical work itself does not excite me?

These are genuinely separate questions worth answering on their own terms, and it is worth being honest with yourself about which one matters more to you at this point in your career. A healthy team doing less exciting work is, for many people, a better place to spend a few years than an unhealthy team doing dazzling work — but that trade-off is personal, and there is no universally correct answer.

Can a healthy culture turn unhealthy later, even after I have verified it carefully?

Yes, and it is worth holding that possibility honestly rather than treating any single evaluation as permanent. Leadership changes, acquisitions, and periods of rapid growth can shift a culture meaningfully, sometimes quickly. The same quiet, ongoing attention you used to evaluate a culture before joining is worth keeping up gently after joining too, not as anxiety, just as steady awareness.

What if I only have one or two interview rounds to work with, not a long process?

Shorter processes simply mean being more deliberate with the limited time you have, rather than giving up on evaluation altogether. Pick two or three of the questions from earlier in this piece and ask them with real intention rather than trying to cover everything. A single well-asked, specific question about a real moment often reveals more in ninety seconds than a dozen generic ones spread thin across an hour.

Does any of this change if I am evaluating a fully remote team?

The signals are the same; the surfaces you read them on are different. Instead of watching a hallway conversation, you read the tone of a shared document or a written message thread. Instead of noticing who talks in a meeting, you notice who participates in a written discussion and whether replies feel curious or clipped. The underlying question — is this a team where honesty and care are ordinary — is unchanged. Only the evidence looks slightly different.

There is no single question, no perfect script, that guarantees you will see a company’s culture clearly before you have actually lived inside it for a while. What careful attention can offer instead is something more modest and more useful: a much better guess, built on real evidence rather than polished language, and a habit of noticing that keeps serving you well past the interview itself.