What is the hardest thing about running a startup?

Someone asked me this recently – it’s been about a year since I left my teaching job to start OnSite – and it made me pause. I could reach for any number of answers. Fundraising. Sleep deprivation. And then the loneliness – not a general loneliness, but the specific kind where all your friends are in stable jobs and have no real clue why you left yours, and love you too much to ask directly, and admire your intensity but from a safe distance. Where the only people who truly understand are your co-founder and your earliest employees, because they’re the ones in the trenches with you, in the nitty-gritty minutiae where everything feels vital and urgent in a way that a regular job doesn’t constantly trigger.

But the answer that surfaced, the one I keep returning to, is this:

I think the hardest thing is the ability to hold two contradictions at once. To see reality clearly, and also to remain optimistic – genuinely, irrationally optimistic – about your company’s ability to crush the category.

Seeing reality clearly requires a certain discipline. You have to pay attention to how your customers are actually interacting with your product – not the story you tell yourself about how they interact with it. Separate signal from noise, index on usefulness and density. Founders who see reality are less surprised when shit happens, because they have been paying attention. They are also capable of having the kinds of uncomfortable but ultimately productive conversations about what’s working and what isn’t. You can’t brainstorm your way out of no-PMF when you don’t have shared reality.

And then on the other side, optimism. Being a founder self-selects for a particular kind of irrational belief – the kind that says, yes, I will sacrifice my sleep and health and income and social life to bring this thing, this hitherto nonexistent idea, into the world.

The problem is that most founder culture resolves this tension by collapsing it – by letting optimism swallow reality whole. When you stop seeing clearly, a kind of narrative inflation sets in. A customer emailing you about the price of your product gets recast as an unequivocal sign of product-market fit. Someone adding themselves to a demo waitlist becomes proof that you are crushing it. A soft landing gets packaged as an ‘acquisition’. Or my personal favourite – “I was too early, the market wasn’t ready for me” – usually offered as an explanation for why the company didn’t work out, right before the person tells you they’ve moved into VC.

Each of these, individually, might be harmless. Collectively, they form a kind of protective fiction – one that allows you to believe that PMF is just around the corner, that the problem will be solved if only you hire the right person, or throw money at it. It is also easy to justify. You’re not delusional, you’re thinking about team morale.

Some founders have this quality – an ineffable one – of creating a reality distortion field so strong that everyone around them begins to inhabit it. And for a while, it works. But I have also seen friends join startups and get burned, badly, because the founders either couldn’t see reality clearly or saw it and chose not to. At some point the field collapses and you’re left standing in the rubble of a story that was never true.

I joke that I’ve learned a whole new vocabulary since joining this industry, because the gap between what people say and what they actually mean is so large. Every interaction is an act of interpretive labour.

I once met someone at an event who told me he had exited and was raising a pre-seed round. Later I looked him up on LinkedIn. He described himself as a co-founder of a company that had exited for several billion dollars. Something didn’t add up. Why would you raise a small round if you presumably already had significant liquidity?

I dug into the details. His company had been acquired – for an undisclosed sum – by a parent company, which was later publicly listed, and at its peak hit a valuation of a few billion. The truer story – that he had built something real, in a genuinely interesting space, that made a social difference, and had maybe made a modest but respectable sum at an acquisition – would have been more impressive to me, though I suspect not to his investors. Instead I left the meeting feeling vaguely disappointed – not angry, just deflated, the way you feel when someone you wanted to like shows you something small about themselves.

What is strange is how ordinary these acts of spin are in the world I now inhabit. Almost nobody bats an eye. A merger that is clearly an acquihire. Numbers that are both facts and fiction, the unspoken knowledge that some figures are meant to be inflated while others are meant to serve a single story.

The thing is, I would have respected the person more if he had been honest. I would respect anyone more – even a founder who told me plainly that they’d raised a lot of money, failed to achieve PMF, returned the capital when they concluded it wasn’t working, and then raised a smaller round to pursue a new thesis. That story has texture. It has the weight of someone who has actually done something difficult. And many of the people I’ve met who engage in these small inflations have turned out to be perfectly normal, even lovely people, some of whom I now consider good friends. But the instinct to spin is so deeply embedded that I think most of them don’t even notice it.

Maybe the explanation is structural. Competition for capital is fierce. The incentives are clear. If every founder around you is inflating their narrative by ten percent, then telling the truth feels like unilateral disarmament. I can extend that charity. There is a whole ecosystem built around this – strategic narrative advisors, PR agencies, communications consultants hired to help founders curate their ‘signature worldview’ and ‘own the narrative architecture’. A well-meaning founder once told me, gently, that my reluctance to spin was a skill issue. I found this clarifying, if not in the way he intended.

Nowhere is this more acute than in fundraising – which is, if I’m being honest, the part of this job I find hardest. Not because of the mechanics, but because it is the place where all of these values collide. Narrative inflation isn’t just ambient in a fundraising process; it is often expected, sometimes explicitly rewarded. A venture partner once wryly explained that the correct euphemism was ‘momentum-driven investing’. But I find myself constitutionally incapable of selling a purely narrative-driven raise – and a part of me quietly dies every time I watch someone else do it well. I’m not sure if this refusal is principled or just stubborn naiveté.

But I still notice what it costs. I experience life in this world as a kind of flattening.

In anthropology, what is rewarded is not the conclusion itself but the precision with which you articulate the diverse and often contradictory threads you are studying. In my dissertation, I was looking at contractors who both hurt and help migrant workers – people who are structurally advantaged and disadvantaged at the same time. You are also rewarded for reflexivity: the ability to consider how your encounters with knowledge, with the world, are mediated through particular worldviews, relationships, power, and biases. Contradiction is not a problem to be resolved. It is the material itself.

Here, contradiction is a liability. Nuance is a luxury. Acknowledging that things are complicated – that your product is good but not great, that you have traction but not enough, that the market is real but the timing is uncertain – feels almost dangerous. Better to pick a lane. Better to say you’re crushing it.

I think what I feel, underlying everything, is compassion. For the people spinning and for the people being spun to. For myself, trying to navigate a world where the rules are so different from the ones I was trained in. And honestly – I question this instinct constantly. Is this a principled take, or am I being sanctimonious? Is it stubbornness masquerading as principles? Does it hurt my company? There are contradictory impulses in me too – the type-A go-getter who wants to optimise relentlessly, who refreshes fundraising announcements and product launches with a kind of feral attention – and equally, the part of me that wants to slow down and ask whether optimisation is even the right frame.

What has helped, I think, is a more comfortable relationship with uncertainty – one that I owe partly to anthropology. The ability to say: here is what I don’t know. Here is what I do. Here are the caveats. Here is what I hope to know, and here is how I think we can test it. I think this serves an early-stage founder more than most people realise. So much of the spin I see around me comes from an inability to sit with not-knowing – the compulsion to narrate yourself out of ambiguity, to perform conviction when the honest answer is: I’m not sure yet.

Of course I still get anxious. It’s impossible not to when doing something this high-stakes, with people who’ve trusted you with their time and their careers. I have a stealth X account I go to for my carefully titrated dose of adrenaline – all these posts from founders crushing it, raising rounds, shipping product at inhuman speed. But I’ve also learned to read past the noise, to close the app and sit with my own reality. What is actually going on. What are the customers saying. What are the blockers. How should we address them.

A few months ago, we signed with a customer – a mid-sized HVAC company. They were excited about the product, ran internal trials, tested end-to-end workflows. And then they told us, honestly, why they couldn’t use it yet. Not in a vague, non-committal way. Over WhatsApp, one of their team sent us a list of eleven points – a mini essay, really – detailing exactly what worked, what didn’t, what they needed before they could adopt it properly. I saw this as an act of love. Despite everything – the gaps in the product, the things that weren’t ready yet – she was happy to bat for us during due diligence with a fund. She wanted to do what she could, she explained, for what she felt was a useful product from a sincere team.

Clear, actionable feedback. Specific features that brought delight, and specific frictions that didn’t. Our job, as I see it, is to fix the delight-to-irritation ratio. But what moved me wasn’t just the utility of the feedback – it was the extraordinary level of detail, the care behind it. It’s a strange place to start from, being moved. But I think that’s the fundamental quality I want to protect – a kind of care that runs in both directions.

My favourite parts of this work are when I get to learn more about the customers themselves. Not just their workflows and processes and hierarchies – though yes, those too – but their individual aspirations. Why they joined this industry. What stresses them. What they’re proud of. These conversations remind me of fieldwork – the good kind, where you arrive with questions and leave with better ones. And they remind me that the product we’re building isn’t abstract. It lives in someone’s worksite, on someone’s phone, in the middle of someone’s day. The delight has to be real. The usefulness has to be real. You can’t spin your way to a product that someone reaches for at 7am on a construction site.

I think there is space for sincerity in this world, that it is not simply naiveté. That you can negotiate hard and still be kind. That you can see reality clearly – all of it, including the parts that are messy and uncertain and contradictory – and still choose to build. I think joy is so underrated. I also think I might be wrong about some of this, or that the world might not always reward it. But the HVAC team didn’t send me eleven points over WhatsApp because I had the best narrative. They did it because they believed the product was useful and the team was sincere. I’d like to keep earning that.