I had a conversation a while back that I keep coming back to.
A client rang me about a piece of software we’d built for them. Nothing was broken. The thing was working perfectly, doing exactly what the brief said it should do. But she was unhappy, and it took us most of the call to work out why.
The system had automated a step that used to involve her ringing a supplier once a week. The automation was correct. It was faster, cheaper and less error-prone in every way you could put on a slide. And it had quietly deleted a fifteen-minute phone call that turned out to be the only reason that supplier ever gave her a heads up when stock was getting tight.
We’d optimised a process and accidentally cut a relationship. Nobody had asked us not to, because nobody knew it was there. It wasn’t in the brief. It wasn’t in anybody’s process document. It was just two people who liked each other, talking on a Thursday.
That call rearranged how I think about this job.
Everything we make is a connection between people
Paul and I talk about interfaces quite a bit, and we came to the conclusion that there’s only people, reaching out to each other through what we make.
A form is somebody trying to talk to somebody else. An invoice is one person asking another for money, which is about as loaded as human interaction gets. A booking page is a stranger deciding whether to trust you with a Saturday. A dashboard is somebody’s boss asking a question in advance, so they don’t have to ask it again on Monday.
Every screen anyone has ever built is a message from a person who isn’t there any more to a person who hasn’t arrived yet. We write these things months before they’re read, to people we will almost certainly never meet, and then we’re not in the room when they land. It’s the most indirect way of talking to somebody ever invented, and we do it constantly.
Which means the craft isn’t really in the code. The code is how you get the message to travel. The craft is in knowing what the message actually is.
Business is a thing that only exists between people
We often talk about businesses as if they were objects. You have one, you build one, you sell one. But you can’t point at a business. Turn off the relationships and there’s nothing left - no revenue, no team, no suppliers, no reason for the lights to be on.
A business is a pattern of connections that has become stable enough to have a name and a bank account.
Everything you’d call an asset is really someone’s willingness. Cash flow is a hundred people who chose you over the alternative. A brand is what people say when you’re not in the room. Credit is somebody deciding you’re good for it. Even your ability to have a bad month is just the accumulated goodwill of people who’ll wait.
None of that is on the balance sheet, and all of it is the actual business.
I say this to people who are starting out and I can see it not landing, because it sounds like a nice sentiment rather than a hard fact. So here’s the practical version: your ability to connect with other people is not a soft skill that sits alongside the real work. It is the compounding asset. It’s the only thing you own that gets more valuable while you sleep, and the only thing that can’t be copied by someone with a bigger budget.
Where I part ways with human-centred design, slightly
Human-centred design was one of the genuinely good ideas to come out of our industry. Start with the person, not the technology. Watch what they actually do rather than what they say they do. Design for the human being, not for the spec.
I agree with all of it. I’ve spent my career doing it. But there’s a shape to it that I think quietly misleads us, and it’s right there in the diagram - one human, in the centre, alone, being served.
Because in my experience there is almost never one human, and they are almost never alone.
The person filling in your form is going to forward the result to someone. The person on your pricing page is going to screenshot it and send it to their business partner with three words of commentary that you will never see, and those three words will decide it. The person clicking “confirm” has already discussed it with their spouse. The person choosing your software will have to defend that choice to a colleague who preferred the other one.
Watch what really happens when someone uses a piece of software and you’ll notice how much of it is social. They ask someone. They show someone. They forward it. They ask “is this normal?” They wait until they can check with the person who knows. Almost nobody reads the help page; they find the one person in the building who’s done it before and ask them, because a colleague who might be slightly wrong is worth more than a manual that’s definitely right.
So the unit we design for isn’t the user. It’s the relationship the user is in.
That’s the difference, and it’s small enough to sound like word games until you try designing for it, at which point it changes nearly everything. We call it human connection design, and it’s what we practise here.
What that actually changes
You ask who else is in the room
Not just “who’s the user”, but who they’ll have to explain this to. Who signs it off. Who gets blamed if it goes wrong. Who they’ll ask before they commit. That person never logs in, and they’re often the one who decides.
Awkwardness is a real cost, and you can put a number on it
This is the one I’d most like people to take away.
Every time your system forces two people to have a conversation neither of them wanted, it charges the relationship a small fee. “Sorry to chase this.” “Could you resend that?” “I think there’s been a mistake with the invoice.” “Are you sure you sent it?”
None of those are bugs. Every one of them is a tiny withdrawal from an account you’re trying to build up over years. Software that removes an awkward conversation is worth far more than software that adds a feature, and it’s almost never valued that way because you can’t demo an absence.
Most of what we’ve built for ourselves came from exactly this. Not from a market gap - from watching a relationship get slightly worse for no good reason and thinking, that’s fixable.
You design for the forward
Assume everything gets sent to someone else, out of context, on a phone, with no explanation. Make the email make sense on its own. Make the page survive being screenshotted. Make the number defensible by someone who wasn’t in the meeting where it was agreed.
You are careful about what you automate away
Efficiency is a wonderful servant and a terrible master. Before you delete a step, find out whether anything human is living in it. Sometimes the slow bit is the value - the phone call where the real information gets exchanged, the check-in that keeps two people warm to each other. Automate the part that’s genuinely a chore, and be very slow to automate the part where people talk.
You accept you’re not in the room
You will never be there when your work is used. You don’t get to explain, apologise or clarify. Whatever you built has to carry your intent on its own, to a tired person at 4:50pm on a Friday who has never met you and has no reason to be generous.
That’s a humbling way to think about a login screen, and I think it’s the correct one.
Why any of this matters to us
We’re a small shop, deliberately. Two of us. We don’t win work by being the biggest name on the list, and we’ve never wanted to.
What we have instead is a set of relationships that have lasted a very long time, some of them well over a decade. Clients who’ve moved companies and brought us with them. People who ring us before they’ve decided what they want, which is the highest compliment in this trade. That’s not a growth strategy - it’s just what happens if you spend twenty years treating the connection as the point and the software as the medium.
And it works in the other direction too. The best briefs we’ve ever had came from people who trusted us enough to tell us the messy truth about their business early. You cannot buy that with process. It comes from being someone worth telling the truth to.
So yes - we build software, and we’re good at it. But what we’re actually doing is helping people reach each other with less friction and less awkwardness than they had before. Everything else is implementation detail.
If any of this sounds like the thing you’ve been trying to put words to, come and have a chat. We’d love to hear about it - and honestly, the conversation is the good bit.