Sensibilities

A shopkeeper in a small town takes a UPI payment. He doesn't look at his phone. A little box on the counter says, out loud, in his language, that fifty rupees came in. He nods and turns to the next customer.

Nothing about that box is the function. The payment already worked. The box exists so he never has to wonder. That's a sensibility: the part of a product that isn't the job, and tells you someone thought about your day.

It cuts both ways. The shopkeeper wants it loud. The person paying might not. Now the whole queue knows what you spent. Walk into a five-star hotel or a big showroom: no box. Nobody announces what you paid. They've made the other call, on purpose. Same sound, two different answers. Sensibilities aren't a coat of polish. They're decisions about people, and every one has a cost.

The writer Scott Berkun has a plain line for this: good depends on context. Nothing is good or bad on its own. The box is right for the shopkeeper and wrong for the person standing in line. Before you decide how a product should feel, ask who is standing there when it speaks.

Every product carries three of them, whether anyone designed them or not.

• 01How it looks. Not decoration. Calm or shouty. Type set for someone reading in sunlight on a bus. People decide in a second whether a thing was made with care. They don't say it. They just pick the one that was.
• 02How it sounds. Car makers tune the sound of a door closing. A thin click makes the whole car feel cheap. Electric cars got so quiet that regulators made them add a sound at low speed, for the people who couldn't hear them coming. A payment app's confirmation tone does the same job as the box on the counter. Silence is a decision too. Usually the wrong one.
• 03How it approaches you. The first screen. Whether it asks for your number before it has done anything for you. Whether it says "you don't know your birth time, that's fine" or throws an error. The tone of a product is a sensibility, and most products let it happen by accident.

A closer look at how it looks.

My son was four. We drove past a hoarding with no name and no words on it, just one big field of a very specific orange. He said the brand's name out loud. He couldn't read yet. The colour was enough. A brand isn't a logo. It's a habit the eye learns, and colour is the fastest way in, because nobody guards that door. Hold the same colour, in the same place, for long enough, and it stops being information and starts being the surroundings. The goal isn't to be seen. It's to be remembered.

Open an AI design tool and ask for a homepage. You get white space, soft greys and thin buttons. It looks made for an apartment in San Francisco. For a lot of India it's the wrong picture. Think of a pan shop, a train departure board, a wedding invitation. Every inch carries something. Nothing is wasted. On an entry-level phone with many apps fighting for one small screen, empty space isn't calm. It's lost signal. And warmth does a job that grey can't, because colour tells people they can trust the place. The tool's default is somebody else's taste. Use it without thinking and you've still made a choice.

We once had to pick an icon for direct messages. Three options: a bell, a send arrow, a double speech bubble. We showed them to a watchman in Hyderabad, a ghee seller on the street, and two Urban Company workers. The bell read as notifications to most of them, and as messages to one. The arrow was the problem. One couldn't tell what it was. One saw "send." Two saw "share." Same arrow, three meanings. The speech bubble read as messages, or WhatsApp, or comments, to every one of them. We shipped the bubble. Not because it was prettier. Because it meant one thing to everyone who'd use it.

Delight isn't a fourth thing you add. It's what people feel when the care is visible and they didn't have to look for it.

Chase delight directly and you get confetti. Get the sensibilities right and you get the shopkeeper's nod.

Before you add a feature, listen to the product with your eyes closed. If it's silent where it should speak, or loud where it should be quiet, fix that first. It's cheaper than the feature, and the user notices it more.