Product Before Profit.

The best of tech, apps, sites and services have proven, like an untended garden, to run to weeds. Which is to say, to get worse. Not everything, and not all at once, but the garden only seems to grow one way, and you feel it most in the products you touch every day — the ones holding your work, your photographs, your money, your conversations.

There's a word for it, coined by the writer Cory Doctorow, and it's built on a fecal metaphor we can't repeat in front of clients. We mangle it every time we try to say it out loud anyway. Enshittification. It comes out of our mouth as entiatification, which is arguably worse.

Nobody sets out to make a bad thing

We can only think of one explanation, and it isn't cynicism. When someone first decides to make a product — a person, a company, a group of people around a table — you'd hope they began with the best intentions. That they wanted to make something great.

And what is a great product? Something that works well for the people using it and solves a problem they actually have. Something that is a pleasure to use, which is not a soft requirement — pleasure is most of the reason anyone returns to a thing without being nudged into it. Something that takes Dieter Rams at his word and uses as little design as possible. Something excellent enough that it might even be loved, although of course you can't love a thing. Like very much, we could say.

So we'll assume the fine ideals were real, and that nobody in that first room was thinking about profit before product.

Then it works

Eventually the thing gets popular. And somewhere in there — usually alongside the growth rather than after it — profit arrives ahead of product. We can think of very few examples where that hasn't happened, and fewer still where it hasn't damaged the very thing that was made well in the first place.

We don't want to name names. This is a thought note on profit and product, not a roll call of the fallen. You know of what we speak.

We're not proposing a solution, either. It's more of a question. Why can't things stay product before profit?

A while back we wrote about productivity without purpose — work done faster and landing worse, and the care that goes missing along the way. This is its bookend. That piece was about how things get said. This one is about what actually gets made.

Why can't things still be good?

In our own small world, they can. We go to bed knowing the product is better than it was that morning. Better is subjective, but not very: it does a better job for the customer, it's simpler, it's clearer, it's faster. What it hasn't done is fall prey to the entropic principle — complication for its own sake, features fudged in to juice a little more out of the customer.

We make four things of our own, and they're the only evidence we can offer that this is more than talk. Not the products themselves, which anyone can describe kindly. The decisions inside them that cost us money.

We left the floor free, and nothing hard-stops. Fly Away Paul, our research and outreach tool, has plans if you want one. It also has a pay-as-you-go floor you never have to leave — you pay for the work Paul actually does, and nothing else. That shape has a consequence we rather like: if we're only paid when the tool does something for you, there's no version of us that profits from it doing more than you asked for. Where there are plans, what's included was measured against what people genuinely use rather than the most we could extract, and running past an allowance doesn't stop you working. It reverts, quietly, to paying by the action. A wall you hit at eleven at night, mid-job, is a product decision as much as a pricing one.

We took the data out. Paul counts page views in about 1.5KB and cannot tell who you are. Reviewing our own installation on this site, we found that client links carrying private tokens could land in the analytics log; the fix went into the product rather than into a policy page, so the tokens are now stripped in the browser before anything is sent. Less data makes it a weaker product commercially. It makes it a better one honestly.

We left the AI out where maths does it better. In Vidual Inbox, a customer mentioning a date gets it surfaced back to you by plain deterministic logic — no model, because maths is right every time, costs nothing and doesn't need supervising. The helper that does use a model, Otto, never sends anything without you.

We made it quieter. Vidual Projects tells you what happened without the frenzy, because the point of a tool for people who make things by hand is to keep them in the flow of making. Spaces answers a plain sentence about your files instead of handing you another dashboard.

Not one of those decisions grew a number. Two of them cost us money outright. One of them — the maths instead of the model — happens to save us money, which makes it the easiest sort of decision to get right, and we'd like to think we'd have made it anyway.

While we're on the subject of shoehorns

The other place you can watch profit arrive ahead of product is the rush to put AI into everything.

We should be careful here, because the argument is not that AI is wrong. It's great, in a great many ways — extraordinary at finding exoplanets and reading scans, quietly useful at a hundred duller things. We use it every day inside our own business, for research and drafting and the machinery nobody sees. What we don't do is put it in front of a customer, because if the answer is AI, why would you pay us?

The fault isn't the tool. A bad workperson blames their tools, and the adage holds even now that the tools can talk back — what goes wrong is any tool applied unthinkingly, dropped into a product where nobody stopped to ask whether it was necessary, and where quite often it wasn't. This one simply happens to be the tool currently within everybody's reach.

Does every business application need it bolted on? Does anyone need another note-taker summarising a meeting they might have attended properly? Some of that is people getting overexcited, which is forgivable and occasionally produces something lovely. Some of it is a feature added because the market rewards the announcement rather than the use. We suspect the second is doing more of the work.

The question is not a hard one to ask. It's the same one behind the deterministic maths in Inbox: was this necessary? It only feels awkward when you suspect the honest answer might cost you the feature.

A thing that hurts isn't a product

While we're here: a great product should never be a hurtful one. Bombs kill. Bullets kill. No euphemism in the brochure changes what the thing does when it arrives, and a thing built for war or for harm isn't a product at all — it isn't making anything better for anyone, and everyone who works on it is cloaked in the same shadow.

That isn't a market position. It's what we think, and it's the one line here we'd hold even if holding it cost us the work.

Nobody has offered us the money yet

Nobody has ever offered us a large sum of money to make any of our products worse. We have no investor asking where the growth went. It is very easy to hold a line that nothing is pushing against, and we're aware that most of the people who ended up on the wrong side of this once wrote something much like what you've just read.

You'd hope that if you were asked, you'd hold it anyway. The world is full of people who found out that they wouldn't.

So none of this makes us better than anyone. Smaller, mostly, and slower. But the four things we're responsible for are better this week than last, we can tell you what every decision inside them was for, and we sleep well.

All we know is that we'd never make a product worse, come what may.

Less data makes it a weaker product commercially. It makes it a better one honestly.