There is no perfect accounting software.

There is no perfect accounting software. I know that isn't a controversial thing to say to anyone who has actually used it for a living, but it's worth sitting with for a moment, because we've quietly accepted something rather strange: that choosing accounting software isn't really a decision, it's a negotiation over which pain you're most willing to live with.

Think about how that choice actually goes. You find a tool with genuinely beautiful reporting, the kind that makes a year-end conversation feel like insight rather than admin, and then you discover the day-to-day bookkeeping fights you at every turn. So you look elsewhere, and you find something with invoice extraction built right in, which feels like a small miracle until the first time you try to book a payment on account and realise the whole thing falls apart in your hands ( I see you Quickbooks ;) ). Fine, you think, I'll skip the built-in extraction, I'll buy a separate tool for that, and I'll choose the software that handles payments on account properly instead. And it does. And then a marketplace settlement lands with a balance offset on it, or a partner nets off their fees before paying you, and you are back at your desk at half past six building a workaround for something that should have taken thirty seconds (sounds a lot like Sage?).

Nobody chose this. We all just arrived here, one reasonable compromise at a time, and ended up with a tech stack held together by the digital sticky tape of CSV imports.

The obvious question is why. Why, after all these years, has no one simply built the accounting software that has all of it? And for a long time I assumed the answer was that it couldn't be done, that accounting was just too sprawling to ever sit inside one tool that did every part of it well. But the more I've looked, the less I believe that. The pieces are all solvable. What's actually happening is that the people building these tools are choosing, again and again, to spend their effort somewhere other than the place accountants feel the pain.

Start with who the software is really for. Most accounting tools are sold to business owners, and business owners are a genuinely different customer from the accountant who has to live in the thing all year. An owner wants to raise an invoice, glance at their cash, and get on with running their business, and they neither know nor particularly care where their bookkeeper's time is quietly leaking away. So that's what gets built beautifully: the invoice, the dashboard, the bank feed, the morning glance. Meanwhile the recurring journal that has to be re-keyed every month, the payment on account that won't sit against the right invoice, the marketplace offset that takes twenty minutes to unpick, these are the accountant's problems, not the owner's, and the accountant isn't the one the product was designed to delight. The pain is real, it's just landing on someone the developer wasn't building for.

And then there's the simpler, slightly more cynical truth about what gets prioritised, which is that some features are easy to market and some are not. You can put a beautiful invoice on a billboard. You can make a thirty-second ad about snapping a photo of a receipt and watching it appear in your books. Nobody is ever going to make a radio advert about recurring journal entries, or about payments on account booking correctly, or about a bulk-edit that lets you fix three hundred miscoded transactions without losing your afternoon. These are the things that make an accountant's working life materially better and they are almost impossible to say out loud in a slogan, so they lose, over and over, to whatever demos well on a stage. The roadmap follows the marketing, and the marketing follows whatever photographs nicely.

So it isn't that the perfect tool is impossible to build. It's that building it would mean pouring real effort into unglamorous, hard-to-advertise features for a user who isn't the one signing up, and almost nobody has decided that's worth doing.

I keep coming back to the same slightly absurd image. We are watching Elon Musk catch returning rockets out of the sky with what amounts to giant chopsticks, meanwhile an accountant somewhere is exporting a CSV, cleaning it by hand, and importing it into a second system because two pieces of software that both do accounting cannot bring themselves to speak to each other. It's tempting to read that gap as proof that accounting is simply too messy to ever get right, but I don't think that's it at all. Catching a rocket is also unglamorous and fiddly and full of edge cases that don't hold still. The difference is that someone decided the unglamorous part was worth their full attention, and poured years of obsession into the tiniest bits nobody would ever put on a poster.

That's the whole of it, really. We are still cleaning CSVs by hand not because the problem is unsolvable, but because the problems that would end the hand-cleaning are the quiet ones, the ones that serve the accountant rather than the buyer and don't fit on a billboard, and so far almost no one has decided they're worth the time it would take to get them right. The recurring journal that just works, the payment on account that lands on the account the right way, the offset payment, that can be posted directly from the bank feed, the two systems that finally speak to each other. None of it is beyond us. It's just been quietly deprioritised, over and over, for years, in favour of whatever demos better.

So no, there is no perfect accounting software, and now I understand why, I find I can't quite make my peace with it. Not because perfection is impossible, but because it's been deemed not worth the effort, and the cost of that decision doesn't land on the people making it. It lands on us, in our own evenings, one hand-cleaned CSV at a time.