Why I stopped paying Splitwise $36/year and built 'Float' in a day
May 25, 2026 • Esteban Selaya
Splitwise has 30 million users, charges $36 a year for Pro, and in 2023 started showing ads to people who refused to pay.
What changed at Splitwise
For years Splitwise was the textbook example of a boring SaaS that people quietly loved. Friends shared rent, groceries, trip costs, and Splitwise tracked who owed whom. Then the team needed revenue. In 2023 they capped free users to a small number of expense entries per day, dropped banner ads into the free tier, and pushed harder on Splitwise Pro at $36 a year.
The reaction went the way these things always go: long-time users felt ambushed, a few dozen switched to spreadsheets, and a handful built their own replacement. The math is hard to argue with from the company side. Splitwise has to charge somebody to keep the lights on, and ads are the cleanest alternative when most users won't pay. But the math from the user side has also changed. Why spend $36 a year forever for a feature you could replicate in under a day?
The hidden tax of every SaaS subscription
Every recurring subscription is a tax on capability you don't have yet.
The capability used to be expensive: a quarter of engineering work to stand up auth, a database, a deploy pipeline. AI is flipping that math. For any tool simple enough to describe in a paragraph, what used to be a quarter now looks more like a side quest.
My partner and I had been using Splitwise for shared expenses for years when the ads started showing up. I already maintain base infrastructure for my client projects, so I could reuse it instead of starting from scratch. The only thing missing was a tiny form to record "X paid Y for Z" and a balance view, plus some nice-to-haves for how the two of us actually use it.
I built it in a day and called it Float. The whole thing piggybacks on infrastructure I already had for other reasons — the same access control, the same database, the same deploy pipeline.
Why it took a day, not a quarter
The reason it took a day and not a quarter is the same reason most build-vs-buy debates go the wrong direction. People estimate the build cost the way it would have looked in 2022: from scratch, with no help. Today the build looks different. AI handles most of the grunt work, and the marginal cost of standing up something small has collapsed.
Splitwise is doing a hard job: multi-platform mobile apps, OCR receipt parsing, currency conversion, settling balances across dozens of users. I was doing the easy version of that job for two people. Those are different products, and only one of them is worth $36 a year to me.
Pay for what you actually need
My one-day timeline is the unfair version. The infrastructure helped, but I also got there fast because I only needed a sliver of what Splitwise actually does. How often are you overpaying for features you don't need or use?
This is a calculation I help my clients run every week, and AI keeps changing the answer. Small builds that used to take a quarter now take an afternoon, which means SaaS invoices that made sense in 2022 don't always make sense in 2026.
If you're staring at a stack of SaaS invoices and wondering which ones are still earning their keep, an AI Strategy Session is a 90-minute working session where we go through your stack one line at a time and decide what to consolidate, replace, or leave alone.