A free personal finance app that connects to all your bank accounts, tracks all your spending, monitors all your investments, and runs analytics on the whole picture isn’t free. It’s priced in something other than money. Probably your data, possibly targeted advertising, sometimes a referral kickback every time you click a credit card recommendation. The bills get paid one way or another.
I’m going to walk through what those bills actually look like. Not to scare you off any specific app, but because the economics are genuinely not obvious from the outside, and they determine how the product treats you.
What it actually costs to run this kind of app
The biggest line item, by far, is bank account aggregation. The company that almost everyone uses for this is Plaid. Plaid’s own pricing page publishes the structure but not the rates: three plan tiers (Pay as You Go, Growth, Custom) and three billing models depending on the product. Some products charge a one-time fee per connected account. Some charge a monthly subscription per connected account. Some charge per successful API call. Auth, Transactions, Investments, Liabilities, Identity, and the rest are each priced separately.
Specific rates are not public. They’re negotiated per customer and depend on volume. The shape is the part that matters for understanding the business model: every connected account a user adds is a recurring line item, every month, for as long as that user keeps the connection open. Add another Plaid product (say, investment holdings on top of plain transactions) and that’s another line item per account.
That’s before:
- Market data. Real-time or near-real-time stock, ETF, and crypto pricing is licensed. Public retail providers like Polygon.io and IEX Cloud publish their tiers, and the institutional-quality feeds (Refinitiv, Bloomberg, NASDAQ) start at five or six figures per year for any meaningful coverage.
- Cloud infrastructure. Compute, storage, bandwidth, encrypted backups. For a finance app with real-time sync and historical tick data, the infrastructure bill is real and grows roughly linearly with active users.
- Engineering, security, compliance. SOC 2 audits, penetration tests, security operations, the actual engineers. None of this is cheap. It’s also not optional when you’re holding bank credentials.
Add it up and the per-user monthly cost of running a serious personal finance app is real money, before any margin and before anyone gets paid a salary. The exact number depends on contracts, scale, and product mix, but it’s never zero, and it grows every time a user adds another account or another data product.
Bank aggregation, market data, infrastructure, and security operations all carry real per-user costs. The sticker is hidden from the user, but somebody is paying it every month.
If users don’t pay, who does?
There are basically three answers.
1. Advertisers
Many free apps include affiliate offers for credit cards, savings accounts, brokerages, and loans. The app gets paid when you sign up. The kickback per credit card signup can run $100-300. This is legal, often disclosed, and the obvious problem is that the app has a financial incentive to recommend products that pay it the most, not products that fit you the best.
The less obvious problem: even when the recommendation is ostensibly “based on your spending,” the spending data itself is the asset. The app sees what you charge, what you carry a balance on, where you bank, what you invest in. That’s a targeting profile, and it’s extraordinarily valuable to financial product issuers.
2. Data buyers
This is the one most people don’t think about. Aggregated financial data, even “anonymized,” is sold to:
- Hedge funds, who use spending trends to predict company earnings (real example: aggregated debit-and-credit data has been a documented input to consumer-stock trading desks for over a decade)
- Credit card issuers, who use cohort spending patterns to benchmark their portfolios and design new products
- Retailers, who use category spending trends to plan assortments
- Researchers and consultants, who package the data into industry reports
The companies that operate the aggregation layer (the ones sitting between consumer apps and banks) have historically been the biggest sellers. Yodlee, owned by Envestnet, has settled regulatory questions about exactly this practice. Plaid’s terms have evolved over the years partly in response to similar scrutiny.
“Anonymized” is a word doing a lot of work in those sentences. The academic literature on re-identification is consistent: financial transaction data is hard to truly anonymize. Three or four high-confidence transactions can usually pinpoint a single household.
3. The free app is a loss leader for something else
Some apps are free because the parent company makes money elsewhere. The free app exists to feed deposits or assets under management to a brokerage, a robo-advisor, or a banking arm. The app itself doesn’t need to be profitable. Your relationship with the parent company is what’s being monetized.
This is the most benign version. It’s also the version most likely to abruptly change product direction, because the free app is always a strategic instrument, not a product the company is committed to in its own right.
What “your data is never sold” usually means
Read the fine print. Most privacy policies for free finance apps contain some version of:
- “We do not sell your personal data.” (But aggregated data is fine.)
- “We may share data with our service providers.” (Without naming who, or limiting use.)
- “We may use your data to improve our services.” (Including training models that benefit other customers.)
- “We may share data with affiliates.” (Often the parent company and its many subsidiaries.)
None of these are necessarily nefarious. They are also not “your data is never used for anything you didn’t directly ask for.” The economic model demands data flow outward in some form. The legal language reflects the model.
The Thermal answer
Thermal charges $14.99 a month for Standard and $29.99 a month for Pro, with a 14-day free trial. Those numbers exist because the underlying costs exist. Plaid bills are real. Market data bills are real. Infrastructure and engineering are real. We pass them through, add a margin so the company can survive, and that is the whole business model.
We don’t sell user data. Not aggregated, not anonymized, not under any creative repackaging. We don’t serve affiliate offers inside the product. We don’t cross-sell to a brokerage arm because there isn’t one. The product makes money in the most boring way available: the people who use it pay for it.
The fourteen-day trial is real, and there’s no friction on cancellation. That’s also part of the same idea. If the product isn’t worth what it costs you, the answer is to not pay for it, not to find some other way to extract value out of you.