176 neobanks won't tell you what they pay. 48 will.
Figures in this post describe a snapshot of 381 tracked neobanks. The dataset now tracks 385, so counts here will not match the live site — the analysis stands, but for current numbers use the browsable cuts, the live directory or data.json.
We record a yield field for every neobank in the directory: what the product pays you for holding a balance, in the company's own words, taken from its own site. It is one of the most requested fields we have, and one of the least fillable.
| What the product says about interest | Count |
|---|---|
| Quotes an actual rate you can read off the page | 48 |
| Mentions yield, savings or interest without a usable number | 157 |
| Says nothing about paying you anything | 176 |
| Total tracked | 381 |
Twelve and a half percent of the neobanks we track will state a number. The largest single group — 176 of them, nearly half — offers no interest story at all, which for most is simply honest: a payments app, a card, a self-custodial wallet. Nobody owes you a rate for money you hold yourself.
The interesting group is the middle one. 157 products talk about savings, yield, interest, pots, vaults or goals, and do not put a number where you can find it.
What "up to" is doing
Even among the 48 that quote something, read what the quote actually is. A representative sample, verbatim from the field:
| Neobank | What the page says |
|---|---|
| Varo | Up to 5% APY (conditions) |
| SoFi | Up to ~4.5% APY (tiers) |
| Current | ~4% savings pods |
| Step | ~4% savings goals |
| Chime | ~2% APY savings |
Three patterns do nearly all the work here, and once you can name them you stop being surprised by your statement.
"Up to" is a ceiling, not a rate. It is the number one customer in a particular configuration receives. The conditions attached — a qualifying direct deposit, a minimum balance, a spending threshold, a paid tier — are the actual product, and they live a click away from the number.
The rate applies to a container, not the account. "Savings pods", "goals", "vaults", "spaces": the headline rate pays on money you have deliberately moved into a sub-account. The balance sitting in your main account, where your salary lands, typically earns nothing. The number is real; it just isn't paid on most of your money.
Tiers move the number, quietly. Where a rate depends on a subscription level or a balance band, the figure on the marketing page is usually the top band, and your rate changes without any notification that feels like a rate change.
The comparison nobody runs
Here is the part that deserves more attention than the rate itself. A headline rate is a claim about return. It says nothing about what happens to your money on a bad day — and those two questions are answered by completely different parts of the product.
Of the 381 neobanks here, 132 hold their own banking licence, 82 run on a partner bank, and the rest operate under e-money, payment, crypto or self-custodial arrangements. Those structures carry genuinely different answers to "who is legally holding this, and what protects it if the company fails" — and the structure is not correlated with the rate. A high rate can sit on top of any of them.
So a rate comparison run on its own is a comparison of one variable while ignoring the one that determines whether the balance exists at all. If you are going to compare two products, compare the pair:
- What is the rate, and on which balance? The main account, or a pot you have to fund deliberately?
- What has to be true for you to get it? Direct deposit, minimum balance, subscription tier, spend threshold.
- Who is holding the money? The company itself under its own licence, a partner bank, or nobody — because it is self-custodial and there is no balance being held for you at all.
- What happens if the company stops operating? Deposit insurance and a regulator-run resolution, or a creditor queue.
Every profile in the directory answers the last two. Compare any two puts them side by side.
What would change this
Not regulation, probably. Rate disclosure rules already exist in most of these markets and the 157 are generally compliant — a rate that is accurate, conditional and buried is not a violation, it is a design choice about where to put the conditions.
What changes it is comparison being cheap. The reason a headline rate can carry conditions a click away is that almost nobody clicks. An open dataset does not fix that on its own, but it makes the gap measurable: we can now say that seven out of eight neobanks won't state a number, which is the kind of sentence that is harder to say about an industry nobody counted.