A note for lender CEOs about a story the industry has stopped telling honestly.
We ran a test. Not by choice.
Credit bureau data was not available at the point of decision — a forced situation, not a designed one. Open banking was all we had to qualify applicants with: transaction data flowing into the lender's affordability rules. Salary in. Bills paid. Surplus to cover the repayment. No bounced direct debits. The affordability check said: lend.
A few hundred customers cleared that gate.
When the full bureau picture came back into view, a different population emerged from inside the same applicant pool. CCJs. Defaults. IVAs running quietly in the background. Customers whose current accounts looked spotless precisely because they had stopped paying anyone back.
Had our clients stayed running on open banking and affordability alone, they would have lent into a loss curve nobody could see.
That is the post. Everything else is detail.
The story the industry has decided to tell
Pick up almost any recent open banking press release and you will read the same script. Nine in ten marginal declines overturned. Fifteen per cent more good borrowers identified than the bureau. Forty-four per cent of declined applicants reconsidered. Vendors are pairing with the big traditional scoring brands. Pilots are being expanded. Cash-flow scores are being launched into the centre of the underwriting stack. There are even vendors now claiming they can do vulnerability assessment from open banking alone — as if a transaction feed could tell you what a properly trained human agent would catch in a five-minute conversation. Same playbook. Same overreach.
The numbers are not lies. They are real outputs of real tests — but they are almost never run at scale, in live production, against the full population, with full bureau and performance data sitting alongside. Pilot conditions are not lending conditions.
And every one of those headline numbers measures the same thing — how many more people you can say yes to — and almost none of them measures the thing that actually matters to a CEO running a lending book: what happens to those people twelve, eighteen, twenty-four months later.
Approvals are an input. Losses are an output. The industry is selling you the input.
And here is the part the press releases consistently fail to make loud enough: open banking data is a complement and an enabler. It is not a silver bullet, and it was never designed to be one. The strongest open banking products in market are powerful precisely because they sit alongside other data — bureau, performance history, fraud signals — not because they replace them. That nuance gets lost in the headline. The headline is what ends up on the procurement deck.
What open banking, on its own, simply cannot see
Open banking is a window. It is a brilliant, real-time, consented, structured window. Used well, it tells you more about a customer's current financial behaviour than the bureau will ever show you.
But a window has edges. And the edges are not really about how far back you can look — depending on the institution and the consent you can stretch the historical view to as much as seven years on the accounts a customer has authorised. The edges are about what falls outside the frame entirely.
Inside the frame, you can see salary and spend. You can infer affordability. You can spot gambling, BNPL stacking, revolving overdrafts.
Outside the frame, you cannot see:
- A CCJ — a court record, not a transaction.
- A default on an unsecured loan from a lender who does not appear in the authorised feed.
- An IVA the customer has been quietly servicing through a different bank.
- The actual evidence of how this customer treats credit — performance on credit cards, personal loans, motor finance, mortgages — as distinct from how they treat their debit card.
- The customer who keeps an immaculate-looking current account by simply not repaying anyone.
The current account is the cleanest part of a financially distressed person's life. That is not a bug. That is how being financially distressed works. You protect the account the salary lands in. You let everything else burn.
A score that only sees the cleanest part of the picture is going to be wrong about exactly the people you most need it to be right about.
Affordability has the same problem, dressed differently
This is not just a credit-scoring story. The same logic applies — arguably more sharply — to the wave of open-banking-only affordability products now being sold into consumer credit, BNPL, motor finance, and mortgage broking.
An open banking affordability snapshot tells you a customer can afford the repayment. Income minus essential outgoings minus existing committed credit visible in the window, divided by whatever stress test you are running. The maths is fine. The window is the problem.
"Can afford" is not "will repay."
A customer with surplus cash flow today and a long history of defaulting on unsecured credit is, statistically, a customer who will default on this one too. Affordability data tells you the wallet is full. It does not tell you the customer has a habit of not opening it for lenders.
If you are making affordability decisions on open banking alone, you are answering one question — does the money exist? — and skipping the one that actually drives losses — does this customer pay people back?
The fix is not to bin open banking. It is to refuse the false choice.
None of this is an argument against open banking. Open banking is, on the whole, the most important step forward in lending data in a generation. Used properly, it expands the population you can responsibly lend to. It catches things the bureau misses — the gig-economy income, the thin-file young adult, the recently arrived professional, the genuinely-improving rebuilder.
The mistake is treating it as a replacement rather than a layer.
A defensible widening of the funnel — the kind that survives an economic downturn, an FCA review, and a board-level loss-curve conversation — uses open banking alongside credit bureau data, past lending performance, fraud signals and alternative data sources. Each layer covers another layer's blind spots. Open banking sees what the bureau cannot. The bureau sees what open banking cannot. Past lending performance sees what neither sees: how this specific customer behaves when they actually owe somebody money.
Build the score on one of those alone and you have built a confident-looking model with a hole in it.
Build it on all of them, and you have something that genuinely widens the funnel without quietly widening losses.
This is the problem Credit Canary was built to solve.
We did not start Credit Canary because we thought open banking was wrong. We started it because we watched lenders being forced into a false choice — bureau or open banking, traditional or modern, narrow funnel or loose funnel — when the answer was always and.
Credit Canary brings the disparate data sets into a single decisioning layer. Credit bureau data. Open banking. Past lending performance across multiple lenders. Fraud signals. Alternative data. The sources that, on their own, each tell a partial story — and, combined, tell the truth.
That is what genuinely widens the funnel without exposing the book to risk that nobody saw coming. It is not a louder press release. It is a quieter loss curve.
The metric that should be making CEOs nervous
If a vendor is showing you a chart of how many more applicants their model would have approved, ask one question:
What did the loss curve on those approvals look like, twelve months later, in a real production book?
If they have that data, take it seriously. If they have only the approval-uplift number — the headline that sells the press release — you are being shown an input dressed up as an outcome.
The lenders who win the next cycle will not be the ones who said yes to the most people at the top of the funnel. They will be the ones whose funnel widened and whose losses did not. That is a much harder problem than open banking alone can solve. It is the problem the industry has, conveniently, stopped talking about.
It is the problem we get up every morning to solve.
Open banking is brilliant — but not in isolation.
In isolation, an open-banking-only credit score, or an open-banking-only affordability decision, is a confident answer to half a question. Combined with bureau, performance history and the rest of the picture, it is one of the most powerful tools a lender has.
The difference between those two sentences is the difference between a wider funnel and a wider loss curve.
Hundreds of customers in our test could have told you the same.
← Back to News & Opinion