Small banks approve more small business credit applications than any other lender type. According to the Federal Reserve’s 2026 Report on Employer Firms, applicants who went to small banks were the most likely to be fully approved, at 57%, and they were far less likely to be surprised by what the money cost: 37% of small-bank borrowers reported higher-than-expected borrowing costs, against 60% of those who borrowed from online lenders.
Better odds. Better terms. And a shrinking share of the applications. Over the same period, the share of applicants seeking financing from online lenders climbed from 17% in the 2020 survey to 29% in 2025.
That gap is not a pricing problem, and it’s not a product problem. It’s an information problem, and it sits on the bank’s side of the relationship. The financial institution that is best positioned to say yes is frequently the last one to learn the question was asked, because it cannot see the conditions that produce a credit need until the business shows up with an application, if it shows up at all.
Customer insights platforms are the category built to close that specific distance: between what is happening inside a small business and what the bank knows about it in time to act. This article covers what a customer insights platform in banking actually is, how it differs from the analytics tooling most institutions already own, the layers it is built from, the teams that consume it, the operating discipline that turns signals into engagement, and how to measure whether any of it is working.
A customer insights platform is a system that combines an institution’s internal account data with consented external data from the customer’s other financial and business applications, resolves it all to a single business entity, and converts it into signals, scores, and recommended actions that frontline, credit, product, and marketing teams can act on inside their existing workflows.
The category is often confused with tools banks already have. The distinction is worth being precise about, because the overlap is real and the gaps are the whole point.
|
System |
What it is built to do |
What it does not do |
|
CRM |
Record interactions, manage pipeline, hold relationship history |
Generate the signal that should have started the interaction |
|
BI / data warehouse |
Answer a question an analyst already thought to ask |
Push anything to anyone without a human initiating the query |
|
Marketing automation |
Deliver campaigns to defined lists on a schedule |
Know which business is in-market this week rather than this quarter |
|
Core reporting |
Report accurately on the accounts the institution holds |
See the financial activity happening outside the institution |
A customer insights platform is not a replacement for any of those. It is the layer that makes them useful for a segment the bank can only partially see. Three characteristics define it: outside-in data (it reaches beyond the institution’s own ledger), entity-level resolution (the unit of analysis is a business, not an account), and an actionable output (the end product is a prompt with an owner attached, not a chart).
Most institutions are not short of data. They are short of analytics that work on the SMB segment. Four structural limits explain why.
Core and digital banking data describes money that has already moved through the institution. It does not describe what is committed and still to come, and it does not describe the accounts, processors, and platforms that the business uses elsewhere. Banks are interpreting an SMB’s financial position from a partial ledger, which is a different problem from the cashflow assembly gap — that one is about forecasting; this is about coverage.
A single small business routinely presents to a bank as an operating account, a savings account, a card, a loan, a merchant relationship, and the owner’s personal checking — sometimes under a legal name, a DBA, and a trading name. Without entity resolution, the institution counts one customer as six or six as one. Every downstream calculation — profitability, share of wallet, attrition risk — inherits that error.
Month-end batch reporting is built to explain a period that has closed. Detection is built to flag a change while it is still a change. An SMB’s working capital position can turn inside two weeks; a report that lands on the 8th of the following month is describing a decision the business has already made.
This is where the cost becomes visible in customer experience. In its 2022 U.S. Small Business Banking Satisfaction Study, J.D. Power found that 58% of small business banking customers were receiving purely transactional advice, 27% constructive advice, and only 15% comprehensive advice — and that customers receiving comprehensive advice rated their bank 195 points higher on a 1,000-point scale than those receiving transactional advice.
The constraint is not appetite. In the 2025 study, 61% of small business owners reported receiving financial advice from their bank, and 94% of those said it positively influenced their business’s financial habits. Advice works; it is welcomed, and it is under-distributed — because a banker who cannot see what has changed has nothing specific to say.
Evaluating vendors is easier when the architecture is broken into its parts. Each layer has a distinct failure mode, and a platform that is strong at four of them and weak at one behaves like a platform that is weak everywhere.
The pipes that bring in data from the SMB’s accounting package, payment processors, other bank accounts, payroll, and commerce platforms — with the business’s explicit permission, and with a clear scope. If this layer is weak, then everything else is weak.
Matching accounts, entities, and owners into one business record. This is unglamorous, but it is where most internal builds stall, and it determines whether portfolio-level numbers can be trusted at all.
Raw feeds arrive in numerous ways. Normalization makes a payroll from one provider comparable to a payroll from another; enrichment turns an opaque merchant string into a categorized, attributed transaction.
Converting normalized data into defined, measured events: a sustained shift in collection timing, a change in revenue concentration, a balance moving to an outside institution, a new recurring outflow to a competitor’s product. The discipline here is restraint — a signal that fires on 40% of the portfolio is a description of the portfolio, not a signal.
This is the layer skipped most often. Insight that lands in a separate portal gets checked weekly at best. Insight that lands in the queue, the CRM record, or the call list that a team already works gets acted on. As covered in Who Owns the Small Business Banking Relationship?, the institution that notices a change first earns the next interaction — but only if the information reaches a person with the authority to respond.
A customer insights platform is usually pitched to one team and paid for by one budget, which is why so many deployments underdeliver. A single signal typically has four or five legitimate consumers, and the return compounds when they share a definition.
Take one example: a business whose receivables are taking materially longer to clear while its card settlement volume climbs.
The governance point matters as much as the capability. If commercial, credit, and marketing each maintain their own definition of "at risk," the institution will run three campaigns against three populations and attribute the same outcome three times. One signal library, versioned and owned, is a prerequisite rather than a refinement.
The difference between a financial institution that bought an insights platform and one that runs an insights program is whether the signals are wired into a repeatable loop. A workable play has five defined components, and skipping any of them is where programs quietly fail:
Two disciplines separate scalable programs from those that get switched off. First, start narrow: three to five well-instrumented plays are better than 20 launched at once because they can actually be evaluated. Second, define suppression rules — how often a business can be contacted, which signals take priority, and what happens when several fire in the same week. Suppression logic is often an afterthought, and it is usually why frontline teams stop trusting the feed.
SMB attrition is rarely a single event. It is a drift: engagement decays, then balances migrate, then products lapse, and the account closure that shows up in the churn report is the last step in a sequence that started months earlier. A customer insights platform earns its keep by making the earlier steps visible.
Every layer above depends on the data the SMB chooses to actively connect to the platform. Therefore, connection rate is the single variable with the largest effect on program value, and it is consistently the least planned-for line in the business case.
The good news is that the value exchange is real and reasonably well evidenced. In a Q1 2025 Datos Insights survey of more than 1,000 US small businesses commissioned by Apiture, 73% of millennial-run small businesses said they were willing to pay for banking products and services that help them manage their finances more effectively, with only 9% unwilling. These owners are already sharing operational data with software: the Federal Reserve found that 46% of small employer firms now use AI in the business, and 51% of those users apply it to planning or analysis. The bank is not asking for something unfamiliar. It is asking later than everyone else.
Three things move connection rates in practice:
Governance belongs in the design phase rather than in a legal review at the end. Scope, purpose limitation, revocation, retention, and auditability all shape what the platform can be built to do, and retrofitting them is considerably more expensive than specifying them.
Spending is not the bottleneck. Cornerstone Advisors’ survey of 416 bank and credit union executives found more than 80% planning to increase technology spending in 2026, while planned system deployments continue to fall short of what actually gets deployed. Jack Henry’s eighth annual Strategy Benchmark, based on 193 institution executives, found data analytics is the third most common planned technology investment at 32%, behind AI at 48% and digital banking at 38%, while growing deposits remains the top strategic priority at 64%. The intent and the budget are both there. Three patterns account for most of the shortfall.
Getting the operating model right is worth the effort. Jack Henry’s research notes that small business deposits are four to five times larger than retail deposits. That gives SMB engagement work a larger deposit impact than comparable retail efforts and ties the insights program directly to the top strategic priority executives identified: deposit growth.
9Spokes is a white-labeled data and insights platform built for financial institutions that serve SMBs, and it is deliberately built as both halves of the loop described above rather than as a standalone analytics product.
The Customer Insights Hub takes consented SMB data and shapes it for the teams that act on it: relationship and commercial teams, credit, risk, product, and marketing. It can be embedded in existing internal channels or run standalone, and it is where wallet-share visibility, cross-sell opportunity identification, early risk detection, and campaign targeting come from — all working from the same resolved entity and the same signal definitions.
The SMB Financial Hub is the customer-facing side, embedded inside the bank’s own digital banking experience under the bank’s brand. Businesses connect accounts across 800+ financial and business service providers and get a consolidated view of their financial life in return. This is what makes the insight layer viable: the value exchange that earns the connection is the same mechanism that supplies the banker-facing intelligence. One platform, two audiences, one data layer.
For how these capabilities fit into a full retention and share-of-wallet strategy, see SMB Financial Insights: How Banks Can Drive Growth, Retention, and Share of Wallet. For where this category sits within the wider stack, read SMB Financial Management Explained.
Financial institutions rarely lose SMB relationships because their prices or products fall short. More often, they lose them to latency: the gap between a shift in a business’s financial position and the bank’s ability to respond. A customer insights platform closes that gap, but it only creates value when connected to clear owners, defined plays, precise thresholds, and measurement that can stand up to scrutiny.
See what customer insights look like inside your own digital channels — book a 9Spokes walkthrough. Or get the full market picture in The State of SMB Financial Health & Banking 2026.
Questions? Reach the team at sales@9spokes.com.