For lending and expense-management platforms using embedded finance, customers’ accounting data can be pivotal when it comes time to make business decisions. Accounting data can indicate whether a business is in a healthy financial position, and it’s often used to make decisions around extending credit, approving spending, and more.
Data Gaps
Unfortunately, though, the data a platform needs to make decisions with confidence isn’t always immediately available. For example, say a customer’s financial position seemed healthy on Monday. Then on Tuesday they upload late-recorded expenses, and there are still bank transactions pending in their accounting software. Without complete accounting data, embedded-finance platforms can’t gain the full financial picture they need to accurately assess a customer’s financial position.
Different Accounting Records
Complicating matters further, every business’s accounting records are going to look a little different. Consider two very different businesses: a small restaurant and a growing construction company. The restaurant most likely uses an accounting or bookkeeping solution designed with owner-operators in mind, while the construction company most likely uses a more complex financial system with multiple entities, employees with different approval permissions, and projects with codes. While many pieces of data will overlap, the format and level of detail can vary widely.
The Month-End Close
This provides a concise explanation of APIs at https://www.coursera.org/articles/what-is-an-api, but how can embedded-finance platforms use accounting APIs to access standardized, real-time financial data regardless of a customer’s accounting solution?
Initial Connections
Once a customer links their accounting data, the first order of business is getting access to as much relevant information as possible. This could include:
- Chart of accounts and account types
- Open invoices and bills
- Bank and credit-card transactions
- Current balances
- Historical profit-and-loss activity
- Vendor, customer, and project details (when available)
It’s important to note that at this stage, none of this data should immediately be used to make a decision, whether that’s approving or declining a line of credit, for example. There are a number of reasons why accounting data may initially seem incomplete, including:
- Only one entity has been connected
- The latest transactions have not yet been pulled from their bank
- The bookkeeper hasn’t had a chance to categorize last week’s transactions
- etc.
Let’s go back to our example of a restaurant owner-operator applying for a line of credit. Their accounting data might show that money consistently comes in from sales each day, but expenses related to payroll and food and beverage suppliers are recorded a bit later. In this case, the restaurant’s cash flow may initially seem stronger than it really is.
Data Standardization
Once embedded-finance platforms have a handle on what data they can access from a customer’s accounting solution, the next step is figuring out how to standardize the data so it can be used consistently. For example, one restaurant owner may record the money they make from a delivery service as revenue, while another records the money coming in from that service as revenue and the delivery fees as an expense. While both methods of recording this information are equally valid, one restaurant’s profit margins are going to look much different from the other’s.
Chart-of-accounts discrepancies like this one are surprisingly common. For example, construction companies may have chart-of-accounts entries like “Job Costs,” “Site Materials,” “Retainage,” and “Subcontractor Clearing,” while others may not. Without context, it would be easy for an embedded-finance platform to draw inaccurate conclusions about these construction companies’ financial health.
This is where a Unified Accounting API can be useful: it enables finance platforms to map all incoming data into buckets they can use to make decisions, like operating revenue, direct costs, payroll, debt payments, receivables, and discretionary spend. At the same time, though, it’s important for these high-level categories to roll up from, and therefore retain, more detailed account data. While underwriting and risk teams may only need the top-level data to make a decision, other teams, like operations, may want to know more, like whether increased costs can be attributed to one supplier category or construction project.
Data Freshness
Just as not all account types mean the same thing from one accounting solution to the next, not all transactions should be given equal weight. Accounting transactions can be pending at the bank, imported but not categorized, categorized incorrectly, categorized after close, and more.
- Confirmed: The transaction is posted, reconciled, and reflected in the account balance.
- Pending: The transaction has been imported but isn’t fully categorized and/or bank activity hasn’t cleared.
- Stale: Data is based on a last successful pull that exceeds your data-freshness threshold.
- Exception: There’s something amiss, like the transaction can’t be mapped, important information is missing, it’s a duplicate, etc.
Normalized Decisions
- For our restaurant owner-operator, an underwriting team might want to understand their cash flow: How consistent are their deposits? Do they run payroll on the same day each month? Do they pay suppliers on time? Do they have any outstanding AP? Are they experiencing any cash-flow crunches? While they may have had a great weekend, that doesn’t necessarily erase the fact that they haven’t paid some of their suppliers in weeks and their cost of goods has been trending up month over month. Rather than basing creditworthiness on a one-time outlier, embedded-finance platforms should strive to get a sense of their customers’ cadence and routine.
- For our construction company, embedded-finance platforms may want to answer a slightly different set of questions when employees submit travel, materials, and equipment expenses for reimbursement. Which project does this expense belong to? Does that project still exist? Does it have a budget? Are there other similar expenses?
- Is this data recent enough to make a decision?
- Does the transaction belong to an active project or account?
- Does the transaction fit within the customer’s available funds or project budget?
- Are there any unusual cash-flow trends or outstanding bills that might factor into a decision?
Transaction Holds
Not every exception or edge case is going to warrant pausing an automated process. Minor variations in account names can probably be mapped once and for all. Missing memos on small expenses may not have much bearing on creditworthiness. But, for example, let’s say our construction company has a sizable AR on their books, but there’s been very little incoming money reflected in their bank account.
Depending on the project, that might not be unusual, but it could also mean that their AR is aging more than expected. In a case like this, it would be a mistake for an embedded-finance platform to automatically factor in this money as available cash flow.
Or, let’s say our restaurant’s sales suddenly increased. This may have been caused by an error in account mapping, a duplicated deposit, or a prior-period adjustment that was booked at month-end. Rather than assume all incoming cash is recurring operational revenue, it’s important for expense-management platforms to figure out if there’s something else going on.
In situations like these, it’s important to understand why a decision was made. Ideally, there’s an audit trail for customers to refer to that includes all the transactions that were used, how accounts were mapped, when data was last successfully pulled, any flagged exceptions, and so on. This way, if data is incomplete or out of date, customers know exactly what steps they can take to get a response as quickly as possible.
For finance platforms, that means deciding up front what will and won’t count as an exception, how stale is too stale, and where flagged transactions can be reviewed. That way, when it’s time to make automated credit and expense decisions based on accounting data, they’re making them based on a holistic view of their customers’ financial health.
