There is a trend in bookkeeping and accounting for small businesses to rely on bank and credit card feeds to handle the bulk of transaction processing. While I applaud the advancement of technology, this trend has two significant limitations.
Limitation 1: the feeds have less than 100 percent accuracy
Every bookkeeping transaction has two sides, a debit side and a credit side. The feed has a 100 percent accuracy rate for one of those sides only. The other side has an accuracy rate of less than 100 percent, often a lot less in my experience.
The side with 100 percent accuracy
This side represents what happened at the bank or credit card. If a paper check cleared the bank, for example, then money was removed from the bank account. This is a credit for the firm’s checking account register in the software. Or, if the firm made a bank deposit, then money was added to the account. This is a debit for the firm’s checking account in the software. What happened at the bank or credit card can be handled by the feed with 100 percent accuracy.
The side with less than 100 percent accuracy
The other side of the transaction will be the opposite sort of entry from the side with 100 percent accuracy. In other words, if the first side was a debit, this side will be a credit, and vice-versa. This side is the reason for the transaction. If a paper check was used to pay for office supplies, the office supplies are the reason for the transaction. The fact that money was a withdrawal from the account will be recorded correctly, but the fact that it was for office supplies might be missed during the import process. That is a very basic example, and, in fact, the software does “learn” what do to with the transaction based on the vendor, assuming the vendor name is included as part of the import. Many debit card transactions from the same Office Depot location, if classified correctly initially, will almost certainly have a 100 percent accuracy rate during the import for both sides of the transaction.
Here are some examples of possible problems with classification on the other side of the transaction:
- New vendors. This includes when a chain or franchise is used at different locations. The vendor name is imported slightly differently for each of them.
- Paper checks not first recorded in the software.
I can imagine how someday both sides of the transaction are imported with a 100% accuracy rate. We are not there yet, and the needle (for accuracy) hasn’t moved much in at least ten years, as far as I can tell.
Limitation 2: accrual-basis reports will not be accurate
Bank and credit card feeds, if relied upon exclusively, make accrual-basis reports inaccurate. For example, business owners may neglect to record Accounts Payable on the books, if all vendors’ payments are recorded using a bank feed. Other liabilities as well may be neglected on the books if the bank feed is used exclusively. This means that the true amount of the liabilities will be understated on the Balance Sheet. For Accounts Receivable, there might also be problems with matching customer payments with their corresponding invoices if the expectation is that the bank feed will handle it all.
Unlike the first limitation, this one will be harder for technology to overcome. It is possible that someday AI will be able to create accrual-basis reports that any size business can rely on for GAAP reporting purposes (and tax and managerial). But we are not there yet. Given that not much improvement has happened with the first limitation in over ten years, I am doubtful that we will see significant improvement on this limitation for a very long time.
At least for now and for the foreseeable future, these limitations mean:
- A human must examine the feed and make adjustments to it.
- If accrual-basis reports are needed, then a human is needed.
- Relying exclusively on these feeds means that cash-basis reporting is the only option.