Part 2: The risks of manual reconciliation and the case for automation
This is part 2 of a summary of the “Ease of Automated Reconciliation” webinar held by Credrails In June 2023. Hosted by Munyaradzi Hoto, the expert panel included:
Credrails is an open banking platform that uses data to build products. The company is currently present in two markets, Kenya and Nigeria, with clients ranging from agency banking to neobanks. Automated reconciliation is one of their flagship products.
How automation is transforming businesses
“Daily reconciliations are indeed painful; sometimes you need to reconcile multiple providers in a given day. Because of customer issues, you may also need to do reconciliation every hour. Some of these reconciliations can’t wait for the following day or even the end of the day. In the case of banks, again, the risk of losing money is so real.
So what has been the case is teams will use Excel, which is very manual and prone to error. It means if you miss just one formula, you will have missed a variance that was supposed to be
picked. It means customers will not be refunded in good time or you will lose money.
When it comes to the benefits we’ve seen our customers enjoy, it’s accuracy:
“There’s no such thing as 100% accuracy. So we are 99.9%. And because it’s an intelligence system, we’re always trying to teach it once it catches an error, to resolve that error and to improve on that going forward. We cut out a lot of the errors that could happen. For example, an Excel sheet can’t do more than a million lines at once. That is eliminated once you start doing
automated reconciliation. It eliminates a majority of the errors that would otherwise be caused by doing lots of reconciliations at once by a number of humans with differnt levels of understanding of the reconciliation process.
Trepidation surrounding adoption of automation tools
“I must be sincere and say that throughout my time and experiences, adoption of automated
systems has not been greeted with excitement. Now, the reasons are:
“We had the same thing in mind when we started [Credrails] and started getting reconciliation going. We thought of systems that were not fully automating the reconciliation process, where the team had to do part of it outside the system and some of it inside. What we have done with our system is make it flexible and try to understand the touch points that the customers have to go through. For example, a lot of the systems we have seen require their users to standardize their statements into a certain kind of format before it can be accepted into the system. We have eliminated that part in our system, so [Credrails] is format-agnostic. It can ingest any time of statement from the banks and providers without asking customers to do anything manual. When it comes to reporting, we also understand that different businesses have different reporting requirements. In that area, we’ve also been very flexible to ensure that if it’s a CFO of a certain business and certain data is important to them, it’ll be available to them on the dashboard.”
“We generally let our users try, in the sense that they’ll have a period before they migrate everything and during that period we help them with whatever issues there might be. Because
it’s machine learning, we need to train the system to understand new patterns. What we’ve found when it comes to that trepidation [of adoption], is that the more responsive we are as a company to the problem they have, the better adoption is. The more people feel that we actually feel their pain point, the better adoption is. I think the most important thing is that people like to know that they have been heard. We do our very best to hear out customers and help solve their problems as they arise."
How does automated reconciliation help businesses where there are lots of transactions with thousands of similar amounts and no unique identifier?
“Our system uses a unique identifier to be able to match but when there is an absence of a unique identifier we concatenate different parameters that would be available on the statement.
So perhaps in one case we could concatenate the customer name or the customer ID, or even the amount they paid or the date when they paid.
On your platform. If you have captured this information on your platform, it can be checked, matched versus the bank statement or the statement of the provider using the same parameters, (date, customer name or ID, amount paid, etc).”
What happens when variance isn’t caught on time?
“Mistakes happen when the entire process is completely manual. Whether it’s due to inexperienced staff, oversight, or fraudulent attacks, sometimes when an error occurs, you don’t even know where to go back to check exactly. It may appear that you are balanced, and that’s the danger of reconciliation. You could appear to be fine for the next year, or ten years. But unfortunately by the time that error of 10 years ago hits you today, you are completely gone and out. [When errors materialize], you put in all the multilevel controls in that entire reconciliation process, in fact, increased levels of manual processes end up being included. The multilevel controls are so you are plugging every possible gap at every point so you can mitigate risks to the organization.”
“It’s extremely difficult to know who you need to refund and how much if you’re not reconciling on a daily basis. Many of us have the experience of having to go back to your bank for chargebacks or reversals of a transaction. It takes a lot of time and it’s extremely painful, and part of this is because there are no records of automated reconciliation. Sometimes I’ll go to a bank office and I’m always surprised when I see banks that have paper as their back office operations and they have to go through all those transactions. It’s costing the customer, and in these days, where there’s so much innovation, it’s extremely difficult to maintain customer loyalty if you’re not giving them the best customer experience."
“There’s also real money getting lost due to system glitches. One transaction could be processed two or three times, meaning the beneficiary got credited three times, and when reconciliation is manual, it means you don’t catch this issue as it’s happening. You only see it two or three days later and recovering that money from the person who has been credited twice is very hard.”