How to Get Started with Accounting Automation

When modernizing their accounting processes, many teams start with automation. It’s likely at the top of your chief financial officer’s (CFO’s) list for transformation, and it’s perhaps at the top of your IT department’s list, too.

The state of the art in automation has changed dramatically over the last few years, with more advanced rules-based tools, machine learning, and transactional matching engines that promise to automate work that has traditionally been done in manual spreadsheets.

But there are also gotchas. Mismatch the automation tool with the accounting process, and you’ll get subpar results — whether that’s too much reliance on IT, not enough intelligence to automate what you need, or scratching your head after the fact trying to figure out exactly why automation did what it did. And there are certainly more than a few horror stories out there of robotic process automation (RPA) deployments gone awry. (Think messing up, but with robotic levels of speed and scale!)

Like many accounting leaders, you’ve likely dipped your toe in and noticed that there’s a lot of automation technology terminology out there — such as RPA, financial close (FC) solutions, and intelligent automation (IA). The following sections offer a quick tour of each.

You can mix and match! RPA, FC solutions, and IA aren’t mutually exclusive. For example, you might use an RPA bot for order entry, because it’s a high-volume, relatively simple task. Then you might use an FC solution to add controls and orchestrate the close. And finally, you might use an IA that understands the ins and outs of receivables automation.

RPA is a way to automate without modernizing the process. RPA developers typically use tools that capture and automate all the clicks and typing that you’d normally do in your day job in your existing applications, user interfaces (UIs), or spreadsheets. (If you’ve ever created macros in desktop applications, it’s a similar concept.)

The recorded actions are then played back on existing systems and spreadsheets — for example, typing, selecting fields, copying, and pasting — by a piece of software called a bot that performs the tasks in place of a human.

RPA often requires defining detailed step-by-step scripts that manage the automation logic (that is where to click, where to type, when to wait, and so on). Alternatively, RPA may involve using a graphical workflow editor where you can draw out the process you’d like to automate, much like a flowchart.

RPA can be an excellent choice for simple, rote legacy processes that rarely change, like entering sales orders or processing refunds. For these kinds of jobs, the cost savings can be significant. Deloitte found that the payback on RPA can be less than 12 months, with an average of 20 percent of full-time equivalent (FTE) capacity provided by robots.

RPA is excellent for automating things like opening email and attachments, moving files and folders, copying and pasting, filling in forms, and performing basic “if/then” decisions. RPA applications can also scrape data from a user interface like a website, or even extract data from documents, importing it into a spreadsheet or another system.

However, RPA tools don’t really understand accounting. Think of RPA as an obedient robot with no accounting smarts whatsoever. It takes everything you say literally and follows every instruction blindly, at superhuman speeds.

You teach it to repeat steps. Click here, type there, copy here, paste there. If there’s a mistake in the actions it recorded or something unanticipated happens (for example, maybe the UI to the enterprise resource planning [ERP] system changes), it’ll either break the automation or, worse, perpetuate the error at mind-bending scale.

Suppose you wanted to change or improve an accounting process managed by RPA. For example, maybe you want to enter a new field to copy from or add an additional “if/then” operation. You would typically work with an RPA developer, often in finance/IT, who’ll make the updates.

Financial close solutions are purpose-built applications specifically designed to be owned and managed by accounting departments. Their goal is to automate and improve the record-to-report (R2R) process.

FC solutions act as a single centralized system of automation and record for core FC processes like reconciliations, journal entries, and intercompany accounting. They also typically provide a single location for storing an audit trail, journal entries, and underlying supporting detail, which is often valuable for enabling an efficient audit and balance sheet integrity.

Unlike RPA tools, FC solutions are built specifically for accounting, so they (typically) provide templates for standard FC automations, dedicated preconfigured dashboards, and best-practice accounting workflows.

They also typically enable accounting departments to manage their close processes and all their dependencies. For example, they can route approvals and signoffs and monitor the close process, looking for things like unreconciled balances or outstanding items awaiting approval.

Crucially, the applications are usually designed to drive continuous improvement. They help accounting departments automate, see the results, and process performance metrics, and then make changes next period, so that things run increasingly smoothly over time.

IA tools are purpose-built, highly evolved algorithms that automate specific everyday tasks. They’re ideal for daily activities that are typically upstream of the FC, like accounts payable (AP) and accounts receivable (AR). In many ways, they’re like a fusion of RPA and FC solutions.

Imagine this scenario: Your company sends out 10,000 invoices a month. At the end of the month, you receive 5,000 to 10,000 payments. But these payments are all applied manually and don’t get done until week two of the following month.

Waiting too long hurts customer satisfaction with inaccuracies around overdue invoices and credit balances. There’s no clear view of the working capital position for the CFO. And accounting wastes time chasing down receivables that are already paid.

Modernizing these kinds of processes can pay big dividends, both in accounting and in the front office. However, doing so successfully takes a more targeted approach, using technology with a deep understanding of the task at hand, which is where IA comes in.

For example, an IA for a domain like cash application would consist of four key areas:

• It knows the typical AR systems. It automatically integrates data from bank files, the ERP system, and remittances.

• It understands the processes. It provides a purpose-built algorithm to match customer payments to invoices without human intervention. A remit engine processes all payments, whether automatically matched or not, and uses the available remittance documents when needed.

• It self-improves. It gets smarter each period via machine learning algorithms that learn customer behavior, typical matches, and frequency by analyzing the payee, address, amount, product, and other fields.

• It acts autonomously. It pushes applied payments to invoices back into the ERP system.

Because IAs know a particular business area so well, such as for cash application in the preceding example, they can determine things like the minimal data on a bank statement to learn precisely which customer the account belongs to. They can understand which invoices a payment relates to and how to manage things like deductions, tolerances, and taxes and fully apply the payments and journals back to the ERP system.

You’ve learned about the pros and cons of the various automation technologies out there. But where should you start to ensure the best bang for your buck?

Because accounting departments typically spend half their time in transactional accounting activities, it’s a safe bet to begin there.

Everyone’s journey is different, but the best automation projects usually start with a quick win because no one wants to get bogged down at the outset of an enormous digital transformation project.

For example, reconciling bank data and substantiating cash balances is a critical yet manual procedure for accounting teams. It typically involves retrieving bank statements, ticking and tying transactions in spreadsheets, and storing supporting documents offline.

A bank account reconciliation modernization project might consist of six steps:

1. Regularly importing bank detail and cash receipts or cash disbursements data from your ERP or other subsystem into your automation solution of choice

2. Configuring your automation system’s business rules to compare your banking data to the data in your accounting system, ideally to match as much as 90 percent of the transactions

3. Incorporating a task management process to manage accounting reviews, exceptions, and reconciling items

4. Posting adjusting journal entries back to your ERP system

5. Storing an audit trail and all supporting documents (preferably automatically)

6. Measuring results using dashboards, with metrics like reconciliation approval and rejection rates, unreconciled balances, and auto-certification rates

You’ve got your first quick win in the bag, so what’s next? There are often plenty of opportunities to streamline items in the FC, but the best automation opportunities typically fall into the following key areas.

Pop the hood on your accounting team, and you’ll find that they’re spending way too much time reconciling GL accounts. You’ll also likely find a lack of standardization. Fortunately, the process you learned earlier for bank account reconciliation can be used for many other types. Just the data and rules will vary.

General ledger account reconciliation is tightly coupled with source data systems to eliminate spreadsheets and emails as well as other manual activities in processing data. It starts with importing and reconciling transactions, and where required, substantiating balances with supporting files or comments.

A unified space for performing and approving GL account reconciliations leads to more collaboration in accounting, where preparers can better communicate with the business units.

Comparing all the details between subsystems and other systems can be very labor-intensive, no matter what industry you’re in.

For example, if you’re in retail, no doubt you’re familiar with the order-to-cash process involving comparing point of sales (POS) and payment processor detail. Painful, right?

In manufacturing, there are in-transit inventory and freight accruals to deal with. In the software industry, you may be dealing with subscriptions or commissions.

The first step is to take the time to figure out where the transactional stressors are for your team.

A new kind of technology, transaction matching engines, are typically the best fit for automating this kind of detail. They’re purpose-built to hunt down relationships in millions of line items of data and connect the dots — so you don’t have to.

These engines are typically trained with machine learning. They often use rules to try to match on every possible field, using fuzzy matches, variance thresholds, lookup tables, many-to-one matching, and other techniques to minimize human intervention. Figure 2-2 shows an example.

Some even post entries back to the GL automatically. They won’t match everything, but they can achieve match rates of more than 90 percent, enabling accounting to shift gears and instead focus on the variances and exceptions.

Some of the highest-value opportunities in transaction matching that cut time and risk typically include the following:

• Credit card (POS to payment processor)

• Intercompany (AR versus AP)

• Processing payments against invoices

• Reconciling payroll (W2 to human resources [HR] and payroll)

• Automatically clearing out open items in your ERP system

Get your free copy of Modernizing Accounting For Dummies to read the rest of the book. It provides a roadmap to help you move to modern accounting and drive finance automation at your F&A organization.