The need to close the books more quickly, accurately and with less risk is a constant pressure that can only be relieved through re-engineering critical components of the financial close process. One of the most critical, manual and time consuming tasks is account reconciliation.

Fortunately, there is new hope for account reconciliation process improvement given the following:

• Cloud-based computing is a more mature and robust platform now than ever before allowing new solutions to be created rapidly and providing great flexibility to F&A staff.
• Cost of storage and processing has come down dramatically making these systems more cost effective than ever
• Technologies like Robotic Process Automation and Machine Learning are improving at a rapid rate and are finally being applied specifically to applications to help the F&A teams.

When we want to improve the process to create more efficiencies, there are two basic strategies:

1. Improve the way you handle the process you currently have - perfect approach for Intelligent Automation.
2. Change the process to be more efficient, eg. eliminate tasks, fewer steps, or less manual work. The art and science of process change management applies here.

Improving the Process You Currently Have

The first step of process improvement is to realistically map out the current process. This would include all of the decisions, actions, tools and delivery artifacts from every step of the workflow. Here is a good description of a process mapping process which includes the 7 types of flow in accounting.

The process mapping process helps us understand where we are spending time and on what tasks. Based on this understanding we can look for solutions to automate the tasks in the process to take out repetitive, manual work or find ways to do the task more efficiently to achieve a similar impact.

RPA Helps Automate Specific, Manual Tasks (not Processes)

Automating repetitive manual tasks decrease overall process time when the manual tasks take a significant amount of time. They are particularly helpful when the tasks span across systems and the user interfaces are static. This normally happens with legacy type systems which are slower to evolve vs newer, cloud-based platforms that tend to update more often since they are not on-premises. Since discrete manual tasks tend to be a modular step in the process, the updating of these automations as data or systems change tend to be much easier as well.

Finance and accounting teams have traditionally relied on using Excel to help with many of the tasks in account reconciliation. Excel Macros, Visual Basic scripting and data manipulation features like pivot tables and advanced filtering have all contributed to improving efficiency in the process, but have limitations across non-Excel systems.

RPA has taken the notion of recording user actions (similar to a macro in Excel), and expanded it software outside of spreadsheets. RPA sits on top of multiple systems and acts as a personal agent that is relatively easy to setup to do things such as:

• Open screens and login to accounts
• Download and open spreadsheets and documents
• Simple processing of invoices
• Analyze forms for missing information

New companies like UI Path and Blue Prism have built easy to use RPA platforms for building rules-based bots that replicate your actions in an automated manner. These platforms can save staff an enormous amount of time by automatically replicating what you are already doing.

However, RPA does not improve the process - it only speeds it up what you already do.

The Downside of RPA

In the Gartner Magic Quadrant report on RPA vendors, they make several points about the downside of RPA. Specifically;

• RPA scripts that connect systems and execute tasks are relatively easy to build, but are not humans who can interpret and adapt as needed. As data sources change or become incorrect, the robot won’t notice and adapt. The script has to be updated.
• RPA does not easily automate long-running processes. RPA is very task driven so stringing together many tasks within one longer process becomes more fragile the more tasks are involved. It’s like a chain that gets more brittle the longer it is because it won’t update as tasks within the process evolve.
• RPA automations create long-term technical debt, rather than overcoming it. As you overlay RPA onto current technology and tasks, you are locking yourself into those technologies instead of updating and evolving. Organizations must manually track the systems, screens and fields that each automation touches in each third-party application and update the RPA scripts as those systems change. Very challenging in a SaaS world in which product updates happen much more regularly than on-prem.

Machine Learning’s Impact on Finance Automation

The use of Artificial Intelligence - specifically Machine Learning - is an opportunity to use next generation technology to improve steps in the process. In the case of account reconciliation, one of the tedious tasks is the matching process of clearing transactions that are good and identifying the items that need to be reconciled.

For simple data sets and matching transactions use cases, you can use a rules-based approach that leverages amounts, invoice numbers or dates and matches them up. One-to-one matches for modest volumes are an example of a simple use case that can benefit from this approach. However, as matching scenarios become more complex, creating rules to match transactions (especially at a large scale) becomes very expensive to implement, the rules tend to break easily, and provide a sub-par matching accuracy.

Scenarios that make matching reconciliations more challenging:

• many-to-one, or one-to-many, or many-to-many matching transactions
• matching from multiple systems with inconsistent data formats
• daisy-chaining or combining multiple files into a matching process
• matching that requires specific timing alignment and restrictions such as clearing international trades
• aggressive acquisition of new data partners or company acquisitions, which introduces new systems
• data that “breaks” regularly due to system or human errors

These scenarios can all be handled much more efficiently with a Machine Learning approach like Sigma IQ's. An ML approach can detect data types and relationships from multiple historically-matched datasets, eliminating the need for creating rules that need a FTE (or more) for maintenance.

Changing the Process

While complex matching reconciliation process can be dramatically improved through the use AI and Machine Learning, a re-thinking of the entire process of account reconciliation is a bigger, more organizationally expensive endeavor that may provide larger, longer-term benefits.

The bigger the process change, the more measured the implementation.

Every business has its own account reconciliation process in place. Each is unique in their business model, maturity level, size and organizational complexity. The following steps can help in planning your process improvement:

1. Clearly define the measurements and objectives. Process change is expensive and disruptive. A clear set of metrics that will be impacted, and the objectives to achieve, paint the picture of a better organization and more efficient business. Specifically, selecting a North Star metric like time to close days, exception rateor FTE's per revenuewill help create the success vision needed to drive alignment.
2. Get an executive champion that partners with the CFO. Since the account reconciliation process sits within the office of the CFO, we are assuming they are onboard with the program. However, the cost and impact will expand across the company, so additional executive visibility and support will help when requests for resources cross the executive team's desk.
3. Build a cross-functional team. Business process change is all about people, and change is fundamentally hard. By having a broader team of people (finance, treasury, IT, etc.), you will bring a diverse set of input and a greater foundation of support for the recommended improvements.
4. Map the current process. The mapping process can often shed light on aspects of the process that can be an easy fix such as data input problem areas, overlapping responsibilities or duplicative activities. It is critical to map the time to complete and current costs of each step (FTE's & technology) to get a complete picture.
5. Talk to a wide variety of team members, partners and prospective software vendors. The more people you interview who touch the process, the more insight you'll gain into the pain, costs and opportunities involved in the process. While talking to frontline staff doing the reconciliations is an obvious starting point, including vendors who have solutions for account reconciliations can often bring new perspectives and possible solutions for consideration. Of course they want to sell you on their solution, but listening and having them evaluate your situation doesn't cost much more than time and can be quite educational.
6. Create a strategy and action plan. Once you are comfortable that the research has been done, problems identified and solutions agreed upon, present the plan with a focus on why the changes are important outside of the core group. This has two main objectives: First, it battle hardens the plan. Journalists call this “using a red-team" — asking hard questions about the story outside of the team that created it. In a large process change environment, this type of scrutiny can catch problems before they get executed. Second, to gain buy-in from a broader audience whose help you may need in implementing, for gaining resources, etc.
7. Implement the plan in iterative time boxed chunks. The bigger the process change, the more measured the implementation. Ideally, this is done in a separate process workflow that does not compromise the current process during proof of concept and development.
8. Document what works, what is challenging and what is learned. The team will invariably learn many things during the proof of concept or initial implementation. It is critical to capture what is learned, iterate the plan and roll out accordingly. The things learned need to be shared with the wider group of stakeholders to help the organization understand the challenges and evolve.
9. Share the change in metrics, improvement in KPI's and movement towards objectives. As the process starts to work (and it will!) keep tracking and posting the selected measurements. This can be in the form of internal emails/posts, regular team updates, company wide presentations or lessons-learned papers.

Process Improvement is a Journey, Not a Destination

New tools, technologies and approaches can help F&A teams reduce their manual work, improve the speed and accuracy of their close process and enable staff to work on higher value, more interesting work. Dedicating time to understanding the process at a deeper level, creating alignment around strategy and generating momentum with small wins can roll into big wins over time and strengthen the organization. Hopefully, this article has sparked some ideas and given you some inspiration for your own account reconciliation process improvement program.