Bits on Bots: An Automation Playbook for Finance & Supply Chain
RPA Deep Dive into Industry Horizontals
In its own small way, the Bits on Bots series has turned into a conversation, which makes me very pleased. This week’s post comes from a comment on our previous blog post.
Retail and hospitals seem slightly ahead in terms of using automation but it would be interesting to read your take on more ‘legacy’ types of B2B businesses such as banking and supply chain as lack of data availability becomes a bigger issue in those mammoth, fragmented businesses.
She’s right. After creating a framework of activity groupings within industry verticals to look at the feasibility of RPA, we should go broader to look into the transformative impact of RPA across a few industry horizontals. I looked into the following horizontals:
After examining the penetration and potential impact of RPA across industry horizontals, this post looks at two ends of the spectrum: Finance & Accounting (F&A), and Supply Chain Management (SCM, including procurement, and order management).
🚢 Supply Chain Management
While the market for RPA in supply chain management is still relatively under-penetrated, I believe the potential for RPA is actually considerable. Many enterprises are adopting a holistic approach to digitizing their overall value chain — procurement, production, and delivery of products and services.
RPA, at the outset, carried the notion that robots will only learn once and that they need to be taught perfect lessons for them to perform later. Due to AI, solution flexibility can now be added to all stages of automation, though agility is perceived as a challenge. We also need to document how a hybrid solution could work.
A hybrid solution combines RPA with other technologies to provide an optimal solution. For example, there are different ways in which classification and data extraction powered by optical character recognition (OCR) and machine learning (ML) leverage RPA. In some cases, it starts with those tools classifying and extracting data. In other cases, the robot handles the orchestration of the activity from an input (email, for example) to output (ERP, perhaps). Invoice processing is a natural use case and discussed below.
Areas Where RPA Solutions Already Exist
To reduce operational costs, it is important to implement RPA throughout the procure-to-pay cycle. Examples of specific user flows:
Managing Requisitions
Taking new user requests ➡ checking them against a list of multiple validation checks (nondisclosure agreements, for example) ➡ entering requisition information into an ERP ➡ creating a purchase order to facilitate straight-through processing.
Generating Invoice Exceptions Reports
Monitoring a report of unmatched invoices and purchase orders for specific error codes ➡ based on code, accessing the invoice and generating relevant messages to the business unit with invoice details ➡ notifying the business unit of follow-up steps required to resolve and match invoices to purchase orders.
Invoice Verification
Matching invoices and goods receipts across multiple different systems ➡ clearing matched items ➡ generating a report of uncleared items in order to update accruals in the balance sheet.
Areas Where RPA solutions Can Be Used
Contract Management
RPA can be used to extract metadata when uploading contracts into a contract repository, and analyze a large number of contracts in a merger/acquisition. Vendors in this space include Seal, Kira Systems, Ravn Systems, Brightleaf Solutions, and eBrevia.
Spend Analytics
There is a need to optimize the conditions around spend analytics for ML technology. RPA platform with capabilities of predictive spend analytics can further improve the planning and prioritization of strategic sourcing activities. Vendors such as Zycus, SpendHQ, and Coupa Spend360 are already using ML to improve spend classification.
Supplier Selection
An RPA bot can be leveraged to monitor and determine the most profitable inventory mix, and the level of inventory, by intelligently conducting this analysis based on data from across your organization saving you multiple hours.
Instead of manually sending RFIs/RFPs to potential suppliers, you can employ RPA tools that generate communications, send updates, and follow-up with companies in a timely, consistent and standardised manner throughout the selection process. Humans will be familiar with only a subset of suppliers, and the solutions marketed towards them. With the help of RPA, there is a big opportunity in casting a wide net, and overcoming human biases.
Performance and Risk Management
A centralized RPA bot can track and alert you when a supplier fails to hit a key performance indicator or there are product shortages obviating the need to manually search for data from different departments to determine on-time delivery of goods. RPA tools can also be set up to send you alerts to potential supply chain risks such as identifying potential suppliers with questionable reference checks.
Supplier Phasing Out or Renewal
RPA bot can automatically alert you when a contract is coming up for renewal alongside actions that must be taken in lieu of manually scrolling through many contracts. Additionally, instead of spending hours gathering information about a supplier’s performance in order to decide whether or not to renew a contract, an RPA bot can be used to pull information from across the company including payment history, on-time delivery record, and customer satisfaction levels. This will also help debias the process.
💰 Finance & Accounting
By the end of 2020, robotic process automation will eliminate 20% of non-value-added tasks within the office of finance, as most organizations are rationalizing their current business applications and RPA will be used to “bridge the gap” before newer finance systems are adopted. I believe, counter-intuitively, many finance systems do not need RPA if they have invested in more fully automated finance systems. RPA is used as a way to automate some of the manual processes when users interface with systems, and it has no finance best practices embedded in the toolsets. A key challenge, therefore, for organizations is knowing when to use RPA and when not to.
Areas Where RPA Solutions Already Exist
- Checking if vendors are already listed in the vendor master file, and adding them if they are not in the file
- Collecting through email/spreadsheets and posting entries into a centralized general ledger
- Collating operational and financial plan data from enterprise sources, combining it, and processing it offline before it is entered into a financial planning and analysis (FP&A) system
Areas Where RPA Solutions Can Be Used
Reconciliation Management
Companies must reconcile their accounts to prevent balance sheet errors, check for fraud, and avoid auditors’ negative opinions. Companies generally perform balance sheet reconciliations each month, after the books are closed for the prior month. This type of account reconciliation involves reviewing all balance sheet accounts to make sure that transactions were appropriately booked into the correct general ledger account.
Few reasons why automating reconciliations has been problematic:
- Security: There is no technical way for your accounts package to automatically synchronize with your bank account given the levels of online security that most bank accounts have applied to them, and rightly so. Think of all the passwords that have to be inputted as well as the manual key fobs with the ever changing code that have to be activated and inputted every single time you log into your bank account.
- Data is simply unavailable: Many organizations may pay you from bank accounts that do not clearly identify the company that was invoiced or with a reference that doesn’t clearly identify the invoice being paid. The problem is compounded when software is attempting to predict payments and receipts from your entered customer and supplier invoices but you haven’t entered all invoices, payments, and receipts. Do you make online payments in the same consistent manner that an automated bank reconciliation process can reliably identify? Also, what about credit notes and refunds?
- Need for intuition and experience: Many argue that automation takes away the insight and supervision required for the financial health of a company. By having vital steps hidden out of sight or carried out automatically, you are removing opportunities for a human user, with her intuition and experience, to flag inaccuracies, misuse of funds, or forthcoming financial difficulties. The accounting software will have no insight into the real world dynamics behind accounting errors, and it will not be able to reconcile what may have happened. In reality, it’s the marriage of a good bookkeeper with an RPA tool that will give a company the assurance that their affairs are clearly in order.
It takes an average of 64 days to onboard a new reconciliation and more than three-quarters of this time is typically spent on analysis and building. This time is significantly reduced to a few hours through RPA reconciliation solutions, which also minimizes risk by reducing manual intervention in matching, and validation, and capturing and correcting errors before they impact any downstream processes.
Increased regulatory scrutiny, larger amounts of data, and increasingly complex financial products have led to operational departments having to operate hundreds of reconciliations daily. RPA can be used to extract data from bank statements into reconciliation management templates, and compare account balances without needing a separate reconciliation management solution.
Integration during Financial Close
In accounting, financial close refers to the process of reducing the balance in nominal accounts, such as revenues and expenses, to zero. The close is part of the accounting cycle, and is necessary to prepare these temporary accounts for the next period’s transactions and events. Once the financial close is complete, a post-closing trial balance is performed to verify that all debits and credits have been posted to the income summary. RPA can be used to collect non-financial system metrics, PDFs/backup details and automating the email confirmation process when needed across the financial close cycle monthly or annually.
Few reasons why automating the close process hasn’t been as effective yet:
- No defined close process: Different people involved in the process “just know how things get done” — and have done it that way for years. A loose list of the tasks that need to be accomplished are kept in spreadsheets, which often aren’t shared across geographies or divisions. Technology can help, but not before the close process is documented and tracked. It’s impossible to automate processes without a clear understanding of dependencies, and task duration. “It’s always been done that way” is not a value adding reason.
- Inaccessible real-time data: Integrated planning — combining strategic, operational, and financial plans– has long been a goal of finance leaders. When they couldn’t trust the underlying data because it was so frequently inaccurate, those leaders put those plans on ice. Additionally, they didn’t have access to systems that could share information between silos — the sheer amount of data required across silos was a challenge. Plus, there are silos between actual and plan data, which must be combined for the purposes of variance reporting. There’s an opportunity for RPA tools to deliver real-time data with disparate systems allowing businesses to adapt and adjust forecasts throughout the year.
- Perception: Many finance leaders remain leery because they feel their own processes are so unique that automation solutions would never work without massive disruption and investment. In truth, it’s possible to take a measured, step-by-step approach with RPA tools that can help them realize efficiencies.
Treasury Connectivity
RPA can be leveraged to upload bank account balances from bank systems to treasury systems and place the data in a format the treasury system can process. Much of this is still manual today. Going a step further, RPA can also be employed to distribute treasury system reports to local finance personnel to communicate balances.
Few reasons why automating this workflow has been hard:
- Connectivity Protocols and File Formats: Given that treasury typically relies heavily on vendors for connectivity, there is often a lot of confusion around the difference between connectivity protocols (FTP, SWIFT, APIs, etc.) and file formats, which are the language of connectivity. Though confusion around this is common, it’s really not terribly complicated when it comes to bank reporting.
- Treasury Portals: Cloud-based software are now being used in new portals for treasury — for foreign exchange, cross-border payments, investments, debt covenants. SaaS platforms are easily suited to connect with each other as they natively support APIs, but also support technology that allows “reskinning” the user interface to make the systems look alike. For example, when managing investments, treasury teams have historically chosen to download investment data and set it up in their Treasury Management System (TMS) as either transactions or summary balances. More recently, organizations are preferring a two-way integration where they wish to make cash decisions in the TMS, digitally present that to their investment portal, and then manage the post-trade activity in their TMS.
While not every organization will demand complicated integration, those that do need to know what information they want downloaded. Do you just want balances for your account, so you can see your overall cash and liquidity? Or do you want to see actual transactions where you can actually apply some workflow to automate trade settlements? Here is where RPA can add value in bringing clarity and efficiency in the data gathering process.
RPA technologies can automate manual finance processes, especially in a fragmented finance system landscape. An RPA tool follows the process flow in the same way an Excel spreadsheet does, except the RPA tool can work across multiple applications on mainframes, client server, web and via the user interface screen interfaces.
The tools can and will be used with a person, an AI tool or two, the Internet of Things (IoT), business process management (BPM) tools, and ERPs in an additive fashion as required. Like in SCM, the crucial part is to become clear on what you actually need to do, and how to accomplish the next best actions for optimality should you use RPA or a different tool or a person.
We could delve deeper into this topic but I’d like to follow the 80/20 rule here — you get 80% of the value from 20% of the information. I hope this post gives you enough insights to get you interested without getting bogged down by unnecessary detail. Holler if you want me to analyze other industry horizontals or write on the interplay between RPA and AI.
Thanks for reading. If you know of great entrepreneurs looking to raise an initial round of funding in the areas we touched upon in this post, I’d love to hear from you on LinkedIn or Twitter. ⛹️♂️
References: ComTec, Frost & Sullivan, Gartner, Wipro
