Record to Report (R2R) - Intelligent Automation in Record to Report Process with AI, ML and RPA

1.0??Preliminaries

Modern businesses operate in a highly unpredictable environment, especially in the aftermath of the global health crisis. Access to accurate and timely information can make or break companies. R2R is a complex F&A process that’s critical to providing deep insights into a company’s strategy, operations and performance – to both internal and external stakeholders. Automating R2R creates a single source of truth and reduces risk while cutting down cycle time by days or even weeks.R2R is one of the most complex functions in enterprise finance. It processes the data that delivers critical insights on an organization’s performance—from both financial and operational angles—to internal and external audiences. R2R encompasses several critical areas within finance: financial/general ledger operations, regulatory and external reporting, treasury operations, and financial planning and analysis (FP&A). Intelligent Automation integrates Robotic Process Automation (RPA) with cognitive technologies, such as computer vision to extract information from unstructured data like invoices and credit notes, and leverages machine learning models for advanced forecasting and budgeting. It also harnesses natural language processing (NLP) techniques to extract entities and transactional data from contracts or to classify email enquiries. The result is end-to-end automation for unparalleled efficiencies, agility and scale. Increasingly, businesses across industries are deploying Intelligent Automation across the F&A function to discover and realize greater value, more so in their R2R function.

Intelligent Automation standardizes and integrates processes using intelligent workflows, enabling faster, better and more accurate reporting and closing. Record to report (R2R) involves collecting, processing, and delivery of accurate financial data. R2R provides strategic, financial, and operational feedback on the performance of the organization to inform management and other stakeholders. R2R automates the reporting and compliance processes through a steady cadence:

·??????Collection – R2R locates data in various systems, and compiles and consolidates it into one location.

·??????Processing – R2R then verifies the information, flags inconsistencies and errors, and ensures everything is validated.

·??????Delivery – Once you have validated data in one location, you can easily provide a wide range of reports for management and compliance.

Record to Report processes covers Reconciliation, Financial Planning and Analysis (FP&A), General Accounting / Trial Balance, Treasury Operations, Compliance and Regulatory Reporting.

2.0??Reconciliation

Financial reconciliations are critical to ensuring accuracy. Businesses typically rely on manual intervention due to the complex nature of reconciliations. Many of these routine processes of validating general ledger data against source systems lend themselves to automation. Bots can access and compare the general ledger and sub-ledger data, identify discrepancies, and pass adjusting entries. The reconciled data is then be uploaded back into the ERP/financial systems.

2.1??????Reconciliation Process

Reconciliation process can be segmented into 4 main sub-modules: document data extraction, matching, reconciliation, and finalization.

·??????Invoice and Journal Processing - As different bank statements and text documents may be in various forms, whether printed or as PDFs, manually performing data collection and extraction may be highly prone to errors. Therefore, it would be necessary to adopt OCR approaches for extraction and insertions into databases for reconciliation. AI tools read incoming invoices or journal instructions in various formats, validate the content based on complex self-learned rules, predict the department and account to which a journal needs to be posted. Finally, these generate and upload the journal entries in the ledger without human intervention. Any noticeable discrepancies could be noted here for further investigation in the later stages.

·??????Matching - Tools analyse input from various sources and formats to perform complex reconciliations such as intercompany, derivatives, cash, bank and other general ledger accounts. They use structured and unstructured data and learn matching rules to continuously improve the auto-matching rate. They also suggest journal entries and actions for unmatched items which can be reviewed and acted upon. Any mismatches will then be sent into the reconciliation stage.

·??????Reconciliation - This is the main stage of investigating reconciling items. Any errors will be reported to companies or individuals for fixing followed by reviews and approvals. This stage is very labour-intensive as the errors are unsolved by automated approaches. Therefore, it is crucial to optimize the previous 2 stages to ensure that the least amount of work is required during this phase.

·??????Finalization - After all the work is checked and errors are fixed, we can then update the checking lists and entries accordingly. This stage is fairly straightforward and thus can be done purely via computer-aided programs for ERP / RPA.

2.2??????Intelligent Reconciliation Systems

A growing number of channels, instrument complexity, and activity spread across multiple service providers and increased transaction frequency by consumers adds to the complexity of the reconciliation process. AI and ML will have a significant upside on the efficiency of the reconciliation process. Advanced ML algorithms can improve process efficiency across multiple reconciliation points. The reconciliation process typically entails tasks such as onboarding payment classes, extracting, and normalizing data from non-standardized file formats, defining matching rules and posting entries for settling accounts. With ML-enabled processes, the?system automatically “learns” the data sources and patterns, analyzes it for likely matches across multiple data sets, highlights reconciliation exceptions / mismatches, and presents actionable “to do” lists to resolve data issues.

RPA can automate routine, manually intensive tasks.?Let me give you an example.?Even today banks with automated reconciliation processes deploy dedicated personnel to fetch files from an interchange portal or a dispute management system, download the files and place them in the right location for the reconciliation system to act on the data.?Such tasks can be automated by use of bots, maximizing value of employee time.

Payment reconciliations have become exceedingly complex, with multiple payment options, channels, combination of product processors for different payment method across line of business. AI-based solution offers reconciliation across payment workflows, with built in support for, multi-source, multi-file many-to-many reconciliation scenarios. Intelligent Reconciliation adds value in the following ways:

·??????A unified web-based platform system to handle end-to-end reconciliation which incorporates data import, transformation and enrichment, data matching, exception management

·??????Supports all classes of digital payments using a single system – GL Reconciliation Tally, ATM/ Card Reconciliation, Online Payments, Wallets, Instant Payments (IMPS and UPI), NEFT, RTGS and QR Code Payments — with flexibility to rapidly onboard new payment channels and schemes

·??????Simplifies reconciliation process via a template-based data-mapping framework.

·??????Provides a detailed audit trail helps users understand the rationale behind a break or match case and address it accordingly.

·??????Advanced Exception Identification and analysis for advising timely action and follow ups to enable closure

·??????AI-based Settlement Processes Leveraging ML algorithms, Intelligent Reconciliation continually learns file patterns and automatically identifies new records, predicts exceptions and performs resolution actions.

·??????Support for dispute and chargeback lifecycle enabling banks to respond to disputes in much shorter time frames – enhancing efficiency as well as customer service.

3.0????General accounting

3.1??????Trial Balance Process

The trial balance is used to prepare financial statements from ledger and journal entries. It is the basis for preparing the financial statements like balance sheet and P&L accounts. The trial balance is the first step of preparing the final financial accounts, where the statements of the closing balance from general ledgers accounts are considered. The steps to prepare the trial balance are:

·??????Firstly prepare the ledger accounts and the closing balances of every account in it. For example, the bank overdraft in trial balance, the commission received in trial balance and general expenses in final accounts, among the others.

·??????Now post these balances into the trial balance’s credit and debit columns.

·??????Expenses and assets are accounted for as debit balances, while income and liabilities are considered credit balances.

·??????Next, calculate the total debit and credit balances.

·??????If the trial balance is accurate, the sum of credit and debit balances should be equal.

·??????In case of any differences in the balances, trial balance error rectification shall be undertaken through an audit of the accounts.

3.2??????Intelligent Trial Balance Automation

RPA automates complex reconciliations involved in trial balance calculations and posts to general ledger accounts. In essence, bots normalize all data, calculate a trial balance, log into the ERP system and open each account in each ledger, compare it with the general ledger balance and pass adjusting entries, as needed. They can also post the trial balance in the appropriate ledger account.

Trial balances validate an ending balance in the general ledger by starting with the previous period-ending balance and adjusting for all the subsequent transactions to calculate the current period-ending balance. One common use of RPA is to automate the trial balance calculations and the posting of any adjustments to the general ledger. RPA bots can extract all of the transaction data for the current period, calculate a trial balance, compare it to the general ledger balance, and prepare an adjusting entry if there is a discrepancy. Also, bots can post evidence of the trial balance calculation in the appropriate general ledger account.

4.0????Compliance and Regulatory Reporting

4.1??????Compliance and Regulatory Reporting Process

Regulatory reporting is the submission of data to a relevant authority in order to demonstrate compliance with the necessary regulatory provisions. In simpler terms, it is the process businesses and individuals must continually go through to show they are following all the rules. Compliance efforts are necessary practices to ensure that a business is keeping up with internal and external regulations and policies, which helps businesses detect violations and fraud, and avoid fines and lawsuits. However, according to Deloitte, accomplishing effective compliance is a challenging process due to cost, quality, and consistency of monitoring and assurance, as well as difficulty of coordination between different business departments.

4.2??????Intelligent Automation

Businesses rely on manual controls, validation and reporting to ensure accuracy and compliance. Bots can easily take over these tasks, establishing necessary audit trails and ensuring proper documentation for regulatory authorities, offering peace of mind to business leaders. To limit the risks of regulatory fines and reputational damage, financial institutions can use RPA to strengthen governance of financial processes. RPA helps consolidate data from specific systems or documents to reduce the manual business processes involved with compliance reporting. ML goes further by deciding what data an auditor might need to review, finding it and storing it in a convenient location for faster decision-making.

·??????Eliminate unauthorized access to privileged data - Different industries leverage customers’ personal data or third-party data (e.g. IoT data) to make data-driven decisions. Any error in processing and storing such data may compromise privileged data and expose it to non-authorized individuals, thus, violating data privacy regulations and policies. RPA bots can replicate data manipulation processes such as entry, transfer, and storage, without human intervention. Leveraging bots to automate data processes will: Reduce the chance of data errors, Eliminate unauthorized access to privileged data, Create an audit trail to review and monitor user activity,

·??????Verify process logs against regulations and policies - According to Deloitte, 17% of businesses do not conduct regular risk assessment to identify compliance risks, exposing the organization to reputational risks. RPAs can be deployed for monitoring and assessment of compliance-related tasks. Bots can verify process logs against regulation and policy documents to detect missing steps and non-compliant processes, and notify auditors of recurring non-compliant processes.

·??????Automate low value regulatory reporting tasks - According to Deloitte, only 24% of the time is spent on analysing reports and making data-driven decisions, remaining 76% of the report delivery time is spent on low value adding tasks, such as generating the reports according to templates and ensuring quality data. RPA bots can be leveraged for generating regulatory reports from business data in designated time of the month or year. RPA bots can be programmed to extract data from business datasets, verify that the data matches original sources (e.g. finance reports, invoices, CRM data), and generate compliance reports and send them to relevant employees for verification. Additionally, bots can identify mismatches in data and notify users of data errors and duplicates to ensure better risk assessment.

·??????Keep up to date with regulatory requirements - In a KPMG survey, 60% of organization leaders reported that they do not use automation for regulatory requirements and laws (e.g. Dodd-Frank Act, HIPAA, FISMA) which are frequently modified by policy makers. Updating internal policies to reflect external policy modifications is a repetitive and time consuming task. Rely on RPA bots to scrape policy makers’ websites, extract news about regulations, laws, or rules, and update regulation data in the organization’s internal regulation databases. Bots can also send emails to notify relevant employees about novel or modified regulations so they can update their policies accordingly.

·??????Minimize errors in onboarding and due diligence - Third-party onboarding and due diligence is done when an organization wants to outsource specific tasks to third parties, which may include access to privileged data. This is a repetitive time consuming process of third-party data entry, verification against existing legal records and documents, and risk score calculation and assignment. However, failure to assess the profiles and risks of third parties can have significant impact on business data; 74% of organizations that faced data breaches said it was a result of providing too much privileged access to third parties. Leverage bots to complete due diligence processes to reduce human errors, increase speed of onboarding completion, and generate a comprehensive audit trail for later compliance analysis. Bots can also be programmed to route third parties to employees if they came out with a risk score higher than a selected threshold.

5.0????Financial Planning and Analysis (FP & A)

When it comes to financial planning and analysis, bots can access and gather information across systems and sources and structure it. They help aggregate and format data, perform standard calculations, create preliminary budgets and management reports, and perform variance analysis. This frees up employees to focus on higher value tasks such as analysing reports and formulating business strategies

5.1??????FP & A Process

FP&A process covers:

·??????Gather and report financial data: The FP&A team works with the accounting team to collect and present a business’ finances to the c-level executives. This data includes profit-and-loss statements (P&L statements); balance sheets (analysing net income); variance analysis; income statements; annual, quarterly, and monthly reports; financial models; cash flow statements; and other financial statements.

·??????Present and compile findings: The team presents their data findings in financial reporting meetings and documents. FP&A analysts may also need to compile ad-hoc reports or reports on specific business areas that higher-ups request.

·??????Make budget recommendations: The FP&A team may also be responsible for budgeting for the entire company—creating fixed yearly budgets with a strategic plan in mind and regularly assessing the business’s budgeting against key performance indicators (KPIs) and other metrics to increase profitability.

·??????Forecast business performance: Historically, FP&A teams focused more on the company’s current financial health. Advances in forecasting technology and automation have encouraged FP&A teams to take a more active role in predicting upcoming financial developments. They also recommend decision-making approaches, often informing business planning in the short-term with rolling forecasts and the long-term—up to three years in the future. The FP&A uses its financial projections to strengthen the company’s future performance.

5.2??????Intelligent FP&A Solutions

FP&A teams deal with huge company-wide datasets. They need to aggregate, categorize, and summarize the data to make it more manageable. AI-powered applications can instantly parse through huge amounts of data to spot errors and alert the user, significantly improving predictions and forecasts. Further, AI, ML and?RPA helps FP&A in the following ways.

·??????Trend Analysis - Trend analysis is all about assessing data for a given period. This is the first AI capability that has become a part of the “predictive analytics” toolkit. It uses human interaction and AI to query data, identify trends, and test assumptions. So, FP&A analysts can use trending to detect patterns in the data of a single variable over time to generate time-based formulae that fit the series. This can then be used to derive further series along with a degree of confidence related to the accuracy of the predicted numbers.

·??????Correlation Analysis - Correlation is a statistical measure of the relationship between two variables. It recognizes that all variables, whether actions or assumptions about a business environment, are influenced by other measures. AI is being applied to automate processes like pattern and anomaly detection, which is a natural step in correlation analysis. This helps teams quickly determine how one variable move or changes with the other variable. Further, it gives them an idea about the relationship between the two variables.

·??????Neural Networks for Higher Forecast Accuracy - FP&A is non-linear, and stock price data may seem random at times. The traditional time series methods like the ARIMA model are effective only when the series is stationary, and this isn’t possible in a live trading system. That’s where neural networks come into the picture. These networks do not need stationarity, and by nature, they effectively determine the relationship between data and use it to predict new data. Further, artificial neural networks have a huge potential to increase forecast accuracy. In fact, these networks can offer far higher accuracy compared to the traditional methods.

·??????Data Models - Previously, the above-mentioned AI capabilities needed data to be stored in some form of model that had a defined structure involving a series of relational database tables, a multidimensional database, or both. Today, we have tables with defined fields and records that do not require data to be kept in a predefined model. Advancement in AI is introducing sophisticated ML algorithms like decision trees and neural networks that are helping teams to solve the issues they face.

·??????Automated Management Processes - AI is helping FP&A to create effective and efficient workflows for their organization. The technology helps adapt and withstand ever-changing business environment and automate the repeated processes, thus improving work efficiency. RPA is empowered by ML and AI and helps FP&A teams manage a large volume of redundant processes that otherwise need a dedicated human workforce. AI-embedded automation management processes are allowing FP&A analysts to overcome performance bottlenecks and improve the quality and efficiency of the operations.

6.0????Treasury Operations

6.1??????Treasury Operations Process

The significant core functions of a corporate treasury department include:

Cash and Liquidity Management - It is treasury's 'primary duty.' Essentially, a company needs to be able to meet its financial obligations as they fall due to pay their employees, suppliers, lenders and shareholders. To maintain liquidity, or solvency: a company needs to have the funds available that will enable it to stay in business. In addition to dealing with payment transactions; cash management also includes planning, account organisation, cash flow monitoring, managing bank accounts, electronic banking, pooling and netting as well as the functions of in-house banks.

Risk Management - Risk management is the discipline of managing financial risks to allow the company to meet its financial obligations and ensure predictable business performance. The aim of Risk Management is to identify, measure, and manage risks that could have a significant impact on the business' goals. The objective is not to eliminate all risk. Taking risk is a critical part of any business – "no risk no profit" considering risk appetite. Treasurers are then typically responsible for managing:

·??????Liquidity risk: the company is unable to fund itself or is unable to meet its obligations;

·??????Market risk (or price risk): changes in market prices (typically foreign exchange, interest rates, commodities) cause losses to the business;

·??????Credit risk: that a counterparty default causes loss to the business;

·??????Operational risk: fraud or error cause losses to the business.

6.2??????Intelligent Treasury Automation

Treasury delivers on its mandate to optimise financial assets and liabilities, drive cash flow improvements, process cash transactions, and manage financial risk is on the face of profound change. The goal: a digital treasury ecosystem where the treasurer makes real-time financial decisions utilising interfaces and intelligent technologies.

Consolidating cash flow and exposure information

·??????RPA consolidates incremental data through automated processes on existing platforms (e.g., ERP).

·??????Data analytics scrubs and validates the consolidated data and studies key variables over time—historic balances, intercompany and third-party payments and receipts, forecasts versus actuals—to understand the real levels of cash required in each business.

·??????AI generates cash flow and non-functional currency projections and highlights anomalies. AI also calculates reliability of forecast up to 18 months by running an algorithm to compare past forecasts against actuals.

·??????Cloud dashboard presents global cash and exposures pulled from all systems

Understanding cash flow and exposures

·??????RPA updates forecast in real time, including market data feeds.

·??????AI analyses forecasted cash flows and exposures against economic scenarios.

·??????AI recommends hedging strategies.

·??????Cloud dashboard updates to reflect recommended hedges, potential exposure position and profit and loss impact.

Approval for trading

·??????Cloud dashboard allows treasurer to review the analysis and approve/modify recommendations to manage cash flow and risk.

·??????AI allows treasurer to run trades against a tweaked scenario.

·??????RPA prepares hedge transaction.

·??????Fintechs / APIs collect quotes from counterparties. AI selects quote based on wallet distribution criteria and past performance.

·??????Blockchain trades and confirms the transactions.

·??????RPA updates detailed cash forecast.

Payment factory

·??????RPA retrieves all three-way matched vendor payments due. AI validates vendors and their bank accounts to black and white lists and highlights modifications in static data compared to approve orders.

·??????AI spots opportunities to reroute transactions to another bank and proposes a payments on behalf (POBO) transaction.

·??????Fintechs support digital payments and collections.

Strategy and projects

·??????Cloud dashboards provide insights into business activities.

·??????Data mining tools source data and identify patterns.

·??????Business partnering to understand projects and financial exposures.

·??????Support initiatives such as M&A and working capital management.

·??????Application programme interface (APIs) supports KYC process.

Settlement

·??????RPA combines all warehoused transactions (Treasury and operation).

·??????AI creates POBO transaction.

·??????Blockchain settles transactions in real time.

·??????RPA posts cash in Treasury ledger.

Digital tools can enable treasurers to better manage Treasury processes, leverage data for decisions, and support their target operating model.

·??????AI, RPA, blockchain, and cloud dashboards can transform the FX management process from initiation to execution

·??????Leverage AI to accurately recommend forecasts, trades and transactions based on historical data trends, key business drivers, and market insights

·??????Utilise cloud to increase access to Treasury Management System, ERP, and banking systems, allowing for payments and reporting anywhere and at any time

·??????A smart contract on the blockchain allows multiple counterparties to interact in a mutually agreeable format, with proper documentation, for the benefit of a commercial transaction

·??????IoT-enabled supply chain leads to quicker conversion cycles and better alignment of financial metrics tied to commercial operations

·??????Process automation through RPA standardises the service level agreement process

7.0??Conclusion

Intelligent record-to-report process brings several benefits including: a process that is extremely streamlined from start to finish, enforce clear communication, create a close process owner that ultimately creates accountability and responsibility, continuous monitoring of KPIs and flash reporting, gives an opportunity to set up record to report best practices. Successful applications of AI in R2R includes the acceleration of data engineering, validation, variance analysis and exception handling, audit support, all the way to report preparation. Unlike humans, robots are specifically designed to compare or reconcile large data sets, or identify exceptions with speed and accuracy. Processes like intercompany accounting, journal entries, GL reconciliations, period-close, statutory and management reporting, and variance analysis all fall into the category of ideal automation candidates. Robots can be trained to assist with regulatory and compliance activities, or to provide support for audits and taxes as well. Any rules-based decision-making can be replicated in a robot’s instructions to be executed with 100% accuracy every time the function is performed. An effective closing process will ensure that companies close their books the very next day. This, however, can be achieved only by having intelligent automation that is integrated with the process.

References

Intelligent Automation in record-to-report: Top 4 use cases, https://www.firstsource.com/blog/intelligent-automation-in-record-to-report-top-4-use-cases

Sathish N, AI, ML and RPA can strengthen reconciliation systems for BFSI sector, October 2020, https://techobserver.in/2020/10/05/ai-ml-and-rpa-can-strengthen-reconciliation-systems-for-bfsi-sector-sathish-n-fss/

Bardia Eshghi, 4 Ways to Improve Record to Report (R2R) by Automation, March 2022, https://research.aimultiple.com/r2r-automation/?

Lucy Manole, 5 AI Technologies Powering Intelligent FP&A Solutions, February 2022, https://itchronicles.com/artificial-intelligence/5-ai-technologies-powering-intelligent-fpa-solutions/

Alamira Jouman Hajjar, 5 Ways to Accurate & Insightful Compliance With RPA in 2022, January 2022, https://research.aimultiple.com/compliance-automation/