The Hive CFO Forum Workshop Series: Record to Report Optimization with AI

As part of their quarterly CFO Forum Workshop series, The Hive Think Tank held their second workshop, co-sponsored by Live Objects, on May 22 and May 23, 2019, on Record to Report (R2R) Optimization with Artificial Intelligence (AI). This workshop explored the emerging role of AI, automation and process optimization in Record to Report and specifically, applications in month-end close activities including transactional processing, sub-ledger aggregation, analysis and adjustments, corporate close and consolidation through to management and financial reporting. The workshops were held in San Francisco, which was moderated by Caroline Abadjian – Principal, Deloitte, and in Mountain View, which was moderated by David Cutbill – Principal, Deloitte. The participants of the workshop were Finance and Accounting leaders in diverse industries. They shared their experiences, best practices and challenges in accelerating the month-end close whilst maintaining and potentially improving the quality and accuracy of the financials. Furthermore, the participants shared their journeys in adopting digital technologies in their Finance Transformation programs – the experiences were varied.  This blog describes the key insights and challenges that came from the session. The session explored the challenges in the month-end close process across a broad spectrum of industry verticals represented by the panelists in the workshop.  These challenges emerged in the following four broader themes:

  • AI applications in Finance Transformation and how to start your AI journey
  • Transaction Processing and Sub-Ledger Close
  • Analysis and Reconciliations
  • Close Consolidation, Management, and Financial Reporting


AI applications in Finance Transformation and how to start your AI journey

Advanced digital technologies and AI can support enterprise finance functions through (a) Automation – e.g. taking manual processes and repetitive processes and enabling technology such as RPA (robotic process automation), (b) Validation – e.g. applying business rules, thresholds, and historical trends to support reviews and reasonableness tests normally conducted by Finance professionals, (c) Optimization – e.g. helps look for and fix inefficiencies such as redundancies, bottlenecks, delays in business processes as well as the ability to suggest new content or configurations.

Participants were curious about how and when to enter into the AI journey. Namely, what types of practical AI applications are being implemented by Finance organizations.  Two important rules of thumb are to (a) know what KPI target you are trying to aim for and (b) to fully understand where your company lies on the maturity cycle.  In other words, if it is for cost reduction, then automation may be the better entry point.  However, if you are looking to streamline and scale your processes, and deliver improved operational, business and customer experience outcomes, then optimization may be a viable option to explore.

A practical example of automation includes the use of RPA bots to accelerate data engineering efforts such as the preparation of month-end close excel workbooks with populated source data on business day one for the accounting team to begin their reconciliations.

Participants who were in the beginning stages of their AI and automation journey discussed how they determined where to dive in. Some practical advice shared among the participants included picking a few areas to do a proof of concept (proof of value) first to show the immediate value and then expanding that to other use cases. To understand automation candidates, it is important to get a deep understanding of tasks performed, which includes getting an understanding down to the click level.


Transaction Processing and Sub-Ledger Close
The participants unanimously agreed that the sales cycle with it typical “hockey-stick sales structure” that tends to see deals closing on the last day and even last hour of the month, presented challenges to transaction processing and validation efforts.  A viable solution available would be to extract all contract data (both from the system as well as pdf contracts) into a streamlined and structured process that can be efficiently evaluated for completeness and to quickly understand (and action) pricing and margin decisions made by the sales team on any last-minute contracts.

The accrual process, especially with general and administrative expenses, also creates challenges at month-end.  Examples include how to exhaustively manage PO data from vendors to ensure all outstanding obligations have been properly accounted for.  AI technologies can help to mine this data in an efficient manner and furthermore take this a step further by generating accruals based on pre-determined business rules provided by the knowledgeable accountant.

In line with the aforementioned example of PO data mining, AI and automation can support the very manual process of vendor follow-up, e.g. finding out billing progress from legal and professional services firms.  Bots can be deployed to do the first few initial email follow-ups.  Any escalations that would require human intervention can be placed be integrated back into workflows and be orchestrated by AI.


Analysis and Reconciliations
A use case for AI is to search for anomalies and to raise alerts based on historical patterns or thresholds.  For Finance and Accounting, the practical use cases can be applied directly to variance analysis and reasonableness reviews.  More specifically, it can be applied to cash settlement processes e.g. transaction matching, to most GL account reconciliations where transactions are compared between source or sub-ledgers to the GL for matching, differences such as in transit items, and/or errors.

Exception and anomaly handling can be done in real time and furthermore, a best practice is to be able to create “what-if” scenarios and build this into intelligent workflows.   Furthermore, AI applications can capture how SMEs are addressing exceptions over a historical data set such that AI can continuously learn from how exceptions are handled. This also has practical applications in account reconciliations.

On the subject of account reconciliations alongside a tight timeline of the close, accounting teams have to prioritize their efforts to the higher risk accounts.  Some AI-enabled best practices among the participants have been use bots to focus on sourcing data from the multiple subledgers, scrubbing the data for initial exceptions and escalating these for SME review and finally, applying excel functions such as generating pivot tables automatically into an excel workbook.


Close, Consolidation, Management and Financial Reporting
Consolidation entries are also great candidates for AI.  Any types of routine consolidations can be automated.  Furthermore, any reasonableness reviews and checkpoints can be incorporated into pre-defined workflows.  In terms of financial report preparation where there is a standardized presentation format, the data collation and validation checking could be applied to help create initial drafts of the reports.  AI can be leveraged to highlight any anomalies and escalated to Finance and Accounting SMEs for further review.

The preparation of investor relations packages and datasheets could be accelerated with technology.  This allows for SMEs to spend more time on reviewing the investor packages and crafting their stories to external analysts.    Similarly, the preparation of the Controller’s packages for internal use could benefit from AI-enabled data handling and workflows.  Additionally, reports generated from a systemized approach provide additional levels of auditability and traceability.  This supports the heavy-lifting involved in generating audit back-up, workpapers, and PBCs.


Conclusion
The workshop highlighted multiple practical applications of AI to the Record to Report process in the Controllership organization.  Successful applications of AI in R2R included the acceleration of data engineering, validation, variance analysis and exception handling, audit support, all the way to report preparation.    Despite the diverse industries represented by the participants, common challenges such as how to decide what, when and how to implement automation and process optimization in Finance were prevalent.  A few important takeaways learned successful applications of AI included (a) understanding what KPI is being addressed (b) identifying the right proof of value projects in order to generate quick wins and in turn, provide ROI to
management (c) continue to stay exposed to emerging AI technologies such as RPA, Process Optimization, Unstructured Data Processing, Machine Learning, etc.


Background of The Hive Think Tank’s Quarterly CFO Forum Workshops
In February 2019, The Hive Think Tank launched the CFO Forum Workshop Series; this is organized as private events that explore emerging applications of artificial intelligence (AI) in enterprise finance functions. This series brings together Finance thought leaders across Fortune 500 companies, industry experts from big advisory firms and entrepreneurs creating new AI-driven enterprise finance applications. Workshops in this series will address the CFO organization’s themes like Revenue Management, Strategy & Planning, Accounting, Internal Control, Audit, GRC, Treasury etc.

How to mine contracts from unstructured data sources with AI and Machine Learning

Companies process a variety of contracts in their business operations. It can be a sales contract, a revenue agreement, a procurement contract, or a legal contract. Authoring a contract with right clauses and obligations is a complex process and requires a lot of manual effort. Many valuable resources are spent in authoring and maintaining these contracts to ensure accuracy and legal compliance. There is contract lifecycle management (CLM) systems available in the market that can streamline the contract processing in an organization. But they cannot address manual touchpoints in the system that span beyond the scope of the system.

Contracts can come from 3rd party sources, external systems, and legacy systems in the form of PDF or word documents. These documents are reviewed by business users and transferred to CLM systems manually. This entire process takes a lot of time and can involve human errors that result in compliance issues and increase the risk. It takes a considerable amount of resources to identify and correct the errors. Many organizations have dedicated resources to manage contracts and handle their complex process. AI and machine learning can learn complex contract processes and rules, and maintain contracts accurately.

The Live Objects platform can read and learn contracts data from unstructured sources like pdf and word documents. The AI algorithms will mine the various components of contracts like clauses, T&C’s and obligations, and maintain a repository of contracts in the CLM system. This will eliminate the manual effort to read, understand, and create contracts. Live objects can create orders, invoices and other related transactions from the contracts and link them to the source. This will help reconcile the transactions, renew contracts and manage mid-term contract changes.

Live Objects can maintain an intelligent clause library in the CLM system based on the learnings from the existing contracts, that can be used to author new contracts or amend existing contracts. Using AI algorithms, the system can extract obligations from unstructured contract data and help to track them during the contract life cycle. The platform can compare new contracts with the existing playbook and help the users to redline the contract based on matching accuracy. It can alert users for possible actions on the contract. This can reduce the cycle time of contracts and the compliance risk.

Live Objects can perform the process mining of an existing CLM process and identify the bottlenecks to optimize the process. This is part of the closed-loop self-optimizing business process transformation strategy of Live Objects that will discover, recommend, and transform the CLM process with process mining using AI and machine learning. Live Objects can work with any CLM product (Apttus, Icertis, SpringCM, and others) and optimize their process.