BeanBlog — AI in Month End Process

Author: Natasha Lam, Head of Finance Automation at Blackbook.ai

How to identify AI opportunities in the Month End Process

Today our subject matter is Month End, the bane of all accountants’ existence, right!! We are going to talk about the challenges with Month End and opportunities where AI can assist.

Ok, let’s break it down.

The KPI

Every accountant loves a good stat. The KPI’s for Month End are;

· According to a general benchmarking survey form APQC last year, Month End generally lasts anywhere between 4 and 10 days or more.

· It is expected that 3–5 days within this timeframe are wasted on rule-based tasks and manual reconciliations.

· In a CFO Insights survey, 78 percent thought Excel skills were the most important skill two years ago, and that number is now 5 percent.

The prime indicator here is that the Month End Process can take up to 1/3 or an accountant’s role every month. Any process that takes this long is a good candidate for AI. Whilst Finance has been evolving over the past few years with the introduction of partnering and advisory in commercial organisations, the fact still remains there is a lot of time wasted on non-value add tasks.

Ok, so that’s the ‘Why AI is a good fit for Month End’. Let’s jump into the ‘How it can be applied’.

Trend Analysis

AI solutions we are seeing currently in the market and their application.

- Prediction/Forecasting using Machine Learning (ML) — Generates and analyses transaction data to produce demand forecasts for products across the business using internal and external factors. For those finance colleagues that are forecasting, machine learning can assist to improve accuracy and assess information that you may not be considering due to lack of time. Algorithms have proven better at predicting market changes, weather, commodities etc

- AI powered email classification using Natural Language Processing (NLP) and ML — These tools can classify and categorise priority emails, urgent documents and notify management in real time. This is popular in shared services inboxes, such as AP/AR, Payroll, Procurement along with other customer facing inboxes.

- Commercial contract reading and interpretation using ML and NLP — Interpreting customer contracts to ensure correct charging mechanism and accurate update of billing modules. Deloitte already trialing Kira systems.

- Inventory levels and tracking behaviors using ML — Makes recommendations for adjustments and orders. Sends notification of volumes, suggested adjustments and reconciles registers.

- Coding behaviour using ML — Suggests account codes, approvers, cost centres, etc. Most upgrades in major ERP’s and new smaller tool such as Xero already have this inbuilt.

Concepts that are in POV stage and are generating a lot of discussions;

- Journal bots — Instructional chat bots, using ML technique that generates structured questions and structured answers in typed form. The information is then used by RPA to perform the transaction in the ERP. General query chat bots and virtual assistants are also proving popular internally.

- Month End Progress monitoring — As accountants progress through Month End, group accountants have the ability to view progress via a dashboard (powered by a combination of PowerBI, ML and RPA) viewing what is outstanding, what has been completed in the system, and predict what will cause delays.

- Narrative generation for reporting — Use Natural Language Generation (NLG) to populate commentary and static notes on monthly reports.

- Audit — KMPG are already trialing IBM Waston.

- Allocations and intercompany transactions — Elimination entries and reconciliations, currency conversion if required. It can also look at smart contract tech (blockchain).

- Fraud detection/prevention — Using ML to detect anomalies and flag security issues/breaches.

That is just scratching the surface of what AI can do, but if you’re thinking “in my organisation these things aren’t inefficient or don’t need to be improved,” that’s ok. This is where I segway into…

Crunching the Numbers — How to Identify AI Opportunities.

There are various ways to help surface up ideas for improvements that will help smooth out Month End, reduce the burden and assist in identifying what is actually critical about this process. Some tried and trusted approaches are;

· Processes that are time sensitive.

· Processes that add a lot of value to the entire Month End process. Get this done quicker, everybody wins.

· Processes that can be performed on a more regular basis, but you wouldn’t attempt on a granular level due to the volume and detail involved.

· Processes that continually cause inaccuracy or errors in data and absorb a lot of time and resources to correct

· Processes that require SharePoint, massive excel models or macros

· Processes that require constant scenario analysis.

Actuals vs Forecast — The Future of AI in Month End.

AI in accounting isn’t very experimental at this stage, RPA is currently the trusted tool and only early adopters are exploring AI opportunities now (and benefiting). However, we now live in a society where we expect instant results and responses; having software that can deliver results in real time will be a game changer for finance functions everywhere. So, I encourage you to explore opportunities where you can speed up Month End, this will allow you to identify errors earlier and understand results quicker, therefore improving internal controls and minimising business risk.

Over the next few years, Month End will be considered a very inefficient working pattern and will slowly no longer be a major part of an accountant’s role. Technology will streamline and optimise. Consequently, activities that remain, will revolve around the presentation, analysis, interpretation and the advisory of the results.

Start Today Tip

Run reports out of your ERP. What transactions in the system are accountants completing, for example what journals are the team completing, how often and how many?

#endmonthendmadness