You’ve spent over a decade leading finance transformations at a Big 4 firm. When a finance integration runs into trouble during an M&A deal, how often does the root cause trace back to the technology itself?
In my experience, technologies for enterprise operations are typically built to handle almost any complex real world use case. Almost every finance M&A integration that runs into delays primarily stems from a lack of governance which involves incorrect scope definition, inadequate time for systems architecture design or insufficient understanding of the client’s business, rather than the underlying technology itself. In all of my implementations, I always ensure that the initial phase is an assessment prior to beginning any sort of “Design” or “Build”. In a recent implementation setting up an ERP for a Fortune 50 client’s recent acquisition, I mandated two weeks of workshops for the client to walk us through all their business processes so the team can validate if there were any nuances that could be a potential technology gap needing to be solved either via a business process change or additional recruiting.
You served as finance integration lead on multiple carve-outs and acquisitions. What does Day 1 readiness involve on the finance side, and what tends to catch acquirers off guard?
In all the M&A engagements I’ve led, Day 1 readiness involves ensuring a cohesive systems architecture, with defined user roles and responsibilities within a singular platform for tracking and reporting financial metrics. Something that tends to catch acquirers off guard is the difference in processes between the different companies. For example, on a previous engagement where a Fortune 50 client acquired a smaller company, the dollar thresholds for approvals were much higher compared to the company that was being acquired. For a smaller company, a purchase of over $1 million going through several rounds of approval, including all the way up to a CEO is not uncommon. On the other hand, there are several layers in a Fortune 50 prior to something reaching a CEO for an approval. Most industry leading approval workflows are built around roles rather than individuals, since this alleviates pain points if a user was to leave the firm. Hence, it’s not possible to directly lift and shift approval workflows and processes. When building out the new framework, I had to push the client team to ensure there was now a common framework for dollar thresholds and approvals to ensure both organizations aligned on the best approach without causing disruption to procurement processes.
You recently stood up a full NetSuite ERP implementation for a Fortune 50 client in under five months, a timeline most practitioners would consider unrealistic. What made that speed possible?
I think the most important element of getting my client on NetSuite in under five months, while integrating the new company was ensuring adequate time spent on planning and designing. Bringing the entire team together in-person for two weeks of workshop at the beginning of the project allowed us to define a common vision for go-live. My one ask for the client team during these workshop sessions was to provide my team with a list of negotiables and non-negotiables. We called the initial launch an “MVP” go-live and every requirement deemed non-negotiable would be part of it. Negotiables were any requirement that could have a workaround that was manual but not tedious. I also aligned with the client team that these “Negotiables” would be prioritized post go-live and deployed to their instance as a fast-follow. For example, the Controller of the Fortune 50 requested a report that would be emailed to them every day to track all the Journal Entries created that day. The system natively creates this report, but does not email it to users. I was able to ensure that the non-negotiable was that this report was produced, but the negotiable was the automatic emailing. In this particular case, the Controller of the organization had to login to the system and navigate to the report to track this metric rather than it be automatically delivered to their inbox. I did ensure that the emailing piece was a functionality that was configured as a fast-follow post go-live. But, setting these types of negotiables vs non-negotiables allowed us to operate at speed setting the stage for a successful go-live.
Where do accelerated ERP timelines run into trouble, and which parts of the process deserve protection even under deal pressure?
One of the most common features of an accelerated ERP implementation is to utilize a majority of the out of the box functionality while keeping the customizations and configurations to a minimum. This requires businesses to adapt to the system to a certain degree. An area that specifically requires protection is the User Acceptance Testing (UAT) phase. It is during this time that the clients begin their hands-on testing of the application and begin to request additional requirements or enhancements that weren’t originally captured as requirements. If there is not sufficient protection during this time, it can lead to scope creep, which can have a direct impact on the go-live. I’ve come across this situation multiple times.
There’s considerable enthusiasm right now for AI in finance and accounting. From where you sit, working in regulated reporting environments with SOX controls and audit requirements, where is that enthusiasm misplaced?
Working in a regulated reporting environment, it is critical that any solution that is implemented does not violate SOX regulations. Due to considerable enthusiasm, organizations are implementing AI at speed, which poses risks in a highly regulated environment. One of the recent examples I’ve seen is AI agents that propose, create and approve journal entries in the ERP system. Typically, journal entries are created by a team member part of the accounting team with approvals routed to the team member’s supervisor. With AI agents replacing this task, it opens up significant risk for an audit failure if there is no human-in-the-loop intervention. Apart from audit failure, it also opens up the risks to incorrect financial transactions being entered into the system that could take significant effort to unwind and correct. While implementing AI agentic solutions such as JE proposal, creation or any other SOX controls requirements, it is imperative to layer a human approval every step of the way to catch any issues along the way. This will allow companies to operate at speed, while complying with audit requirements.
For finance leaders weighing AI adoption in their close, reporting, or compliance processes over the next few years, what would you tell them to focus on first?
For any finance leader weighing AI adoption in their close processes, my recommendation is to deep dive into the problem statement prior to implementing any AI solution. In my experience there are four ways to utilize AI to solve most of the close, reporting or compliance processes. The four are below:
- Utilize AI to write code that can be deployed in the ERP instance to solve the problem statement. This would be the most cost efficient way, since this is a one time activity that does not require on-going maintenance with respect to token costs, or user access.
- Connect the ERP system to any of the LLMs such as Claude or ChatGPT using API to make calls and solve the problem. This has recurring token costs due to API usage thus increasing overall costs.
- Procure and connect an MCP so that users can directly interact with LLMs such as Claude or ChatGPT to solve the problem statement. This solution also would be more expensive due to user access costs.
- Procure a dedicated AI native 3rd party solution that solves the problem statement. This would be the most expensive solution due to integration with ERP, on-going maintenance and any additional token costs.
Picking the correct solution will have long term impacts on technology architecture stack as well as overall operational costs. Hence, my recommendation to any financial leader is to invest time and effort in understanding the problem and picking the correct solution for the long term future.
