The Pursuit of Process Standardization
Process standardization has been a guiding principle for shared services and Global Business Services (GBS) since their inception. Traditionally, business services organizations have adhered to the 80/15/5 rule:
- 80% of a process is globally standardized
- 15% is regionally standardized
- 5% is tailored to local needs
The ambition has been to push for even greater global standardization, based on the belief that scalability and productivity improvements stem from uniformity. This model relied on humans executing processes within Enterprise Resourcing Planning (ERP) and other structured, rules-based enterprise systems. Introducing variability often made these systems more complex, reduced productivity, and increased the risk of errors.
What are the Limitations of Standardization?
Although standardization offers clear advantages, it also imposes constraints. There are genuine differences in business conditions that require deviations from standard processes. AI offers a way to balance the benefits of standardization with the need for flexibility, enabling optionality at scale. Unlike traditional enterprise systems using deterministic logic ("if this is true, then do this"), AI operates primarily on probabilistic reasoning. This allows AI to evaluate different variables and recommend optimal, tailored processes according to specific business conditions.
Applying AI to GBS Processes for Optionality
Consider accounts payable (AP), one of the most transactional and rules-based GBS processes. Multiple variables – such as purchase amount, type of purchase, supplier nature, and whether the purchase is one-off or repetitive – determine the preferred AP process (e.g., 3-way match, Evaluated Receipts Settlement, post-payment audit and approval, Procurement card).
GBS leaders have historically set standard processes for the most common variable combinations, such as repetitive purchases under $10,000 from selected vendors, using a Procurement Card. However, it is impractical to program ERP systems and train staff to choose the optimal process across too many interdependent variables.
AI can address this challenge by evaluating these variables probabilistically and recommending the best process. As processes become less transactional and more analytical or planning-oriented, AI's ability to guide optimal decisions becomes increasingly impactful.
Data Flexibility and AI
Alongside process flexibility, AI also enables greater data flexibility. Data is essential to AI, and organizations have sought a single version of the truth (SVOT). However, a single data definition often lacks context. For example, Finance may define revenue on an accrual basis, while Marketing may use a cash basis for decision-making. AI can apply the correct data definitions according to the business context, eliminating reliance on a SVOT and ensuring data is used appropriately.
Mass Customization: The New Frontier
AI empowers GBS organizations to achieve mass customization – the ability to process at scale without sacrificing flexibility needed to adapt to varying business conditions. As organizations pursue this goal, several practical guidelines should be considered.
- Augment, Don't Abandon Enterprise Systems and Processing Standards. Enterprise systems are designed to enable standard processing and decision-making for the most common business conditions. There is no need to abandon these systems for most work. AI can enhance these systems when exceptions arise. Enterprise systems and process standards should be used when there is a single correct answer and accuracy and compliance are critical. AI is best applied when multiple correct answers exist and nuance, innovation, and creativity are required.
- Contextualize, Don't Standardize Data. AI can make probabilistic inferences from disparate data sets, but this capability also carries risk. It can create inconsistent or inaccurate perspectives, potentially masking them as clarity. To mitigate this risk, AI should be used to highlight differences in data sets concerning applicability, recency, and source. This approach enables the creation of a contextual, rather than absolute, version of the truth, allowing users to understand both AI's output and its relevance to specific situations.
- Establish Foundational Organizational and Governance Elements. Effective process and data management typically require a federated model with centralized policy and direction and more distributed execution. This involves orchestrators such as Global Process Owners for each key end-to-end process and Data Owners for each critical data domain. Process owners must focus not just on adopting standard leading practices, but also training AI on the underlying logic and assumptions beneath these practices. Data owners should aim to create a contextual, not absolute, version of the truth.
AI Strikes the Balance of Standardization and Agility
Business services organizations have historically struggled to balance the control and scale benefits of centralization and standardization with the adaptability and customization benefits of decentralization. When applied within the right process and data organization and governance framework, AI can be a powerful enabler for achieving flexibility at scale.
