Data Here, Data There. What’s Missing in Your AI Strategy?

Data is everywhere. Internally across business functions, externally with vendors, suppliers and third-party partners. Surely, there is no dearth of data. 

But is all this data structured and synchronized across different systems? Does it offer a single source of truth to all stakeholders? And does it empower internal teams such as procurement to make the best use of artificial intelligence (AI)? 

As the race to deploy AI gains momentum, many businesses are faltering at the first stage. They are struggling to align data across different systems. Data is stored in different formats, spread across platforms and managed by separate teams. 

A recent GEP-sponsored Procurement Leaders survey shows data quality remains a significant hurdle for businesses, with less than half (42%) of respondents rating their data quality as good or excellent.

The key issue here is integrating data across disparate systems that cannot connect and communicate effectively. This creates a siloed, fragmented landscape where data remains isolated. 

Lack of standard data formats and governance is another key challenge. This prevents businesses from harmonizing data into a standard, machine-readable format. Teams spend too much time pulling information manually. Reporting is delayed, and AI tools cannot decode the context and access the full picture. 

Who Owns the Data Infrastructure? 

Data management has traditionally been seen as an IT responsibility. However, this viewpoint is now changing, with many businesses increasingly recognizing data as a strategic asset. 

Around three-quarters of survey respondents have a hybrid data infrastructure owned jointly by procurement and IT. The hybrid model allows procurement to bridge knowledge and skills gaps by accessing specialized IT expertise and resources. 

While IT builds and manages centralized data infrastructure to connect disparate data sources, procurement gets an opportunity to influence data-related decisions. This also drives better collaboration between procurement and IT. 

The joint ownership model supports better design, stronger alignment and faster results. Chief procurement officers (CPO) can ensure that systems are not built in isolation and meet real business needs. 

To get AI-ready, you need to identify your data owners, stewards and custodians as well as AI use cases, says Sonali Bhavsar, head of data and AI at GEP in this webcast. 

Shared ownership of data infrastructure extends to data governance, as procurement teams establish dedicated forums and data-focused roles. There are data owners who are responsible for strategic direction of data categories such as invoices, purchase orders and contracts. 

Then there are data stewards who manage day-to-day operations and address data quality and integration issues. The stewards are more of change agents working very closely with the distributed stakeholders, says Bhavsar. They could be CPOs, CTOs, or even supply chain officers or CIOs, she adds. 

Where Should You Store Data 

Lack of data storage infrastructure also prevents teams from making the best use of AI. The majority of Procurement Leaders survey respondents currently do not have AI-friendly data storage infrastructure. Up to 51% of respondents do not have a data lake, while 76% do not have a data lakehouse. 

Compared to traditional data warehouses, data lakes and lakehouses are considered more agile and robust for supporting AI applications. They can handle structured, semi-structured and unstructured data. They can scale cost-effectively and support advanced analytics and machine learning workloads. 

With these capabilities, data lakes and lakehouses can accelerate the implementation of AI across all business functions and provide a single source for diverse data types. Teams can move quickly and increase the number of use cases. 

Keep a Human in the Loop 

Preparing the human workforce is as important as building a robust data foundation. The popular opinion among the workforce today is that AI will not replace humans but augment human capabilities. 

However, they still need to learn how to leverage advanced AI tools. Their role will change from manual, repetitive tasks to overseeing and validating results provided by technology. Training and change management will therefore be vital to work alongside sophisticated systems and make the best use of technology. 

Businesses that can organize and integrate data and train their staff will be best placed to turn business strategy into action and achieve measurable results. 

The Next Steps 

It is now clear that successful AI implementation depends on how well you can organize and integrate data. To start with, procurement should centralize and standardize existing data. Next, it should work with IT to align infrastructure with the function’s strategic goals. Finally, procurement must define clear roles, responsibilities and data standards to maintain data quality. 

Download the GEP and Procurement Leaders report to learn how leading teams are preparing their data for autonomous decision-making and long-term AI success.