
Blog
Next Phase of AI & Digital Category Management
By Mark Hubbard |
The rapid growth of Artificial Intelligence as a core subject in many areas of life and the incorporation of the phrase ‘AI’ into a range of products and services is a reality. At this stage, deciding how and where to embrace AI in Procurement is the challenge many centres of excellence face.
A lot of current focus is on AI tools that leverage Large Language Models and use text driven interfaces and queries – Chat GPT started the craze and is probably the most prominent one, but others like Meta’s Llama, Google’s Gemini and Microsoft’s Copilot are catching up with speed. However, there are also AI based tools which focus more on image manipulation, decision support, numerical analysis and behavioural predictions. It is far from a one solution fits all position, and understanding the marketplace for AI support, and how that fits against our individual requirements, is key at this stage.
Clearly, there is a potent argument for developing a category strategy for AI support in procurement, as all the key stages in category management immediately apply to understanding what we want and need.
Articulating business requirements for AI
The 2024 Category Management Report shows that category knowledge, skills in the application of category management, and stakeholder management are all critical to effectively deliver value. Finding an AI solution in these spaces would address many of the challenges we currently have. However, these are three distinct requirements, so we need to get more granular about exactly what we need to solve here, for our own organisation.
The global category management report shows that category knowledge, skills in application of category management and stakeholder management are all critical to effective delivery of value, so finding an AI solution in these spaces would directly address the regular challenges we have. However, these are three distinct requirements, so we need to get more granular about exactly what we need to solve here, for our own organisation.
We are likely to have data security business requirements, as well as cost of service provision. As this is an evolving technology, we are likely to have requirements around evolution of service and how to protect ourselves from that. Our largest challenge and requirement is how to integrate the approach into our overall delivery, and how to train our people to use the tools effectively.
From this we will have our own blend of requirements for AI, and, if we do it well, a balanced scorecard of those requirements showing importance and ranking.

Data analytics: bridging the gap
As we outlined in our article on Data-Driven Procurement, we are often rich in data but poor in knowledge. Using AI supported analytical tools is a sensible step as LLM AI tools are already good at pulling market overviews together or cleansing and harmonising data. But getting to an analysis of spend or understanding supply chain value quickly continues to require human intervention and validation. High quality analytics which drive insights and initiatives in an understandable way is the goal and AI tools for accelerated category strategy development are starting to help with this.
Decision support: unleashing AI’s potential
One of the poorly understood challenges of category management is the decision-making process around which option, or combination of initiatives, is most likely to deliver value to the organisation. This is often a manual process and influenced by recency, familiarity, history, and data-free opinions. At a simple level, a decision tree with value attribution can help. However, a better solution is an AI-driven decision engine that suggests the most effective combination of business requirement inputs and available value levers to recommend relevant initiatives.
In a perfect world, getting to the right combination includes the implementation approach and required resources to reflect the reality that outcomes are often blunted by the limited ability to deliver change.
The general approach here is the use of Intelligent Decision Support Systems, which typically use Natural Language Processing approaches to integrate broad data sets and use AI decision support software to guide the decision-making process.
Governance and overview: navigating complexity
Many category management programs falter as activities stall, both in strategy development and within implementation. AI can support in keeping track of the wide range of teams, activities and initiatives that all need to move together and intertwine. Being able to analyse the content of these interactions can help identify intervention needs before projects become unrecoverable. Resource management and reporting is something AI can contribute to as humans tend to be less diligent with mundane tasks.
Conclusions
AI isn’t a one-size-fits-all solution and use cases will continue to emerge. We really need to focus on what we need to achieve and test the most promising solutions. We also need to ensure that we allow for flexibility, experimentation, and fast failure. Building a category strategy to identify and deliver the best AI solution is the way to go here; although we may have to do that without AI help at the start.
Further reading:
Blog: An AI Driven Approach That Dramatically Accelerates Category Strategies delivery

About Mark Hubbard
Director
30+ years experience in procurement and supplier management, in line and consulting roles
Previous employment: Positive Purchasing Ltd, SITA,
QP Group, BMW, SWWS, Rover
Education: BSc in Engineering Metallurgy, MBA University of Plymouth
CIPS: Member