通过人工智能驱动的支出分析将数据转化为可操作的智能

DRIVING INSIGHT AND ACTIONS Data leads to information which, in turn, becomes insight that caninform the actions leaders should take. But following this path is much A recent survey found 71% of chief data officers spend more than one-quarter of their time preparing data for reporting and decision-making.Some 78% want to leverage proprietary data to differentiate their Data leaders believe that AI is the answer. For procurementprofessionals inundated with data, can AI help transform this Data Integrity Trends and Insights report, IBM, 2025 The proportion of decisions made bythe average procurement function MOVING BEYOND AD-HOC ANALYTICS Considering the amount of investment in tech, data platforms andpeople, analytics are not yet translating into consistent and repeatable Procurement teams are inconsistent in deciding what information they rely onto make decisions (see Figure 1, right). Very few have a defined cadence onusing analytics: how and when and using what data? How do buyers translate The use of analytics is ad-hoc; often based on adoption by individuals ratherthan across the function. Dashboards provide information, but that requires “Procurement needs to make analyticsa part of the workflow and embeddedin business operations. So, how can Alex Zhong, senior director, global head ofproduct marketing, GEP “I think that dashboards today arealready kind of dead... we are soreactive and there’s so muchnoise in our data... I think 57 The percentage of procurementprofessionals that says integrationdifficulties are a barrier to data quality OVERCOMING OBSTACLES TO DELIVERING AUTONOMOUS ANALYTICS The multiple barriers to driving AI in the analytics space (see Figure 2, right) areanalogous to the issues faced in leveraging other types of technology: from lackof quality (or ‘clean’) data, as well as integration issues, through to sponsorship But the different ways in which procurement organisations can improve theirspend analytics are also multiple. Moving from small data to big data will helpidentify spending patterns. Predictive analytics will spot problems such as Then, agentic AI will not only be able to analyse and recommend actions – it willtake them. For example, seeing that freight costs are above budget; analysingthe cost drivers and undertaking basic sourcing. Finally, multiple AI agents willwork together like a ‘procurement taskforce’; one updating cost forecasts; the 1. Data quality and availability2. Lack of domain knowledge and talent3. Integration and legacy systems4. Governance, risk and trust concerns5. Technology limitations (current AI models)6. Fragmented efforts and sponsorship AI for Spend Analytics: Unlocking Smarter Forecasting, Budgeting, and Benchmarking,Procurement Leaders, GEP, 2025 “Analytics is not merely a tool; it mustbecome an integral part of the businessfunction – where insight converges “Once you havestandardised data,the number of use Alex Zhong, senior director, global head ofproduct marketing, GEP The share of value gained from workflow-embedded data and analytics – more thanany other aspect of data and analytics RETHINKING ROLES TO MAXIMISE THE ROI OF AGENTIC AI To benefit from agentic AI, procurement must rethink how work is designedand managed. It’s not just about using smarter tools, it’s about reshapingdecision-making and how value is delivered. Many organisations are beginning Agentic AI does not eliminate the need for humans: it shifts their role fromdoing the work to managing how it gets done. Procurement teams will neednew skills to work with autonomous systems. These include reviewing AI- Procurement Data and Analytics Survey, Gartner, 2024From chaos to clarity: delivering data to drive AI-powered procurement, Procurement Leaders, 2025 “It is increasingly clear analytics cannotremain siloed; it must be embeddeddirectly into operational workflows.Technology alone accounts for only asmall share of the value – approximately ADOPT A HUMAN-CENTRIC ROLLOUT TOEASE CONCERNS OVER AI FOCUS ON FEASIBLE USE CASES GOVERNANCE AND GUARDRAILS Don’t try and ‘fix’ everything. Find use caseswith a clear ROI; build trust and progress to The cost of one bad decision can outweigh10 good ones. Make sure decision thresholdsare governed appropriately through escalation Don’t aim for fully autonomous today. Instead,look to create some initial value; augment TRACK AND COMMUNICATE VALUE EARLYIt’s important to tell stories. If value is invisible,then you lose the right to scale. DON’T AUTOMATE TASKS,REDESIGN PROCESSES TACKLING INTEGRATION AND DATABe honest about data readiness. If datais fragmented, you need to create anorchestration layer to seamlessly drive AI driven analytics is more than just a tool,it’s something that needs to be in yourdaily workspace. The only way in which ABOUT OUR PARTNER PROCUREMENT LEADERS Head of partnerships and digital engagement:Ian lawlessSenior editor:Peter EllenderArt director:Salvatore SpagnuoloWriter:Kevin Reed