Gradually, the decision-making systems used for purchases incorporate a predictive dimension, so that they are no longer just reacting but also in anticipation. By relying on simulations and forecasts, the function can adapt its strategy well in advance.
To act with a head start, anticipating market changes or changes in demand, purchasing authorities understood that they had to look to the future. And so add a predictive dimension to their decision-making systems. While the projects initiated by a few large groups in this field are still far from having achieved their objectives, interest in the subject remains strong. Provided you master its contours, since predictive analysis encompasses multiple concepts, around two main approaches: simulation and forecasting.
In the first case, it is a question of testing different scenarios, playing on the possible options, to measure their impact and make the best choices. In purchasing, the simulation allows to calculate the potential gains and adjust the strategy, as part of budgeting, but also to plan the activity from an organizational perspective, to know what timetable to adopt, which buyers to mobilize, etc. The more advanced forward-looking approach follows the opposite of reasoning: relying onartificial intelligence, it aims to identify future situations in order to prepare for them in advance. If the forecasts apply particularly to financial information (currency prices, price changes, etc.), everything is possible, especially in terms of supplier risk management. Purchasing management, however, has an interest in limiting the fields of investigation, instead positioning itself on subjects with relatively short-term impacts, in areas where it can find alternative solutions.
Before embarking on the predictive, several steps are required. Having a system of expenditure analysis, to know what is consumed and by whom, by using which suppliers, is an essential prerequisite. What good would it be for a company to look to the future if it does not already know what it is buying? It is then necessary to identify all other potential sources of information and extract relevant data, possibly using knowledge management solutions, particularly automated monitoring tools or search engines. The use of datamining software to sort or establish correlations can also be very useful. Not to mention the big data-oriented decision-making systems which are beginning to multiply in the market, capable of processing large volumes of data of any kind, in very short time frames.
This complexity explains the timid progress of the purchasing departments in this field. The most advanced use the tools made available by finance, which allow to do only part of the work. This is often the decision-making module of the company's ERP, sometimes enriched with specific developments. The use of dedicated brick e-purchase solutions is much less common. It must be said that decision-making, let alone predictive analysis, is a project that goes beyond The purchasing function, with technical constraints in terms of data extraction and rules programming, which requires the intervention of the CIO. Even statisticians or data scientists, a profession in the making.
To support purchasing directorates, most business publishers, at least those already positioned on decision-making (B-Pack, IBM-Emptoris, Ivalua, SAP-Ariba, SynerTrade, Zycus, etc.), ensure that they are working on the subject. More often than not, they approach predictive analysis first through budget planning, for example, to propose six-month or one-year spending forecasts. In seeking to go further, they fear not only to move away from their profession, but also to confront hyper-specialized tool publishers, much more experienced and recognized in their market. Until e-purchase solutions are up to standard, the use of general decision-making solutions (IBM Cognos, SAP Business Objects, Oracle Hyperion, etc.) remains preferable. Especially if the project is limited to modeling, simulation or projection to identify major trends. However, more functionality is essential to know and anticipate future events.