In the procurement domain, I feel like Spend Forecasting is the most underused analytical component. While everyone agrees that it would be wonderful to predict the future costs, few are willing to implement proper tools for it. Perhaps we feel that ‘predicting the future’ is too magical of a concept to be taken seriously.
Admittedly, telling the future for certain is impossible, but perhaps our approach in this case is wrong. We don’t really have to master the future right away to start seeing benefits from a proper forecasting process. The prize is not a magical gold chest at the end of the rainbow, but spread out across the road there.
Starting the journey itself offers maybe the biggest pay-outs. Most organizations probably are doing some form of forecasting on Excel-based processes. While it’s a good first step to get started, the problems with the approach are evident. These forecasts are hard to update and maintain, collaboration across regions and units is difficult, and the numbers themselves don’t elicit trust.
This is where the low-hanging fruit is. A harmonized and centralized forecasting tool and process can bring about the two major prizes: speed and trust.
Trust between procurement and other parties is something we in Sievo feel strongly about. So strong that we’ve even written a whole book about it, called Procurement Loves Finance. For finance to trust the forecast, they need to trust the data. Therefore the forecast needs to be built up from shared and agreed ground truths. Robust integration to internal (such as ERP / MRP data for accurate production volume forecasts) and external (commodity index data) sources will build up the basis for the forecast automatically. A process and data that is accepted and trusted across the functions enables us to skip the bickering between parties and make decisions as a single company that can be easily justified by pointing at the shared truths.
The second benefit comes from the speed and reduced effort of generating forecasts. Automatic integration to sources, without manual efforts and with periodical refreshes – a machine-assisted forecasting process that handles the mundane data operations automatically frees up time and effort so we can actually start focusing on the most impactful items.
This is already a big step. Without even having to obtain a crystal ball yet, we can create fairly reliable forecasts, just based on simple available data. Just estimating our own costs based on contract prices and lengths, along with category manager estimates of future trends, is already very impressive and gives valuable inputs to our stakeholders on strategic decisions and budgeting processes.
After this, the journey is endless. We can take the process to the next level, and plug in more detailed cost structures and do commodity price development-based scenarios. We can employ predictive analytics and machine learning to improve our forecasts for the ‘tail’ spending, and get ourselves to focus on the most important items. At this stage, we can start talking about being proactive, and actually impacting future profitability. This is our ambition in spend forecasting, and we are excited for the journey ahead.