Most chief procurement officers (CPOs) recognize that data is instrumental – yet the reality is that many CPOs don’t leverage data as much as they could. From defining advanced procurement strategies, to working with suppliers more efficiently, or developing predictive models, data is a critical differentiator for procurement leaders.
In our work, we have encountered many use cases demonstrating how advanced data management can offer CPOs practical advantages. In today’s unpredictable world, leveraging data gives CPOs a vital means to understanding the impact of market volatility and sudden shocks (such as pandemics and geopolitical turmoil) on external spend. For example, over the past few years, advanced data management has helped CPOs better appreciate raw material pricing trends and their impact on supplier contracts which can put the cost base at risk (especially when tied to “no index” agreements). Data enables CPOs to quantify the financial risk for categories to which the company has high exposure, share more reliable financial insights with their chief operating officers (COOs) and chief financial officers (CFOs), and develop strategies to mitigate risk.
Good data management also offers advantages in soliciting a request for proposal (RFP) process. One automotive supplier we worked with used data to better assess their costs in areas such as production costs and stock-keeping units (SKUs). The company combined three data sources per SKU into a data model: current actual costs obtained from site visits of their own factories, RFP quoted price points from various suppliers, and past purchasing prices. This allowed them to leverage up to 25 data points per SKU, enabling them to improve their knowledge on should-costing and putting them in a stronger position in negotiating with suppliers.
Top-notch procurement data management
Access to reliable data is critical, but it takes effort to overcome common pitfalls. Good data management practices cannot be implemented overnight, and many organizations struggle with the incorrect perception that data is the sole responsibility of the IT function. Likewise, traditional category managers see effective data management as a pain point. Transforming such attitudes involves a great deal of change management across multiple departments.
Another problem we see is the tendency to “boil the ocean” where teams request data for every spend item, which is not only unnecessary but is also impossible in the short term. Instead, procurement leadership should take a more selective approach – understanding which use case they need data for and where the data can be sourced. Some data will be held by procurement, such as prices and rebates. But some data may reside with other teams (such as the accounting department for customer billing data, engineering department for product specifications data, or external parties such as market volatility data). Here, it is also crucial that the data be usable, be formatted correctly, and be linked to the data that procurement already has.
It’s therefore essential that CPOs have a clear vision, specifying the use case and defining, with their teams, which data is needed. Governance plays a huge role, as data cannot be used effectively if it isn’t maintained properly. Likewise, talent can prove decisive when it comes to good data management: skilled resources who understand data and its importance within the procurement function. Therefore, CPOs should be prepared to invest in the skills they need to run the procurement department of tomorrow. Take the role of buyers, for example, where having solid competencies in data management will enable them to better leverage their organization’s data to spot where the best value lies.
Data lakes: What CPOs need to know
Procurement leaders need to ensure that their data is well-organized and can be easily accessed by their teams to make sound decisions. In our experience, data lakes can offer an enormous tactical advantage to procurement leaders. Firstly, these offer flexible solutions that can integrate unstructured data (such as images) with structured data stored in warehouses. Secondly, data lakes can help business leaders extract more insights from more data and sources, which can support strategic thinking and lead to more agile, data-driven organizations.
Data lake implementation: Best practices
Data lake implementation can be fraught with difficulty if not approached correctly. Indeed, this has sometimes put off companies from implementing data lakes. That need not be the case. Most of the hazards can be avoided by following a few best practices.
As with all cross-departmental projects, the first challenge is to define the right approach to implementing the data lake, including what sort of data lake would be most beneficial.
1. Planning the data lake
Frame the impact in terms of strategic goals: Accurately quantifying the return-on-investment (ROI) of a data lake can take more time than its actual implementation. The case for implementation is better made by framing the impact in terms of how the data lake can help achieve strategic goals. In the case of one procurement leader, for instance, it was demonstrated that the data lake would enable the company to achieve its goal of providing customized healthcare to all its patients by 2025.
Ensure a multi-disciplinary team is involved in its planning: Ensure shared ownership of the data lake; don’t let it become the “pet project” of IT. To this end, it is best practice for the planning team to be comprised of all the key stakeholders close to the data lake process. The team should include both IT team members along with other key stakeholders, such as those from procurement. The defining characteristic of the planning team should be that it includes members who have both technical skills (such as data science) and content expertise.
Make data lake management a cultural exercise: Data lake implementation requires changes in how data is used and collected. For that to happen, it is essential to secure buy-in from the procurement staff beforehand. It’s vital that the staff understand what data lakes are and how they work. This understanding will enable them to better appreciate the benefits that will ensue from its implementation and will help ensure that they continue feeding procurement data into the data lake once it is implemented – critical for the data lake to continue yielding timely insights.
Plan time expectations: As with all cross-functional projects, it is important to manage expectations regarding the implementation timeframe. Getting a data lake up and running is likely to take between 12 and 18 months. Knowing this at the outset can help pace the project.
Ensure you have the necessary talent on board at the outset: To ensure smooth implementation, it is imperative to have all the necessary talent on board to manage the implementation, including those who have the knowhow to access and manage the data, such as data scientists and data engineers. If the company already has a well-resourced IT department, an effective approach can be to develop dedicated data lake talent from within the organization.
2. Activating the data lake
Ensure the content and objectives of the data lake are clear: Ensure staff are on board. If it is believed that the implementation of the data lake is merely a matter of reducing headcount, then its implementation is unlikely to gain full cooperation. It is essential that the real purpose and objectives in enhancing data management are communicated clearly to all those involved.
Embed agility to ensure sustained impact: The chosen data model needs to be both agile and flexible. Procurement’s data needs are likely to be very different in the medium- to long-term horizon vs. what they are today. Consequently, the data model needs to offer the flexibility to enable the company to add new data sources further down the road.
Scope out all the key data requirements prior to implementation: It is vital to get the basics right. This includes defining the scope of the project and identifying all the required types and categories of data prior to the start of data lake ingestion – it is surprising how often this isn’t done. Not doing so creates significant risk. Data ingestion is an “all or nothing” process. While it is possible to adjust the scope as you go, much time and effort can be saved by getting it right at the outset.
Phase implementation: The key to successful data ingestion is to carry it out in several different phases or stages. The best approach is to start with one or two specific categories of data (such as indirect spend). Each category of data will then be fed into the data lake at the appropriate stage in the implementation. This will ensure that the data is complete for that particular category. If multiple data categories are loaded simultaneously, all are likely to be incomplete.
3. Maintaining the data lake
Develop resources for data lake management across procurement: To ensure that best use is made of the data lake, it’s important to develop specialist knowledge and expertise across the procurement function. This can include locating data specialists within procurement itself, as well as identifying data stewards in each department. Their role includes monitoring and enhancing data lake use and performance, and continuously maintaining the data lake (periodically updating or removing outdated data, ensuring appropriate metadata is captured, cleansing data).
Implement rules that instill data discipline and strong governance: Once the lake is up and running, it will be necessary to ensure that users adjust how they access it. In addition to feeding the data lake with fresh data, behavioral change will be required to ensure that all procurement staff are only looking at data through the analytics application that sits on top of the lake and are not continuing to store data locally on their desktops. Following the data lake implementation, it is also necessary to update the data governance operating model (to clearly document data lake management roles, responsibilities, and requirements) and to have management enforce data governance standards to support maintenance, so as to avoid turning the lake into a swamp.
When properly set up and maintained, data lakes can open the door for significant untapped opportunities for procurement, not only via cost savings but also by providing the CPO with a comprehensive and accurate overview of the available data. The increased transparency can play a critical role in helping companies gain strategic insight. A data lake can be a way to achieve long-term impact very quickly. CPOs who realize this and who understand the value their organizations can bring should focus on data, guide their teams towards internalizing its importance, and ensure to onboard the right talent to activate it. Be it supplier performance management, risk management and spend analyses for category management, or sustainability, supplier relationship management and should-costing, efficient data management is the key.
Thierry Ajaltouni also contributed to this article.