Digital Analytics in Procurement-How decisions will be made in the future

Digital Analytics in Procurement-How decisions will be made in the future

Digital Analytics in Procurement-How decisions will be made in the future
 Digital Analytics in Procurement: How decisions will be made in the future 

Digital Analytics: Foundation For The Next Procurement Revolution

The race of staying competitive in the digitized world has only just begun. Within that race, the procurement function plays an important acceleration driver for the entire business. Digital analytics and the smart application of it will revolutionize the way procurement is done in the future, fundamentally impacting how CPOs manage their department and how decisions are taken.
At the foundation of every successful future procurement strategy lies the smart application of digital analytics, deeply embedded within the existing system landscape. The Value creation and all Enablers of the BCG’s Procurement House can only unleash their full potential if they are based on the smart design of a solid data foundation. As digital analytics detect patterns in existing data and predict future developments, they advance buyers to make smarter, better informed business decisions than ever before.  
Download Also:
Figure 1 – BCG’s Digital Procurement House  
Leveraging digital analytics creates value along all procurement value dimensions as shown above. As illustration, a global car manufacturer utilized an extensive comparison of planned ordering volumes with actual volumes for re-negotiations, especially where actuals exceeded plan, thereby cutting costs on relevant materials by 10%. As a second example, a global truck producer deployed an algorithm on part similarity across suppliers to identify parts as ‘similar’, enabling buyers to select the supplier with the lowest price, thereby cutting costs on specific components by up to 30%. 
CPOs now have the opportunity to create positive impact with digital analytics, supporting decision-making and delivering in the respective value dimensions of BCG’s Procurement House. In this paper, we focus on how smart leverage of digital analytics by using existing data can positively impact all Values and Enablers of the procurement function, and we answer the question of how CPOs can adjust their department for the digital future.

Four Simple Steps to Unleash Full Value Potential  

Any digital analytics model must be tailored to the specific company, as every business’s data structure and strategic focus differs. To unleash full value potential of digital in procurement, CPOs must follow four crucial steps.

1. Gather relevant data and create a holistic database
First, it is essential to gather all existing data. Moreover, potentially relevant external data sources like financials of suppliers or traffic data are included to execute more precise analyses. In the following, the data is brought into a single format and prepared for further processing by correcting format errors or excluding outliers, e.g. purchase order dates seemingly hundred years ago. Finally, all scattered data is integrated into one structured, holistic Data Lake for further development.  
2. Visualize data to fine-tune the scope
The next step is to generate hypotheses based on an initial visualization of the data. By using stochastic techniques like correlations and clustering, data scientists can prioritize variables and determine which should be ultimately integrated into the data model, e.g. if meteorological effects systematically influence timeliness in logistics. Clear hypotheses help to refine the problem set that ultimately lays the foundation for constructing the analytical model. Key outcome of this step is to have an alignment on the questions that need to be answered by the analytical model.  
3. Build the digital analytics model
With defined questions to be answered, the digital analytics model is developed by data scientists. The model construction cycle is initiated by selecting the model that best fits the data and desired outcome, like the demand forecast for a certain commodity. This is followed by the process of model training, where relationships are coded in order to “train” the model. In this process, the model learns e.g. how expected demand of the end product translates to needed purchase orders for parts of a commodity. The final step in the cycle is model evaluation, which helps to revise the model’s performance. The model is then iterated by running through the cycle until the desired model quality is reached. Lastly, the desired output format is selected to meet users’ requirements.  
4. Automate and customize the model
Once the data modeling has been completed, the algorithms to be used are automated. Moreover, a fixed database connection is usually established and a user interface for day-to-day use is developed. With a convenient interface, commodity buyers can e.g. immediately create an overview on how a changed demand forecast of the end product translates to volume changes of parts in their commodities. This interface can then display a ranking of most effective tender approaches or illustrate a selection of suppliers for certain part purchases based on the changed demand forecast, supplier financials, geopolitical risk data and other variables.

How a large food producer digitized its decision-making  

A global food producer intended to optimize its procurement process and reduce spend. The company had an excellent track record in the commercial aspects of procurement. However, it had no substantial understanding of IT to enhance the impact of traditional procurement levers, e.g., by applying advanced analytics to consolidate fragmented spend that comprised over 400 ingredients and 1.800 packaging types. 
It therefore brought in external data scientists to work with the procurement function, gathering and integrating data, resulting in a holistic database.  In the next step, the team jointly prioritized relevant data based on initial hypotheses, discussing generated insights with posters in a “war room.” In this process, first insights especially on saving potentials were identified, for instance on price-elasticity analyses, as shown in figure 2. Based on the generated insights, about 100 analyses with relevant variables were refined and the scope for further modeling was detailed.
Figure 2 – Volume vs. price increase: classical procurement analysis can now be applied to complete spend with real-time updates  Based on the prioritized variables, data scientists initiated the modeling process cycle. They back-tested the model by comparing historical results with predicted results for the same period until the desired accuracy was met.  The analytics model has meanwhile been successfully integrated into the procurement department and reduced spend by 8%. Furthermore, the supplier default risk has decreased, as potential supplier quality issues are much more transparent.

The Digital Organization to Ensure Long-Lasting change  

Businesses looking to revolutionize their procurement department need to consider implementing certain fundamental changes. Five major factors are relevant within that change.

Create a digital analytics group
In order for digital analytics to reveal potential in procurement, an advanced analytics team must be established. This dedicated group of data experts creates a holistic data view, connects data sources and extracts value from data lakes. By developing advanced algorithms, this translates in strategic buyers being more efficient and focusing on crucial business aspects. The advanced analytics team could be located as a supporting function within the procurement department, strongly interacting with buyers and other business units in daily operations.

Ensure data rigidity in processes
As digital analytics bring change throughout the procurement department, new processes must be set. This is essential to promote speed and cost-efficient handling of data and its models. Clear process descriptions must be defined for new procurement data collection, its quality requirements, and their integration into the analytics model. As an example, it’s crucial to define the process to capture all data for executed purchase orders including volume, price, agreed terms and conditions as well as to ensure their integration into models for improving future negotiation strategies.
Define a clear digital capability landscape
In order to effectively use data models in the current system landscape, businesses should focus on a specific team composition. We identified three main capability types in the procurement department of the future. The smallest group are Digital Experts, who account for around 10% of employees. These employees’ core competency is modeling and controlling data models, as well as developing advanced algorithms. 
Around this core, we see trained Users of digital technologies, who understand how the data system works and are able to interpret the output. By that, they can derive profound decisions that create value within their procurement category. Lastly, there will still be Operational & Supporting Staff, who have no fundamental knowledge of the digital system and how to handle it. Companies will have to upgrade their skills by procurement training programs to understand and execute selected tasks, fulfilling the new, data-centered job requirements.
Figure 3 – A digitized procurement department should have three capability types

Create an agile IT infrastructure
To guarantee high-quality data input, procurement functions must ensure comprehensive data gathering with subsequent data handling. Doing so, it will be necessary to extract data from different functions and sources and to consolidate the data into one procurement database. As final step, calculated output and extracted data must be included into an IT solution to fit into the existing procurement tool landscape.  Overall, even though it’s possible to harness considerable value with smart bolt-on niche applications, a more fundamental system integration and upgrade will become necessary as companies progress in their digital procurement ambitions.

Establish governance to monitor the change process
A governance body should be put in place to monitor the mentioned changes and early identify need for action. Furthermore, we suggest implementing performance indicators for procurement employees to better monitor individual performance and to customize trainings based on specific needs. These can include e.g. percentages of decisions made that are based upon digital analytics’ recommendations.

A successful transformation story leads by example
A European-wide oil and gas company introduced advanced analytics. Two main goals have been to improve quality and risk prevention. In regards to quality, a dedicated program was started that focused on instantaneous data availability. Users in sites were enabled to provide direct feedback on supplier quality by an application installed on their smartphone. 
Suppliers on the other hand regularly received a transparent evaluation of users’ feedback and were incentivized to improve quality compared to a jointly defined baseline. Eventually, overall quality performance improved by 38%, as measured by a lower PPM. For risk prevention, three dimensions were identified driving bad quality and non-delivery of suppliers:

1)Financial pressure/distress of the supplier  
2)Disruptions in the supply chain  
3)Delivery shortages caused by a suboptimal, vertical organization of the supplier

By applying advanced analytics to a comprehensive dataset, the main risk drivers could be clustered and detailed in these dimensions. Through ongoing analyses, high risk suppliers were identified, so only c-parts were sourced from them. As a result in an internal regular survey, the perceived risk exposure to suppliers sank by 16%.

To achieve these results, the company needed to transform its procurement department along the five previously defined factors, starting by establishing a direct report to the CPO for Research & Analytics. As next step, processes for procurement data handling were defined. Part of the new performance management was the way employees passed on data to the Research & Analytics group, backed by an individual performance-based bonus for clean data handling. 
Addressing capability types, new profiles were defined especially for the Research & Analytics unit, to manage procurement data, specialize in the application of analyses and develop advanced digital analytics tools. Implementing these fundamental changes ensured lasting improvements along the value dimensions in scope.

How to Navigate the Jungle of Possibilities to Unleash the Full Value Potential  

We believe that procurement functions that successfully implement digital analytics within their existing system landscape must follow a “think big, start small, grow fast” mindset. The foundation can be laid by integrating digital analytics into the existing system landscape. However, the entire procurement department will have to be adapted to keep on track in the digital age to come. Overall, decision-makers have to seize the opportunity to unlock the hidden potential of their existing data that can be leveraged to elevate their entire business towards a sustainable future in the digital age.
Related Topics:
The Author: Daniel Weise

                                         Daniel Weise
I am a procurement enthusiast and have the privilege to lead BCGs procurement business line globally. Supporting my clients globally and across industries, I focus on value delivery - beyond cost and including resilient supply chains and sustainability, operating model redesign and digitization programs. 
I have also supported many of my clients in PMI and restructuring settings. Recently, I have published my first book summarizing my experiences in digitizing procurement functions: "Jumpstart to Digital Procurement". In BCG, I am also part of our global Operations Practice Area leadership team.

Post a Comment

Previous Post Next Post