Over the next two years, artificial intelligence and other data-driven techniques will change the public sector procurement workforce significantly.
Beyond this bold statement, European public procurement has traditionally lagged behind when it comes to technology, but do new data techniques offer an opportunity for things to change? Artificial Intelligence, as well as other data management techniques, have already proved capable of improving government operations and meet the needs of citizens in several ways, ranging from traffic management to processing tax forms. However, European public procurement institutions are still struggling in harnessing these powerful technologies.
A smarter public procurement endorsed by governments
The benefits of relying on data to enhance public procurement can no longer be ignored. Indeed, several European government reports have already highlighted the need to leverage data for better decision-making processes. However, some countries seem to be ahead of others.
In the UK, the House of Lords Select Committee on Artificial Intelligence was told that the use of AI across government departments and the public sector could save an estimated £4bn a year and “enable more informed policy decisions”. In its report[1], the Committee said the government’s annual spend of £45bn on goods and services gave it an “immense power in encouraging the adoption of new behaviors and practices in its supply chains”. The report also said that the Brexit provides an opportunity to alter EU public procurement rules "to ensure that these rules and thresholds benefit businesses in the UK, in particular when it comes to public sector procurement and the stimulation of a fertile AI development sector, as long as it is still a competitive process".
In Belgium, the committee called AI4Belgium has highlighted the need for Belgium SMEs and Public Procurement to closely collaborate around projects related to AI and Procurement. This report[2] invites Belgium authorities to “Redesign public procurement processes to enable trial and error, not excluding young organizations”.
In France, Deputy and Mathematician, Cédric Villani was given the task to analyze how AI could help public procurement enhance existing workflows. In his report[3] called « For a meaningful Artificial Intelligence: Towards a French and European strategy », he promoted the idea of AI solutions to further improve French public procurement.
Despite these reports, new data techniques adoption rate is still low in European public procurement administrations. We believe that internal work must be accomplished in order to change the mindset of most buyers.
Innovation is still not a priority
Based on a series of interview and own expertise advising public procurement decision-makers, we have concluded that the overall benefits of data techniques are more or less accepted. Indeed, decision-makers do realize that these techniques can enable them to save and gain efficiency when it comes to buyer’s workflow management. However, the majority of those we have interviewed have declared that out of the 250 000 public authorities in the EU, a vast majority do not consider innovation as a priority.
Some of them have doubts related to the overall financial health of so-called startup data companies, as well as skepticism regarding the solutions presented to them. From the startups, we noticed a real lack of knowledge related to public procurement processes.
Obviously, public procurement decision-makers can, through sourcing steps, spend time looking for what the market has to offer a given issue. And yet, this task is still very tedious and buyers are often running against time on some kind of just-in-time workflow. These low added-value tasks can be automated through data-driven solutions.
Furthermore, sourcing within the EU single market is still an uncommon practice for the public buyers. Cross-border penetration in public procurement may be dramatically boosted by new AI (NLP) technologies whereas it doesn’t exceed 6% of the overall EU award value (Source : Border Effects in European Public Procurement[4], Measurement of impact of cross-border penetration in public procurement [5])
We firmly believe that it is crucial to promote and accelerate the dialogue between public buyers and data companies. At the level of the EU, the purchasing power of public buyers’ accounts for around 16% of the European GDP (source: Benefits of Modernized European Public Procurement). Public buyers have the possibility to boost innovation among existing market players, but also provide opportunities to SMEs and startups who may have developed AI solutions to satisfy unmet needs but face difficulties in bringing them to the market.
« Improving public procurement can yield big savings: even a 1% efficiency gain could save €20 billions per year. » (source: ec.europa.eu)
Based on our expertise, public procurement remains insufficiently oriented towards innovative purchasing. The lack of culture when it comes to innovative purchasing is certainly an issue, as well as legal risk aversion in exploiting current regulations, and operational risk aversion in purchasing innovative solutions. Indeed, the purchase must meet a need of the public authority and is subject to an obligation of result, which in turn is passed on to the contract holder.
The European Union is openly promoting the creation of an innovation procurement framework to help close the gap. A comprehensive policy framework that provides vision, strategy and appropriate means is essential for the transformation of European public procurement. Through a more decisive role of public authorities and an increased R&D procurement budget, it would become possible to motivate the adoption of innovative procurement solutions. When it comes to investment, European countries are not as active as other countries. Indeed, as mentioned in a European report[6] “authorities around the world have set targets to direct a percentage of their public procurement budgets to research and development and innovation. For example, the US strives to spend at least $50 million on research and development to help accelerate the adoption of innovative procurement solution.”
The US is also an interesting example through their “cloud-first” policy[7]. Indeed, the US government is promoting the use of innovative procurement tools and new talent development models to help buyers use these new data-driven solutions. Buyers can rely on a living document called “TechFar” to enable them to better understand industry best practices, initiatives to simplify procurement procedures and feedback from internal and external stakeholders. When it comes to data management, Chief Information Officers are invited to identify “must-move” services and create plans to migrate them to the cloud. Each US procurement agency is being monitored based on their performance related to the “cloud-first” policy. The goal is to reach specific KPIs for each public procurement organizations.
A last worth-mentioning example of innovative procurement policy is in South Korea. The country has adopted a bold plan to spend 5% of its public procurement resources on developing and 20% on deploying innovative procurement solutions[8].
Current status within Procurement organizations
Beyond the difficulties, a limited number of public organizations have started opening to innovative solutions. So far, most of what we are seeing is either aggregating data or speeding up a process. At CKS, we have developed a unique tool capable of helping public procurement managers better organize and supervise the work of buyers. By leveraging data visualization techniques and tailored-made algorithms, our solution can anticipate the workload required depending on the project and make recommendations to help managers allocate work between buyers. In a very near future, we expect to leverage machine learning algorithms to predict and automatically create a full work schedule per buyer.
Indeed, Machine learning (ML) is well-suited to procurement because it spots patterns in the large volumes of data generated through purchasing and then forecasts future trends. We believe that ML is particularly suited to the public procurement since it is in the public sector, where supply management professionals are tasked with identifying critical cost-reductions, that the automation and granular insight of machine learning is particularly synonym of added-value.
It is key to remember that ML doesn’t work unless it has a vast number of accurate data. Public procurement will need help from data experts to structure their data. Even though, we are seeing a legitimate will to promote an open data culture, the public sector has been slow to improve how it uses data and, going forward, public procurement leaders will have to prioritize data quality.
Another important element slowing down the adoption rate of new data-based solution is the issue of sovereignty in cloud-based tools. Public procurement organizations do not want their data to be processed and ultimately belong to companies complying with foreign rules. As such, companies willing to collaborate with public organizations must comply with some rules for data usage. Moreover, data is being viewed as a national asset that countries do not want to share. National leading providers, like Outscale in France, have appeared to close the gap between foreign cloud solutions and national ones but this is still a work in progress. For instance, AI-based solutions often rely on third-party serviced provided by Microsoft (Azure) or Amazon (AWS).
At CKS, we have understood that while public sector decision-makers are increasingly aware of the decisive challenge of data for advanced data techniques embedded in new solutions, the data is often neither accessible nor discoverable. Consequently, we have decided to build our solutions by partnering with Forepaas, a leading data end-to-end platform (PaaS) that enables us to structure data coming from different sources, both internal and external, and create procurement-oriented unique KPIs that can be displayed in an easy-to-use dashboard thanks to data visualization techniques. We aim to deliver a decision-making tool that can represent a meeting point for different public decision-makers such as procurement, financial or controlling directors. Public sector officials may also lack the appropriate knowledge and expertise to make strategic buying decisions for AI-powered solutions.
Data-driven solutions to help public procurement
A lot can be said on how new data-driven techniques will impact public procurement. Let us begin with government procurement contracts.
1/ Tendering, Contract Management and Risk Assessment
Buyers know that they are complex and that public-sector processes are special in their requirements. State and local entities must adhere to a very different set of rules, regulatory requirements and other elements that set restrictions on government contracts.
Through the use of Natural Language Processing (NLP, a subfield of AI) based solutions, it becomes possible to remove most of the manually intensive tendering and contract management tasks. Applying automation can accelerate any number of manual tasks present throughout the entire tendering and contract execution processes — from the identification of requirements to creation and approval of key contracts, to their possible renewal. Not only does public procurement from better streamlined process, but with assisted renewals based on data, cost optimization becomes easier.
In addition to eliminating redundant tasks in the tendering and contract management processes, NLP can also assist the public buyers manage their exposure to risks. For instance, by automatically identifying inconsistencies between the tendering documents or assessing financial risks in negotiated agreements.
« The ultimate goal through smart contract management combined with NLP is to enable the buyer to spend more time on added-value tasks»
Moreover, we can envision the development of an AI that could analyze and flag all potential issues in a given contract before signature or reduce the time required to reach contract signature through cycle time prediction based on similar agreements, as well as recommend clauses to negotiate based on the supplier’s data.
This analysis based on different data sources is highly relevant given the key role of compliance in public procurement tasks. We have identified a real interest from decision-makers to gain visibility on the entire procurement process through the available data. At CKS, we also envision a future in which a data-based solution can suggest a scenario for awarding to the buyer. This would be beneficial when the buyer has to choose between several offers.
2/ Enforcing supplier obligations and better manage buyers work
Data-based solutions can also help manage, monitor and enforce supplier obligations, as well as make the public contracting requirements easier for suppliers to meet.
Indeed, by providing public procurement teams a way to track their progress toward the percentage of work accomplished, most buyers can experience a great improvement in the visibility and reporting of supplier activity and costs compared to a contract previously analyzed. We could foresee a system monitoring the supplier relationship and performance throughout the entire contract life-cycle and alerting the buyer when necessary by providing recommendations.
A two-steps transformation
The transformation of the public procurement will go through two steps following a technological maturity trend. The first goal is to deliver tools capable of predictive analysis. The idea is to predict market trends, orders, payment and awarding/supply delays, and thus assist buyers in their decision-making. This analysis will be customized according to suppliers, etc. The use of algorithms and big data is mandatory to reach this level. CKS is currently ready to assist organizations at this level with a proven track record.
The secondary step is called prescriptive analysis. This type of analysis based on ML algorithms will allow public procurement decision-makers to suggest different decisions at key points in time, such as taking advantage of a current/future opportunity or mitigating future risk.
For instance, a buyer might want to know what would happen if a supplier has failed to reach a specific KPI or the consequences of changing supplier in the future. In other words, it will illustrate the implications of each decision, based on past decisions and market trends. If the technology is also able to learn how buyers works, it will also be able to better understand their needs and how to better meet them. We are trying to create the “augmented buyer” rather than replacing them with machines. It will be useful at key moments when the user is open to suggestions. CKS has started building prototypes and gain experience through partnerships.
The need to train buyers
In our opinion, there is little choice but to train the public buyer in the particularities of data-based solutions and their potential risks (regulatory, ethical, etc.). As with IT, which is constantly changing and where the balance between real breakthrough innovation and hype marketing is difficult to understand, it will be challenging for buyers to assess the relevance of offers.
CKS will soon share the experience being accumulated thanks to our current pilot projects. Stay tuned for more information on innovative public procurement.
[1] https://publications.parliament.uk/pa/ld201719/ldselect/ldai/100/100.pdf
[2] https://www.ai4belgium.be/wp-content/uploads/2019/04/report_en.pdf
[3] https://www.aiforhumanity.fr/pdfs/MissionVillani_Report_ENG-VF.pdf
[4] https://ec.europa.eu/docsroom/documents/26201/attachments/1/translations/en/renditions/native
[5]https://publications.europa.eu/en/publication-detail/-/publication/5c148423-39e2-11e7-a08e-01aa75ed71a1
[6]https://ec.europa.eu/digital-single-market/en/news/commission-notice-guidance-innovation-procurement-available-now-all-eu-languages
[7]https://read.oecd-ilibrary.org/governance/public-procurement-for-innovation_9789264265820-en#page170
[8] https://ec.europa.eu/transparency/regdoc/rep/3/2018/EN/C-2018-3051-F1-EN-MAIN-PART-1.PDF
