Boosting Supply Chain Insight with Neo4j
SaaS for proactive procurement provider Scoutbee needed to give clients in-depth insights into suppliers. By using Neo4j knowledge graphs, business users and clients are able to access and visualize complex, real-world supplier interdependencies, enabling them to select the optimum supply chain partner.
By the numbers: Scoutbee Supplier Graph
- Supplier discovery: 75% faster
- Working hours: 85% fewer
- Platform: Neo4j AuraDB Enterprise on AWS with Neo4j Bloom
Global procurement services company Scoutbee operates in the procurement and supply chain industry, offering supplier intelligence and discovery solutions to help companies improve their procurement processes.
Scoutbee’s AI-powered supplier data platform helps clients build a resilient supply chain. To do this, it’s important to not just identify new suppliers, but to understand how existing and potential suppliers operate in the context of the entire supply chain.
Scoutbee’s challenge was to provide intelligence insights that would support clients in making the best data-driven decisions.
Nischal Harohalli Padmanabha, Vice President of Data and ML at Scoutbee, said: “Having a whole lot of data is great, but if you don’t find a way to understand this data and connect the dots by bridging the data and business language/vocabulary, it’s just overwhelming and, to a certain extent, not very useful.”
Some of the manufacturing giants that Scoutbee works with such as Unilever, Siemens, AUDI (to mention but a few) have data spread across as many as ten ERP systems. On top of that, data comes from third-party sources outside the organization. The data is heterogeneous in nature, and there are concerns about data governance, privacy, and security.
Another issue that arises with data that’s spread across different systems is that there’s no single source of truth. If this is lacking, it leads to reduced trust in using the data in decision making.
“We wanted to centralize all of that data and provide it with a data fabric by applying ontology to the data coming from various sources and domains, thereby creating a knowledge graph. We couldn’t go from zero to hero in just one step, so the strategy for the second half of last year was for us to consolidate all of this data, to eliminate data silos, and get all of the applications and products to consume data with a semantic layer from one central place,” said Nischal.
Scoutbee’s goal was data democratization, building an organization where people working in different departments could all access data insights and trust the data powering those insights. To achieve this, the company implemented a strategy to leverage all possible data connections to provide better supplier information, richer market information, to discover more alternative suppliers, and to make supplier collaboration easier.
Scoutbee’s goal was to provide a data foundation, knowledge as an operating system, to answer complex questions in the field of procurement that go beyond understanding supplier profiles to understanding supplier interactions past tier 1.
Scoutbee uses AWS and had initially considered other graph database technologies. However, after examining the options more closely, Scoutbee chose Neo4j. Nischal explains why: “One of the biggest reasons for us to start working with Neo4j was its visualization capabilities in Neo4j Bloom. We wanted to build the capability for a larger part of our organization to understand the data, to understand the entities, understand the graph, and navigate it by themselves.”
For Nischal, rather than restricting data usage to data engineers and data scientists, interactive graph visualization would open up the knowledge graphs to business users in the organization.
“Neo4j graph database tech lets us build knowledge graphs that everyone can use. Neo4j has helped us reach our goal of data democratization,” he said. “The solution connects the dots between business users, data scientists and data engineers, while maintaining compliance. That way, everyone gets the best out of the data.”
With Scoutbee’s data democratization goal, it is important for Scoutbee to have private, secure, and scalable network connections to the Neo4j database server from other AWS services without exposing their traffic to the public internet. Scoutbee decided to use AWS PrivateLink to transfer Scoutbee’s critical data like supplier information across different systems to enable simplified network and firewall management rules and reduced data output and NAT costs.
The solution also provides the explainability that clients demand. “Not only do we want to answer our customers’ questions as quickly and accurately as possible, but we also want to provide the reasoning behind our answers,” said Nischal.
As a fully managed graph database, Neo4j AuraDB Enterprise frees up the team to focus solely on their solution: “We don’t have to manage our own databases, we can rely on Neo4j to help with performance tuning, take care of backups, restore them, and provide that support in terms of growing our team and leading them on their journey.”
“There’s also a huge community around the Neo4j solution, which is something that’s very useful, especially if you’re writing complex queries or you want to understand how to do domain modeling. There’s a whole lot of value in terms of support for working with graphs once that data is in there,” Nischal continued.
Scoutbee is now able to offer a unique value proposition to the market, providing intelligence on suppliers as they exist in the real world with semantic relationships to data from various domains. “We offer 360-degree supplier views, where we’re not just looking at a supplier from one data perspective, but bringing in all the data we have on a supplier, revealing the entire spectrum of relationships and the interconnection between them,” said Nischal.
Knowledge graphs unlock both graph analytics and machine learning that accesses the wisdom of the knowledge graphs: “Our customers see this as a high value proposition. It’s not just about the problems we can solve for them right now. It’s also about revealing considerations they’re not aware of that might open up in the future by going up the data value pyramid.”
Traditional supplier discovery processes take approximately 100-180 working hours over 24 weeks. Now, with the new smarter, automated approach, the process takes 8-12 hours over six weeks.
“A big win for us is iterative data modeling. Once products and applications are set up to work with a schema, it can be challenging to change the schema to adapt to evolving business needs. Iterative data modeling addresses that issue,” said Nischal. ‘When you start to work with graphs, you’re building a system for master data management. Data integrity, compliance, and governance are key too. This solution offers data integrity, displaying data sources and updates clearly.”
Now, with better data, Scoutbee’s clients can make better decisions, and this leads to a better world where companies supply the best possible products.