Can Graph Analytics Solve The Problem of The World’s Messed Up Supply Chains?

Updated: Mar 14
By: Admin



When the cargo ship Ever Given got stuck in the Suez Canal in June 2021, insurers estimated it was holding up an estimated 3.3 million tonnes of cargo an hour. The final cost was put at $60 billion.

That's the damage that a single stranded ship can do. But, of course, this incident happened during a global pandemic. The wreckage wrought by COVID-19 now makes the Ever Given look like, well, a drop in the ocean.

The $4 trillion price tag of the supply chain crisis
Earlier this year a report by The Economist Intelligence Unit and GEP revealed that supply chain disruptions might have caused up to $4 trillion in lost revenues. And yet COVID-19 was not the only disruptive factor. The report also highlighted other forces such as cyberattacks, commodity price fluctuations and regulation. These numbers are so big they can be a bit meaningless.


So, let's consider the impact on individual businesses. Take Asos, the British fashion online retailer. It announced in October that its profit could slump next year by more than 40% due to supply chain pressures. It sounds bleak. And it is. But there is some good news.


Other companies have weathered the supply chain storm with negligible impact. Levi Strauss is a good example. It says supply chain issues shaved just $10 million off its annual revenue of $1.5 billion.

Why did Levi Strauss achieve this? Largely thanks to diversified manufacturing. The company decided years ago that it would not source more than 20 percent of its product from any one country.


It also committed to “cross-sourcing” across two or more countries so it can shift production when necessary.


“Our supply chain really is a source of competitive advantage,” CEO Chip Bergh told CNBC. “We can move product around with a lot of agility. ... We’ve been running the business against different scenarios for the last 18 months.”

Supply chain modelling: Has the time come for graph analytics?
Clearly manufacturers, distributors and logistics companies need a more agile way to manage the vast network of components needed to move items around the world. Solutions built on graph theory offer a particularly compelling solution.


Graphs represent any network of relationships in a mathematical way. A supply chain graph, therefore, will plot suppliers, transport providers, warehouses, components, services and finished products as a series of vertices and edges. Analysts can use these graphs to create a digital twin and track all parts through their lifecycle – from raw material to final destination.


At present, most companies do not use graphs to scrutinise their supply chain resilience. Typically, they employ more rudimentary tools. Some rely on various ERP systems or even use simple spreadsheets held in departmental silos. Others might bring in data scientists to extract insights from a variety of data sources. This is a more sophisticated approach, but it is still unsuited to modelling the sheer interconnected complexity of a modern supply chain.

Introducing the LynxKite open source tool



Here at Lynx Analytics, we have deep experience of using graphs to solve supply chain issues.


In 2014, we launched a dedicated graph tool called LynxKite. It scales to billions of edges (thanks to its underlying Apache Spark cluster computing engine), comprises a friendly graphical interface and a powerful Python API. In 2020, almost 16000 commits later, we made LynxKite open source to accelerate
its adoption.


Today, we use our expertise with LynxKite to help data scientists at a diverse range of companies. The goal is not to give real-time visibility into supply chains, but to map them – and then use these maps to model scenarios, create simulations and explore potential impacts and outcomes.


In so doing, we can answer questions such as:

  • How will a shortage of a particular raw material affect production?
  • What are the trade-offs and impacts of using alternate routes?
  • Should we use ships with different cargo capacity or air freight for some products?
  • What is the impact of a particular border closing?
  • How quickly can we adapt to a rise in demand from one region and a fall in another?

Of course, graphs can be used to do more than merely plot the potential impact of supply chain disruption.


Logistics managers can also use them to make their networks more resilient in the long term. For example, a company could put systems in place to make it easier to switch to shipping more profitable items or use alternate transportation methods in the event that a route becomes blocked or unreliable.

The world's biggest companies are already doing this kind of work. Their businesses depend on it. But most smaller players are still in a world of home-made tools, spreadsheets and SQL queries.


It doesn't have to be this way. If you would like to know more about LynxKite and how it can help you resolve your supply chain issues, you can contact us here.