Skip to content

Latest products and reports

Adopting predictive analytics in responsible sourcing

Adopting predictive analytics in responsible sourcing

Think big, start small and get the data right!

The hype surrounding predictive analytics continues to grow as business leaders recognise the potential this type of analysis has to unlock hidden value and reveal hitherto unseen risks. Data science tools and techniques are rapidly moving out of specialist silos and into the day-to-day operations of most business functions, including responsible sourcing.

Predictive models can help identify which factories are most likely to perform poorly on audits (thereby helping companies more efficiently target their resources); where costly and reputationally damaging strikes could break out; or where worker safety is at risk.

But deploying any sort of analytics at scale is a complex undertaking that requires time and resources. Before jumping on the bandwagon, managers looking to integrate predictive analytics into responsible sourcing should bear three things in mind: think big, start small, and get the data right.

Think big…

Taking a step back and looking at the bigger organisational picture before diving into implementing predictive analytics is fundamental. Most companies are already considering how to work with big data and predictive analytics. Positioning analytics programmes within this wider context means going with the grain – in addition to buy-in this can also open doors to in-house data scientists, the company’s nascent data lake, or centralised funding for analytical tools.

Thinking big also means changing how the department views data. Rather than disregarding data as a by-produce of various activities or as information to apply in specialised situations, it should be viewed as a valuable resource, even if not currently being fully utlised. Simple changes to what data is captured today and how it is stored can provide key inputs into future predictive models.

…start small

Given the massive promise of predictive analytics, it’s easy to get carried away. However, for responsible sourcing teams starting on this journey, it’s worth reigning in ambitions and treating it as a scalable tool.

If predictive analytics are implemented correctly, much can be achieved in terms of knowledge development, colleague buy-in and even return-on-investment from relatively small projects.

A few simple considerations can improve the likelihood of achieving positive outcomes:

  • Start with a very clear vision of what should be achieved and what specific challenges the new data is helping to address
  • Tackle a problem that can demonstrate real business value, such as cost savings or dealing with a material issue
  • Work with already available or easily obtainable data
  • Communicate clearly with key internal stakeholders, from a project’s inception, during implementation and through to its conclusion
  • Complete initial work and obtain results in just a few weeks or months

…and get the data right!

Predictive modelling projects are only as good as the data that goes into them. The data structure, type and update frequency all determine the types of questions that can be asked.

The promise of ‘big data’ often accompanies predictive analytics, with a common perception that more data must always be more accurate. However, larger datasets often present their own challenges. Specifically, if the inputs are skewed – because the sample is unrepresentative, which can be the case with social media data, for example – then the outputs will be skewed as well, despite the supposed benefits from having used large volumes of data to train the model. Using high-quality, trusted data is a must.

What next?

Predictive analytics has huge potential to add to the effectiveness of responsible sourcing activities; adding it to the team’s toolkit can increase both its existing skills and knowledge. For managers considering how predictive analytics can help their organisations, the best course of action is to just start experimenting today.

Case study:

Using predictive analytics to improve factory auditing regime

Audits are an effective tool for improving compliance with responsible sourcing policies; however, they can be very expensive. A global food and beverage company wanted to plan a cost-effective multi-year auditing programme, which prioritises the highest risk facilities and reduces oversight of lower risk facilities.

We used machine learning methods to develop a predictive model that forecasts how the company’s factories are likely to perform in future audits. This incorporated datasets measuringinternal issues (such as historical audit performance), and external factors, such as the population density around the factory and our own labour rights risk indices.

The model showed that the company’s factories varied significantly in how they were likely to perform in future audits, ranging from an 85% likelihood of failure to 0.01% (see figure below). These results can be fed back into ensuring resources are targeted at maintaining compliance in the highest risk locations, generating better outcomes and better returns on investment for the programme.

By Dr James Allan, Head of EMEA & APAC, Consulting; Guy Bailey, Head of Analytics

Filter Results


Press enquiries:

Jason McGeown, Head of Communications
Tel: +44 (0)1225 420000
press@maplecroft.com

To find out more about Verisk Maplecroft’s country risk reports, briefings and election monitors
Enquire

Verisk Logo

Verisk Maplecroft is a Verisk business.

Verisk Analytics®