Our client is a leading consumer healthcare organisation manufacturing millions of household name SKUs at a global network of factories serviced by thousands of suppliers.
They wanted to take advantage of AI to evolve their risk management processes to a proactive approach to predict and catch manufacturing incidents before they happen—preventing waste, reducing the impact of recall, and improving customer satisfaction.
Our client supplies more than 3.5 billion healthcare items each year to consumers around the globe, patient wellbeing and product quality is their priority. Product recalls are rare but their impact is significant when they do happen—products need to be destroyed or reworked, there are logistical and remedial costs to incur and there is reputational impact to manage.
Our client needed a way to better prevent recalls and “near miss” product defects that are identified before a product is distributed to consumers. The existing, mainly manual risk profiling process was conducted by reviewing historical data every 6 months, at which point if a risk profile had deteriorated a mitigation plan was put in place. This was resulting in some quality incidents being missed with millions of units a year being destroyed and multi-million-pound cost impact. Their challenge was clearly of a complexity that lends itself to a custom data & AI solution.
Datasparq worked closely with the Business, Data and Technology teams to develop a new suite of risk models to improve manufacturing risk identification by 250% through transforming the reactive half-yearly risk management process to a proactive and continuous approach:
Predicting rare events like recalls is hard—like trying to find a needle in a haystack of data. We used advanced machine learning techniques leveraging large quantities of data to predict anomalies and events that are likely to lead to a recall including near misses and customer complaints.
The supplier risk model predicts the risk of a quality incident that could lead to a product recall relating to the thousands of suppliers that provide raw materials and packaging.
It uses a machine learning model to predict the probability and the impact of a given incident—alongside providing an overall risk score for each supplier.
The Supplier Management Team, Procurement and Manufacturing Sites can access a dashboard to see both the incident and supplier risk scores, ranking of suppliers and a clear explanation of the scores; allowing users to invest their time in managing the highest priority risks and make targeted mitigation and improvement plans.
The manufacturing site risk model predicts the risk of a quality incident that could lead to a product recall at 25 global manufacturing sites—should no intervention occur.
Using data on indicators such as previous incidents, inspections, complaints and more, it predicts the precise manufacturing activity the incident will relate to at each site.
Central Quality Teams and Manufacturing Sites can use a dashboard to identify and interrogate risk, reviewing clear explanations and signposts to where intervention is needed.
Datasparq is an end-to-end AI & data transformation company. We help enterprises tackle complex, high-value business challenges using AI. Working as an integrated partner, we help the world’s best-known companies navigate the threats & opportunities of AI by designing, building and running cutting-edge solutions.
