Michelin meeting 28/06/21

• Problem Statement: Inventory reduction, not only on Michelin side but also on customer side

• Root cause analysis: Forecast accuracy, inability to leverage → HOW?

• Customer internal forecasts: MAPE M-3 = 45% → Improvement achieved by 10% through internal Planning Clinic; Tomasz would expect at least MAPE = 30% (or even 25%).

• Customer is providing 12-months rolling forecast

• Service level of 95% but not reactive enough → but this is the Michelin business model

• Workshop with 11 customers
  • Diversified portfolio, big, small, digitized, no IT maturity

Customer feedback from workshop:

• We are supplying Europe from Europe with weekly signal; good reactivity with short lead times compared to competitors in Asia

• Yearly forecast shared by customers, monthly bucket on product level

• Contractually agreed bands of flexibility → 95% availability is only measured on the contractually committed demand but Michelin would like to deliver also on the non-contractually agreed Sales ("spot orders")

Scope:

Tier 3: 2000

Tier 2: 400

Locations: 2 sites

Customers: 110, all share forecasts

4 years of data, comparison to 2019 results

Results:

• Michelin has a group target to reduce scrapping of tyres very soon

⛔ Michelin would like to provide AI explainabilty. Why is the value as it is?

—> They are looking more for trends

External data from Michelin Market Intelligence team:

• Market data = composite index

• OEM data

Potential additional data points: Sell-out from customers (customer demand)

Timeline:

• 2 years for implementation of the final solution

• Customers dont expect to see much before the end of 2021

• For example, first internal validation in the end of the year

Discussion (Gelald's questions)

• How to measure success?
  • Gerald: What if customer is changing priority of stock-out and stock reduction in the middle of the project?
  • What is the reference when 2020 was crisis mode? → historic references not really valid.

• What is the Flowlity business model? - How do we share value with the customers? Pricing?

• What is the right way to start?
  • How many customers?
  • How many products?
  • What is the architecture for capturing and sending data?

• Complexity of the model: Is it a problem that only some customer are part of the pilot.

• Working together:
  • Digital factory in Pune → there is a possibility to source internally

• Pricing