Strategic Agility and Absorbing Supply Chain Shocks

2021-01-20
By
Gonzalo Benedit
Strategic Agility and Absorbing Supply Chain Shocks

Evolving Beyond Extraordinary Reaction to Ordinary Response.

A Humanitarian Crisis Transitioning to Response and Recovery

The coronavirus (COVID-19) is not only impacting organizations around the world, but is deeply affecting many of us on a personal level--from our families to our peers to the employees whose welfare we have a responsibility for. Fortunately, protective measures are being implemented to safeguard our communities and our workers. Business owners and leaders have to begin thinking ahead as we craft strategic plans to guide our companies through this crisis. Many of us are dusting off playbooks and response plans from previous events. But we believe it is time to completely rethink the model, centered around the ability of supply chains to self-heal and be resilient. 

Here is a different model that focuses on capabilities that build everyday agility that can handle crises systematically. It focuses on building systems that have a real-time understanding of the entire operation, that anticipate and advise proactively, and that augment human decision-making seamlessly across the supply chains.


Groundhog Day and a Broken Model

As of March 2020, a look at current events already shows response patterns similar to previous supply chain events like the 2010 Eyjafjallajökull eruption, 2011 Tsunami, 2011 Thailand Floods, 2015 Tianjin Explosion, and numerous smaller events like the 2016 Hanjin Bankruptcy. All were considered extraordinary events managed by supply chains acting in crisis mode. The common thread in all these responses was a heavy reliance on emergency crisis management teams, large group conference calls, and endless spreadsheets and powerpoints to manage critical event decision making. A deeper look at these events yields a set of repeating patterns and exposes an inherent weaknesses in our supply chains:

  • The underlying assumption of homeostasis, with spikes (demand, supply, ramp-up/down) handled outside of the regular process (SOPs)
  • The inflexibility of monolithic transactional systems to respond to rapid operational changes, requiring heavy manual intervention
  • The inability to re-task data lakes and BI projects for unforeseen problems

Systems are almost entirely disassociated with the “questions” humans are trying to find answers for, and built to define pre-programmed answers. The underlying enterprise structure and slow batch-orientation of supply chain systems supporting this have largely remained unchanged since the 1990s despite major advances in cloud architectures and database scalability. This exposes the core major problem -- an orientation toward the wrong goal.


Build for Questions, Not Answers

For at least the past 15 years, major IT and supply chain technology projects have been oriented around building for specific goals, KPIs and measurements. Such task-specific efforts are not unlike the mechanical calculators of the computer science eras, from 1800’s Babbage Engines to 1940’s Bombe Calculators. But progressively, we built modern computer systems with generalized processing that could respond and adapt to a wide range of tasks. Instead of building machines for specific tasks, we built machines with dynamic capabilities to which we could apply different questions.


Transit from the Age of Visibility to Cognitive Automation

Cognitive automation is the process of digitizing, augmenting and automating enterprise-wide decision-making processes. If we could redesign our supply chain with this model, we would build it as a series of skills and capabilities that could be re-used and constantly re-ordered to perform more general tasks. Very much like how our supply chain professionals learn fundamental scientific, operational, and management skills in the supply chain and reapply them everyday for ever changing problems and scenarios.

Let’s take a moment to make this more tangible using a real-world example: inventory stockout prediction. If you had a professional dedicated to this problem, you would essentially write the job description as:

  1. Monitor inventory measurements in finished goods, on-hand and in-transit. Monitor production output and raw material availability for monitored products. Understand sale orders, work orders, shipment orders, and supply purchase orders in order to understand potential distressed orders.
  2. Predict future inventory requirements based on experience and observed material flow history of elements listed above. Predict real demand, considering consensus forecast and actual demand via orders. Predict inventory at future dates based on material lead times and expected levels, including customer allocation. Have an understanding of data modeling, statistical analysis, and machine learning algorithms and be able to implement in large scale transactions and fine scale granularity.
  3. Immediately identify over-sell (and undersell) situations and mitigating actions to correct for predicted shortages early in material flows. Recommend optimal mitigation actions given company service, cost, and revenue goals.
  4. Ability to directly act on recommendations below certain thresholds and/or escalated to appropriate management as needed. Must have experience working with carrier systems and TMS, sales order systems, production systems, and internal stock transfers.

In this role, we just ‘built’ a supply chain capability that solves for inventory shortages. But we also gained an asset that could do much, much, more if we just asked it different questions. It wasn’t an algorithm for calculating inventory for a specific SKU/Location based on a series of fixed formulas. It was a system designed to answer a question: “If you see an inventory shortage, what should we do about it?”


Developing the Skills of Agility

Here is how cognitive automation builds these capabilities in our example listed above, asking, “Do you foresee any inventory shortages, and, if so, how can we resolve it before it’s a crisis?”

  1. Understands: Integrates with existing transactional systems with system-specific communication protocols and target-specific extraction rules. Integrate with ERPs, MRPs, TMS, and other source systems for inventory monitoring.
  2. Predicts: Employs continuous data science analysis to predict future inventory levels, timing, and allocation. Predict demand deviations against plan and shortages or excesses. Model various response scenarios and optimize for product specific business goals.
  3. Recommends: Proposes one or more solutions to issues, with full transparency to logic used.
  4. Acts: Automatically executes actions within allowable rules, escalate exceptions according to business rules. Automatically creates/updates required actions in transactional systems in #1 above.

We’ve answered our original question while also building a system that could answer a lot of other interesting questions:

  1. What is my current available-to-promise/capable-to-promise data for my products?
  2. What is the optimal safety stock given my service levels, lead times, and cost trade-offs?
  3. What would happen if my demand goes to zero? What raw material purchase, production runs, etc. can I cancel to minimize my exposure?
  4. What actions would I need to take if I re-optimized for cash flow rather than revenue growth?
  5. What substitutions, orders, and actions do I need to take to shorten the lead time and increase the volume for Y products?


Applying the Skills of Agility: The COVID-19 Sequence of Questions

Phase 1: Pre-Incident Operations

  1. Are there any products currently experiencing a demand spike (negative and/or positive)?
  2. What would happen if my demand goes to zero? What raw material purchase, production runs, etc. can I cancel to minimize my exposure?
  3. Are there products at risk from possible supply shortages, and what are mitigation options?

Phase 2: Post-Incident Operations

  1. What actions would I need to take if I re-optimized for cash flow rather than revenue growth?
  2. What substitutions, orders, and actions do I need to take to shorten the lead time and increase the volume for Y products?
  3. Do I have supplier contracts to cover my material needs and ability to quickly adjust POs for it?
  4. What logistics and inventory options do I have available to stage the right material during ramp-up times to meet surges?
  5. What disruptions can I anticipate for non-essential products given shifts of material and production for critical materials?

Phase 3: Recovery

  1. What changes to orders and schedules do I need to unwind my emergency measures?
  2. Can I reallocate excess inventory or prevent obsolescence as we wind down activities?
  3. How is the market returning to normal post-incident and what is the new demand?
  4. What do I need to do to continuously fine-tune my supply chain as it returns back to normal?
  5. What promotional activities can I implement to accelerate revenue recovery?

Each one of the questions above could have been a project unto itself in the traditional, purpose-built model. But in a model designed around questions, the same set of skills is applied toward different queries. That is what cognitive automation can mean when we need to turn to our supply chains for help.

In Conclusion

COVID-19’s effects are going to affect supply chains for a significant amount of 2020. But the ripple effects of single-sourced material, supplier force majeure and insolvency, and likely uneven demand resumption patterns are likely to last even longer. Let’s get ahead of this one. Let’s out-think this one. And let’s build a model where systems gain cognitive abilities and serve us solutions this time. Then we can not only survive this crisis but ensure we thrive when we emerge from the other side.

Get social with us!

Follow us on LinkedIn: Cognitive Automation Community
Follow us on Twitter: CognitiveAutomation


By
Gonzalo Benedit
,
General Manager EMEA at Aera Technology
Published:
January 20, 2021
Share:
SUBSCRIBE TO WEEKLY EMAIL
-->