Post COVID19 modern SC solutions
The need for speed, agility, personalization, and cost-effectiveness
The ripples of the COVID19 crises are still evident in the global supply chain. Companies struggle to get their raw materials and parts and deliver their goods downstream. More and more companies are investing in understanding different approaches for managing supply chains and overcoming the inherent uncertainty and variability in supply and demand. These uncertainties and variabilities result in severe shortages in some SKUs and, at the same time, some significant surpluses in others. It is evident that the common practices of relying only on forecasts, Min-Max methods, or “sell one, get one”, are totally not effective anymore.
What should be our considerations when evaluating modern SC planning and execution systems? We used to say, “forecast is always wrong”, but the belief is that forecast is a MUST in SC planning and execution systems. Indeed, the forecast is a must for long-term planning in SC upstream aggregation points. This is needed because of the vast difference in lead times upstream (slow or very slow) to downstream (fast or very fast). But, trying to plan the POS where the forecast accuracy of the SKU/Stock location is inferior is a big mistake. Reality is calling for a different approach. Moving from push to pull systems (based on actual consumption and timely demand sensing) is a significantly more effective SC solution.
What is needed from modern SC systems?
- Dynamic buffers – to deal with the inherent uncertainty and variability in the SC, the system must calculate the SKU/Stock location targets (Buffers) daily, based mainly on actual demand and affected by the available forecast (primarily to be able to react to time events like promotions, etc.).
- Setting the initial buffers – this first challenge is met mainly by using state-of-the-art AI and ML to go over 2-3 years of SKU/Stock location consumption to learn the consumption patterns. This should learn patterns like seasonality, special events (like high weekends sales), etc. The algorithms should consider out-of-stock events, bulk orders, and much more.
- Calculating the buffer –the SKU/Stock location inventory target based on consumption today, near-future forecast (while calculating & considering the forecast accuracy), and additional inputs.
- Correction mechanism – such a mechanism should be in place to check the buffer calculation vs. the actual consumption, so if the buffer was too small (high buffer penetration), the buffer will be increased automatically. If we have much more inventory compared to the buffer size, the buffer will be reduced automatically. Doing that in millions of buffers without overshoot or undershoot is a real art.
- Specific constraints modeling – Calculating the required replenishment and calculating the dynamic buffers in not synonyms. For each particular supply chain, different constraints need to be modeled (and fast) into the system.
- Being able to customize FAST all the relevant supply chain constraints is a MUST in modern systems. The constraints changes between different customers are meaningful, and without the ability to model them, one cannot produce a relevant, effective system.
- There are different types of constraints in different systems, to mention a few –
- Physical constraints – MOQ, Full truck, WH starvation, WH capacity, picking capacity, distribution days, and more.
- Policy constraints – Max category, Full shop, trigger point, economic order size, transfer between shops, packaging, and more.
- SC architecture constraints – Omni Channel: today’s SC is typified by a challenging combination of physical shops and hyper online activity. Seamlessly working with shops, online, wholesalers, and distributors. Vertically integrated customers (from production to shelves).
- Merchandise constraints – all kinds of marketing constraints like “complete set” (of measures) in fashion, the timing of returning SKUs back to the warehouse, and much more.
- Cost-effectiveness – The lockdowns put massive pressure on resellers (mainly retailers) to cut costs. Any modern system must be a SaaS system with very fast integration to present a speedy ROI.
- Any system should present the following characteristics to give a swift ROI.
- Speedy time to value – fast integration both on the data level (using WebAPI or secure file exchange) and the ability to model the relevant constraints.
- SaaS model – Software as a service, moving from the Capex pocket to Opex. A monthly fee with no strings. If the system does not bring value, we do not use it.
- ERP/MRP/WMS agnostic – as an add-on system, it should be integrated with any system and not be limited to specific infrastructure.
- Easy to use – The majority of the work is done automatically. Millions of calculations and receiving back the results are seamlessly done within seconds. Nevertheless, capabilities of a working tool to planners with strong “what if” simulation capabilities must be in place.
Progressive Labs Unique Value Proposition –
- Many companies today are already offering dynamic buffers. This is a significant step forward! Nevertheless, this is only the first step in creating a capable system to deal with the market’s needs. The same goes for better systems working on the cloud. The SaaS business model and modern AI and ML capabilities are all needed to answer the market’s needs, and they are all offered by some other companies. But the ability to create, and fast, Specific constraints modeling, is the real challenge!
- Progressive Labs technical capabilities with the extensive knowledge in working with so many customers all over the world, big and small, in a different position along the supply chain, with different areas of products and processes, allowing us not only to model any specific new customer fast but also offer new relevant constraints, customers and prospects never could use in their day-to-day life as they did not have a way to consider in the past.