The Truth About Size Zero

I came across an interesting article from Wayne Brown on AT Kearney website in their Automotive industry's Ideas & Insights section. The article titled 'Making Sure Lean Stocks Are Properly Nourished' talks about the importance of optimal inventory, not zero inventory as some lean practitioners believe to be the holy grail. The article is well written and hits the mark mostly. There is a key miss though when it comes to planning accuracy, which has prompted me to write this blog post. 

The gist of the AT Kearney article is as follows: 

  • Lean inventory principles have been in vogue in the automotive industry.
  • Lean has put inventory on extreme diets much like some of the superthin models.
  • Like these supermodels pursuing a false ideal of perfection, some automotive companies are seeking unrealistic inventory levels: as minimal as possible.
  • Some even seek the holy grail of zero inventory with Just In Time inbound supply chains feeding the assembly line or build to order supply chains.
  • While lean has benefited some supply chains with excess fat, it has taken things too far in some cases.
  • For example, a car manufacturer had to charter private jets to get material on time from one of its tier one suppliers. 
  • It's time to restore a healthy balance: not size zero, but right sized inventory.
  • A better place to focus efforts to optimize inventory might be the huge volumes of finished cars to decouple artificially stable production volumes with dynamic customer demand.
  • Lean has mistaught the world that all inventory is always a form of waste. In reality, it is the excess inventory that is waste, not the strategic buffers needed to protect customer service levels against unpredictable supply chain disruptions.
  • Inventory optimization policy must be built on a paradigm to balance the conflicting drivers and trade-offs between forecast accuracy, supply chain resiliency, and customer segments that have their own demands on service levels, pricing, and response times. 

All good points. I agree. Now for the part that caught my eye and resulted in this blog post.

Wayne Brown talks about Planning Accuracy in the following terms: " The one sure thing about sales forecasts is that they will be wrong. So unless your customers are willing to wait longer than your total supply lead time, you will need buffer stock somewhere. This can be an issue for manufacturing resource planning systems, which because of their focus on the plan are notoriously inflexible—and particularly bad at accommodating make-to-order products. Higher-performing companies acknowledge that sales forecasts are at best a guide and therefore hold stock at strategic locations to absorb fluctuations in demand." 

At GitaCloud, we see it differently. In our opinion, this mindset trivializes the importance of an accurate forecast and looks to pay for sales forecasting sins by simply holding excess stocks to cover for it. This is similar to the parenting mindset of buying excess candy to keep kids happy instead of getting a handle on how much candy the kids should be eating. No amount of candy is too much in such scenarios.

We believe it is simplistic to hide behind the position that forecasts will always be wrong and hence why bother. We also believe it is misguided to accept high levels of forecast error as an impossible problem to solve and move focus to inventory optimization or supply chain resiliency to deal with the noise.

Demand Forecasting, Inventory Optimization, and Supply Chain Responsiveness/Resiliency are all needed, but in that order. Focus must be on solving downstream problems first before moving upstream: fix demand forecast before you set inventory buffers. Strategic buffer stocks will always be needed, but they should be set on the back of the most accurate long term demand sensing signal you can get. Not just the manual forecast from the sales team, but an optimally sensed and shaped demand signal across machines and humans.

Some SCM software players in Response Management domain similarly tout Supply chain resiliency and responsiveness as a panacea for all demand and inventory planning evils. Supply chain resiliency does not and cannot eliminate the need for robust demand optimization and inventory optimization upfront.  Supply side resiliency is absolutely necessary, but hardly sufficient.

Better planning and scenario management across demand and inventory upfront does not and will never eliminate all surprises, what it does is reduce risk to financial performance by reducing the intensity of impact form such eventual disruptions. 

Our recommendations:

  • Invest in robust, granular, and forward looking demand forecast signal. Forecast will never be perfect, but it can be significantly better than what you have with today's self-learning and powerful demand modeling / sensing systems.
  • Build a forward looking demand model based on all relevant demand drivers and a self-tuning software that can understand the non-linear relationships of demand drivers like pricing and consumer preferences to demand for cars, for aftermarket parts (spares), and for remanufacturing in the reverse supply chain.
  • Allow for effective demand shaping and demand optimization (prescriptive analytics: what should we do to make the future get more in alignment with our goals) on top of demand sensing (predictive analytics: what is likely to happen). 
  • Allow humans to override the demand sensed signal, but rigorously check for value-add (or value leakage) as a result of human intelligence on top of the system generated signal. 
  • Drive inventory optimization and buffer stock setting as an upstream planning task to deal with the noise (error) in the optimized demand forecast (based on sensing as well as shaping). Understand the full range of stock to service level curves in the context of customer segments. The trick lies in knowing when your right sized strategic buffers start becoming excessive. This is where inventory optimization built on top of deep demand sensing can really multiply value.
  • Develop supply chain resiliency as a further upstream activity to deal with the eventual fluctuations and disruptions AFTER your forecasting and inventory optimization processes are working well. 

We believe you need to combine smart machines (for demand sensing) with smart humans (for demand shaping) to get your optimal demand signal for rest of your upstream processes to consume. GitaCloud offers the smart machines part as well as managed services to provide demand sensing as a service.

Forecast needs to be credible & actionable for it to be relevant to supply chain operations. Do not blame supply chain operations to toss the S&OP plans out the door as they sense demand to be different than the one in agreed plan. Also, do not get carried away be the promise of lean inventory, balance it with customer service and market share goals. 

Watch out for your ERP/APS vendor's incorrect understanding of demand sensing: Demand sensing is not limited to a short 4-6 week execution horizon, it is a fundamentally different approach to forecasting your full horizon. We are talking about demand modeling & optimization using advanced pattern recognition and downstream sensing of demand, it has no business constraining itself to a short 4-6 week horizon. 

There is money being left on the table here: most enterprises can get a significant boost to their financial performance delivered by reduction in lost sales and freed up working capital. It is possible to have you cake and eat it too: deliver customer segment specific service levels with an order of magnitude less inventory.  

In conclusion, excess of anything is bad and this is also true with running excessively lean when it leads to lost sales and unhappy customers. I agree with Wayne here: optimal inventory is key, zero inventory is a dangerous idea.  



Ashutosh Bansal is the Founder & CEO of GitaCloud. You can reach him at 


GitaCloud is on a mission to enable Integrated Business Planning with machine-human combined optimal decision making through our predictive, prescriptive, and cognitive advanced analytics based SaaS offerings. Visit for more details.