We Know How Much Money You Lost In Sales This Year

Ikigai Labs can identify lost sales and help optimize purchasing decisions while minimizing inventory cost

Ikigai Labs
3 min readFeb 1, 2022

by Nate Lanier and John Tsitsiklis

Figure 1. A plausible trade-off between missed sales [x-axis] and daily average cost of inventory [y-axis] achievable with a carefully designed ordering policy by Ikigai Labs. The existing ordering policy utilized by an e-comm retailer corresponds to the “red” dot. Ikigai Labs provides various “operating options” for the retailer. For example, the retailer could decrease the daily average cost of inventory by ~49% while retaining the same missed sales (i.e. revenue) or continue to have the same daily average cost of inventory and increase revenues by ~6.5% (or a 33% reduction in missed sales) or various options in between. In short, Ikigai Labs provides much better returns on investment for the retailer!

Many of the biggest challenges that consumer goods companies face revolve around making intelligent decisions about managing inventory levels. On one hand, if inventory levels go too low, demand can surge unexpectedly and large quantities of sales can be lost. On the other, if inventory levels go too high a company can have massive amounts of cash tied up in inventory. Also depending on the products being sold, poor purchase order decision making can impose additional costs in terms of overstocked warehouses or expired products. This fundamental trade off between being overstocked and understocked is at the heart of any purchase order decision maker’s daily frustrations. Getting this problem right can truly make or break a company.

If you are feeling this frustration, Ikigai Labs has the answer you’ve been looking for. We use a company’s sales data to model the demand for each specific sku, identify lost sales and arrive at a purchase order decision making algorithm to strike the optimal balance between being overstocked and understocked. This algorithm can be customized based on the specific use case. Are you ok with having a larger amount of inventory on hand to ensure minimal lost sales? No problem. Are you working with tight cash flow constraints and want to keep your inventory levels as low as possible? No problem. Also the algorithm could be tweaked to account for specific constraints on individual SKUs such as high cost per unit or larger physical size which would take up a disproportionate amount of space in the warehouse.

In the plot above we have missed sales on the x-axis and average inventory on the y-axis. In a perfect world, a company would be at the origin (meaning $0 in lost sales and $0 in average inventory). Here in the real world, although we recognize that this is not possible, we can identify what is likely to be the best possible trade-off between lost sales and average cost of held inventory.

In the example at hand, as shown in Figure 1, our method provides a trade-off curve that shows significant improvement over the state-of-the-art of current business practices. Eighteen months of real sales and purchase order data was used to generate the red dot on the plot. The company had approximately $23.7M in missed sales and $27.4M of average inventory per day across the eighteen months. Each blue dot represents a unique purchase order decision making algorithm. We generated the points by simulating our algorithm against real sales data over the eighteen month period. A company could select the optimal algorithm for their use case by selecting a dot with the balance between lost sales and average inventory that best suits them. Just to highlight a few points, it would be possible to hold the missed sales constant and minimize average inventory by approximately $13.6M, meaning on any given day, the company would have, on average $13.6M in extra cash on hand. Similarly, it would be possible to hold the average inventory constant and increase sales by approximately $8.1M. That is, the business has increased sales by approximately 6.5%. Who doesn’t like more money?

About the Authors

Nate Lanier (Applied Data Scientist)

Nate Lanier is an Applied Data Scientist at Ikigai Labs where he is working to build out the platform’s AI / ML capabilities. He holds a Masters Degree in Data Science from Brown University and a BS in Resource Economics from UMass-Amherst. His interests include all things Machine Learning, Data Science and AI ranging from Natural Language Processing to Time Series Analysis.

John Tsitsiklis (Professor @ MIT)

Professor John Tsitsiklis is a Clarence J. Lebel Professor of Electrical Engineering at The Massachusetts Institute of Technology in the Department of Electrical Engineering and Computer Science. He holds a BS degree in Mathematics (1980) as well as his BS (1980), MS (1981) and PhD (1984) degrees in Electrical Engineering from MIT. His research interests include the fields of systems, optimization, control, and operations research and he has co-authored many texts on these topics.



Ikigai Labs

The only operational BI platform that turns actionable insights into insightful actions.