Guest opinion: There is no last mile in supply chain

Date Posted: October 17, 2019 Author: Greg White
A new perspective

To truly address supply chain issues, we need to surface some immutable truths that will lend a new perspective and help you create some breakthroughs. In this article, I’m asking you to put your experience on pause and trust in the blessing of naiveté. It’s not easy, but be 8 years old again and look with new eyes on the world.

The supply chain doesn’t end at the “last mile”

This seems so obvious, but how many times do you perceive the supply chain ending at the dock or the customers’ door? Count me as guilty for years of excluding the consumer from “Supply Chain.” I bet you know someone else who thinks this way. We often don’t, but need to remember that consumers are part of the supply chain, and now consumers are in greater control than ever before. Consumers are where the supply chain starts and ends. In fact, supply chain is not truly a chain. It’s a network, so there is no last mile. Let’s unpack that.

Think of items in a new way

Though we often talk about items like people, the items you’re managing aren’t even actors in the supply chain. How many times have you heard someone say?

This item is trending”

That item’s seasonality”

The other item is intermittent”

An item isn’t trending, seasonal or intermittent. Customers’ action on that item is. In fact, an item doesn’t DO anything! They are merely objects of a customer’s desire or purchase action. Look at the items in front of you. What have they done while you’ve been reading this? Nothing. Items merely wait for action. Action by whom? Consumers. So, now we’re rolling!

Why then, do we forecast items?

OK, so let’s agree that the best solution is not to forecast items, but rather it is to forecast future customer purchases, and what influences those purchases. If that’s the case, why do we forecast items based on sales history? Because items, sales, and history have been–in the past–the most available surrogate data for forecasting customers’ future actions and their influences.

Additionally, forecasting items based on sales history was based on a couple of flawed assumptions:

  1. The future will look a lot like the past, and
  2. It is, in fact, the item that we are forecasting

If you have any doubt how stagnant forecast science is, read this Harvard Business Review article and only afterward look at the date (they sagely signal some coming changes in the last paragraphs). Even today, we use statistical models that are a hundred or more years old, because legacy systems can’t process the proper data, or the proper data to produce a complete forecast isn’t captured. Even when adequate data is available, the outdated data management, forecasting and optimization techniques of legacy solutions can’t take advantage of it. These solutions lack the ability to properly organize and use robust data to generate forecasts. So, how do we solve that?


Recognize what demand is, and is not

In order to make the leap, it is important to accept this brutal fact:

All demand is causal

That is to say, there is a reason, an influence–a cause–for every purchase. There is no such thing as statistical or stochastic demand. All demand happens for a reason. We have traditionally fallen back to statistical methodologies because in the past, the reason for demand cannot always be known. Rest assured, however that only forecasting methods are statistical or stochastic, not demand.

What is demand? Demand can, of course, be what is purchased, but it is actually what is desired, whether purchased or not.

“I wanted red, but you had maroon, so I purchased that”

Plus, what is desired, but not purchased.

“You didn’t have a large, so I went elsewhere, but I still wanted large, and from you.”

You can see in the example above that for a forecast to be accurate, it can’t simply rest on the how much of sales. It must include the why – the influences that cause demand, or the factors that interfere with demand. This way when we see the why occur again, we can predict the who, how much, and when.

Demand translation is critical

Accurate demand translation throughout the supply network is critical, and now that we have this new perspective on demand, let’s explore what to do with that. We’ve established that even at the consumer, sales is not demand. Sales is at best an incomplete measure of demand, and at worst it is a completely inaccurate representation of future demand.

For brands, manufacturers, and even distributors (suppliers), the issue of demand definition is magnified. Particularly upstream from consumers, demand is not properly represented by legacy industry definitions. As in our retail example, our perception of “best practice” forecasting has been warped based on the broad–and at the time, necessary–acceptance of inferior methods from long ago, because data to forecast the reason for demand was not available long ago.

To ensure the accuracy of fulfillment throughout the supply network, we need a way to capture the data relevant to suppliers, and then we need a way to translate, transform, and transmit it.

For decades, leaders in our industry have sought out historical retail consumer sales data and/or historical customer orders, rolled up the data, and run numerous statistical algorithms over it to determine what the future will bring. This has been considered best-practice, and multitudes of tech providers have built their business based on 7, 15, or 21 “best pick” forecast algorithms. All of this “advanced” and accepted science provides an elegantly calculated average of volume, trends, and patterns. I call this PostcastingTM, not forecasting.

For suppliers, here’s what’s wrong with that strategy. The history of what was sold through, or even what was ordered by customers, has nothing to do with the demand that will be received upstream. Even for staple items, if Target ordered an average of 500 boxes of tissue last year at this time, there is no reason to believe they will do the same this year.

Why is that?

  1. Many of the conditions aside from demand are different today: stock on hand, expected deliveries, pricing, display capacity, budget, order minimums, pack sizes, and more.
  2. Though many try, the ordering mechanism that customers use is too complex to simulate effectively by their supplier partners

To ensure the accuracy of fulfillment throughout the supply network, we need a way to capture the data relevant to suppliers, and then we need a way to translate, transform, and transmit it. Retailers, as the partner closest to consumers, are potentially in the best position to perform this translation. The ideal demand signal for suppliers would be a retailer’s prediction of their future orders, which accurately considers not just sales, but influences on demand like out-of-stocks, as well as the timing, requirements and constraints of the supplier/customer relationship.

These time-phased projections from the customer order technology would work for two important reasons:

  1. It considers the customer’s current conditions aside from demand: inventory levels, inventory and fill-rate/service/in-stock goals, order mechanisms/schedules, inventory optimizations & order triggers. It also considers things like supply chain constraints and requirements – order minimums, pack sizes, etc. For example, last year my average order was 1000, due to low inventory, larger pack size requirements, and full truckload requirements. Yet this year I am flush with inventory, have smaller pack-size requirements. I can order LTL.
  2. This data projection will actually get shared! When a retailer can transmit projections from their order technology, they don’t have to expose consumer demand (which, we must recognize that thanks to direct-to-consumer vendors and marketplace poachers, they are loathe to do), plus the supplier gets a more accurate demand signal at the same time. Win win!


Of course, one important thing to keep in mind here is that the retailer’s prediction of their future orders must accurately account for true demand and current supply chain conditions. While I’ve seen some retailers successfully take this approach, it’s more the exception than the norm. As a supplier, if you can’t trust that the retailer has intelligently forecasted their orders, you’ll have to find alternative ways to take all the same things into account to develop a projection.

Get the right data at the right time, and get it right

So, arguably, we’ve resolved the issue for one element of the supplier network – demand. Now we have to communicate the game-changing accuracy of our demand throughout the network. This is where communication gaps come in. The truth is that communication between supply network parties are lacking in many forms, but particularly regarding data sharing.

Consumers, retailers and suppliers all speak different languages. How each party values, identifies and even refers to a product is different throughout the supply network. What they do with the data varies greatly. Intuitively, it seems simple, but it is a difficult task to coordinate, collect, and collate data, even on something as basic as tissue.

To be able to translate and transmit our forecasts and other data between parties, we have to be able to harmonize this data. Harmonization is the process of unifying data that is in different formats or layouts, naming conventions (I call it “tissue” you call it “facial tissue”) and more, so that every party defines products, forecasts, etc. in the same way. Data accuracy is so critical to implementing our new perspective above, that companies need to prioritize this data harmonization to enable such critical and game-changing business transformations.

If you’ve ever been in retail, you’ve experienced the painful drama in trying to find even internal information in your own data and spreadsheets. And while we’re on spreadsheets, data harmonization technologies assure that data shared electronically need not be pulled out and manipulated (harmonized) manually in spreadsheets. When manual intervention and human discernment occur in a rapidly-moving environment is when errors or deviations in execution occur.

This manual intervention is mundane and repetitive time, wasted by humans and more aptly handled by data technologies. I think we can all agree that if we never again have to do manual data entry into a product data management (PDM) solution, we’ll all be better off. Just typing “data entry” makes my brain hurt.

The technologies that harmonize and unify this data are breaking down barriers that enable companies to coordinate and collaborate in this ever more transparent world. This collaboration is providing visibility across the supply chain that companies desire and that empowered consumers demand. An informed consumer is a happy consumer, and a happy consumer is gold in today’s environment.

The advancements in AI enable computers to take the mundane, repetitive, and potentially variable data and rationalize in consistent fashion. History has shown that when technology is introduced into work environments it enables we humans to focus on truly complex and ethereal issues that require our rapid and pre-trained, human intelligence to resolve. We get to use our gifts in a way that is more beneficial to our team and more satisfying to our esteem. If a computer can solve the issue, let it. Humans can solve issues for which computers lack the intellect.

To compete in today’s rapidly-evolving commerce environment, we need to open our minds to the new abilities that robust data and data sharing provide us. Transparency is coming and it’s beneficial to every party. Enhance your success by employing new understandings and new solutions. As you evolve:

  1. Open your mind to new perspectives, like our demand example and recognize where legacy views have distorted current perspectives
  2. Take comfort that even in sharing, you can defend your turf, while enhancing your partners’ ability to help you succeed
  3. Recognize that to share you must be speaking the same language, and that technology is an opportunity to shed the mundane and repetitive to free your gifts to assess and resolve the challenges your mind was built for.

I’ve shared some radical concepts in this article, and some on which I wasn’t able to go to great depth. If there is anything you want to understand, explore or argue, don’t hesitate to reach out to me on LinkedIn or Twitter.

About the Author:

Greg White

Greg White is a founder, CEO, board director and advisor in B2B technology with multiple successful exits. He is a principal and host with Supply Chain Now Radio and recently joined Trefoil Advisory as a Partner to further their vision of stronger companies by delivering practical solutions to the highest-stakes challenges.

Prior to Trefoil, Greg served as CEO at Curo, a field service management solution most notably used by Amazon to direct their fulfillment center deployment workforce. He is most known for founding Blue Ridge Solutions and serving as President & CEO for the Gartner Magic Quadrant Leader of cloud-native supply chain applications. Greg has also held leadership roles with Servigistics, and E3 Corporation, where he pioneered their cloud supply chain offering.  He has led over 100 supply chain initiatives at companies like Amazon, Best Buy, the Limited, Dick’s Sporting Goods, Arc’teryx, Michaels, Dell and more.

Related resources


Talking better product launch and allocation decisions with Ferrero USA

The global confectioner mitigates waste, improves service levels and controls costs by connecting digital supply chain visibility with POS analytics.

Keep reading

Say goodbye to constant supply chain firefighting: A guide

How to take an iterative approach to digital supply chain transformation with real-time alerts that motivate teams to collaborate on issue resolution

Keep reading

New white paper exposes the gap between planning and execution

Understand how gaps between systems, teams and processes are keeping you constantly firefighting and hurting your supply chain resilience

Keep reading