ERP driven streamlined flows are being replaced with Brownian process flows which dynamically adapt to the highly uncertain and highly temporal demand for products and services. Discover the Brownian catalysts—An Unbounded Selection, Crowdsourced Information, End of the normal distribution and Random is the new normal. The Brownian Multiplier to Idea process will facilitate the harvesting of digital and physical ideas to leverage uncertainty into a success.
Brownian Motion—The erratic random movement of microscopic particles in a fluid, as a result of continuous bombardment from molecules of the surrounding medium (Oxford Dictionary). Streamlined Motion—The organized and repetitive flow of fluid particles in a constrained environment with little-to-no uncertainty. It is obvious from these definitions, that a machine with streamlined flows is way more efficient than a Brownian motion machine, both in terms of cost and speed. Modern supply chains and industry have for decades been designed around the streamlined motion principle. Design and deliver a focused portfolio of products through a standardized process and you have a winner. The strategy has worked very well, starting from Industry 1.0 to 4.0, and companies with streamlined business process flows have grown and achieved great success. The building blocks of Enterprise Resource Planning or ERP systems are business process models that permeate streamlined and standardized operations throughout the firm. This chapter describes how Brownian process flows are evolving into the new normal, and explains why companies need to organize themselves to efficiently and rapidly respond to the varying demand trends in an online economy.
Brownian motion is not a common concept, so let’s first understand the analogical relationship to business operations. Scientifically, it describes the random motion of particles suspended in a fluid. The two primary factors driving this motion are the kinetic energy of the particles and the collision with other particles. The motion was first described by the botanist Robert Brown, who observed it while studying the dispersion of pollen in water. The interpretation is that there is a very high degree of flow uncertainty in the fluid space and every instance or snapshot is different. The business analogy is that online customers have a wide variety of selections or choices, allowing them to exhibit their specific preferential idiosyncrasies. The implication is that the demand for products and services is going to be highly uncertain, highly temporal, and more a function of current dynamics and less of historical trends. Parcels in a fulfillment machine are like particles in fluid flow and if you have stable demand behavior, then you can build a wonderful, streamlined flow process. But online customers are not constrained into normal or constant behavior and every customer and every day is different. The fulfillment machine needs to anticipate the Brownian flow of parcels and create models that optimize these flows for each day. One difference between natural Brownian flows and a fulfillment machine flow is that parcels do not circulate endlessly. Each parcel has a definite start and end, and the flow time determines the speed performance. As we build transfer functions to control the flows, the parameters of each order or an aggregate of orders must link and prioritize the conflicting objectives so that the overall system objective is optimized.
Looking at the definitions of Brownian and Streamline motion, you see that the key differentiator is uncertainty. If you know what you are selling, where you are selling, and how many you are selling, then you have a deterministic flow process and you can streamline and optimize every aspect of it. Since product demand is inherently uncertain, a retailer will smartly attempt to minimize that uncertainty by managing the product portfolio, selecting optimal retail points of sale, and launching an array of marketing campaigns. The benefit of all this is a streamlined product delivery machine. Uncertainty mitigation, or what is more eloquently termed risk management, is not unique to retailing and is practiced in every industry. Henry Ford, the pioneer of streamlined efficient flow systems, very famously said “A customer can have a car painted any color he wants as long as it’s black.” What he meant was, optimize your product-mix offering so that a majority of customers are satisfied. Then design-build a fulfillment system to serve these customers and you have a winning solution. This is the economies of scale mantra, which has been the underlying productivity principle of modern industrialization. If everybody in the world ate the same cookie, engineers would design and build a highly efficient streamlined system to deliver those cookies to each of us at an unbelievable price.
The basic premise in designing a streamlined supply chain is the existence of an identifiable customer or demand majority, with a common or similar set of wants. The larger this majority the more valuable the supply chain and its associated processes. The keyword in Ford’s explanation is the majority, and my argument is that since 2000 the majority has been disbanding and dispersing helter-skelter. In the 1960s, companies were generating products driven by inventive and functional innovation. They were effectively specifying or describing the customer wants, and there was a clear and dominant majority. By the 1980s, many retailers had built streamlined machines to efficiently transfer these products to the customer. Malls across the United States were dotted with these familiar retailing names. By the 1990s, the strongest retailers, such as Walmart and Best Buy, had continued to innovate the streamlined machine and had built formidable supply chains.
Figure 8.1 Dispersion in customer wants and demands
Brownian Catalysts
For a long time, most customers were satisfied with, or accepted the standardized catalog of products or services. Ten-day delivery with a hefty shipping charge was acceptable. The customer wants progressively expanded over time, and the majority started to differentiate and disperse (Figure 8.1). The majority shrank first to 40 percent, then to 20 percent, and then helter-skelter into full dispersion. The streamlined supply chain was sufficiently robust to accommodate the partial dispersions up to a 40 percent majority. Streamlined chains were able to increase product selection and distribute the product through multiple channels. But with further dispersion, these traditional chains struggled to fulfill demand and the disruptive innovators quickly moved in.
So, what has led to the Brownian dispersion of customer wants? Our research identifies four sequential catalysts that are driving customer demands. These catalysts are requiring companies to not just redesign but build completely new fulfillment machines to meet the behavior trends of the online customer.
1. An Unbounded Selection—You may have heard of the maker revolution but are not sure what it is or how it affects the product catalog. It’s the ability of an individual or small company (start-up) to innovate and rapidly design-build a product or service. Makers cracked the economies of scale mantra and could produce new products in low volumes at a reasonable cost. Whether it was cookies, t-Shirts, or handbags, there were choices outside the grocery store and shopping mall. The majority cannot disperse if they do not have choices. The maker revolution provided these choices, and progressively eroded the economies of the scale mantra, giving customers a large selection of products and services.
2. Crowdsourced Information—Social media is an effective influencer and has become a dominant variable in the transfer process that determines consumer opinions, preferences, and ultimately a selection decision. The underlying information is crowdsourced, that is, it originates not from a few verified expert sources but a very large number of our peers. This information resides in blogs, online reviews, webzines, e-mails, Wikipedia, Facebook pages, the list is endless. This information is instantly available to all of us but is also constantly changing as the crowd adds new opinions. The effect of this information on consumer decisions is also unpredictable. Why? Each of us reads a small and different portion of the information, since the information is usually in the form of opinions, we all draw different inferences, and finally we value each source differently. A group of us is convinced that mountain-grown Peruvian quinoa flavored with Kangaroo island honey is the best breakfast cereal in the world. TripAdvisor, Yelp, and Influencers (e.g., the Kardashian sisters) are all examples of companies and organizations that have successfully deployed the crowdsourced information model.
3. End of the Normal Distribution—The normal distribution is an excellent and effective descriptor of the majority. Create a streamlined machine that focuses on the mean ±1 sigma (standard deviation) population, and you have an efficient fulfillment machine. This is exactly what many great companies have done (Coca Cola, Ford, and Ikea). But as sigma becomes larger this becomes increasingly difficult. In the Internet age, customers are seeking a wider range of choices, options, and configurations. Sigma is becoming hyper large and the central tendency is crumbling. To be clear, it’s not the end of the normal distribution, but the end of its widespread use to design and build products and the accompanying fulfillment machine. With a flattening out of the demand curve, the five-sigma customer has the same market importance as the one-sigma customer. The influencers, both individuals and blogs, are not only shifting demand away from the mean but they are also making them fickle, so they choose a different flavor every time they make a buy decision.
4. Random is the New Normal—The customer has many choices and the fulfillment machines make it readily and economically available. Consumer behavior kicks in and you make the selections you like best. Even further, you can make selections that are best suited for you: a 7.25 mg dosage of Lipitor or size 9.65 sneakers. With so many choices, customers have the freedom to keep changing the selection. Every Friday you visit the 87 Prime beer pub in your local downtown, you can choose to order your favorite or try a new craft beer every visit, in effect you are randomizing the demand. The Budweiser customers (the dominant majority of the past) are now a minority in the present. Crowdsourced information is an exogenous factor that promotes randomness, and companies are unlikely to effectively control and/or predict the behavior of this factor.
Some will argue that these catalysts are nothing new and are a constant in business. So how do we validate that these catalysts are radically changing customer demands such that traditional fulfillment systems cannot meet the expectations of online customers? The best way to prove this hypothesis is to look at some innovators who have already implemented Brownian flows.
• Facebook—Advertising machine: Streamline flow—Print or media advertisements are delivered to magazines and then bulk distributed to the audience. The primary differentiating variable is regional or market-specific content. Ads are aggregated or standardized for the majority, so everyone in New York City who reads the Wall Street Journal sees the same ad content. Brownian flow—Facebook delivers ads specifically to individuals, it’s an advertising machine with billions of different delivery points. Ad flow is a function of many differentiating variables, possibly including, your ad click history, friends’ network, likes, news history, and many others. The delivery algorithms are proprietary to Facebook and part of their AI-driven knowledge base. Every user is unique with a transient likelihood of clicking on a particular advertisement.
• Uber—Taxi Service: Streamline flow—Vehicles move on primary roads or wait at defined stands. These roads or arteries are aggregate pints that are closest to the majority of riders. Riders either walk to a primary road or a stand, then wait for a vehicle to initiate a ride. Vehicle availability is uncertain and waiting times can be short- or long based on location. Brownian Flow—Uber accepts the rider’s current location as the start point for all rides, creating a large and random set of ride requests at any time. The current status and location of all drivers are tracked and known. Matching algorithms generate optimal assignments using multiple attributes. Algorithms are proprietary to Uber but most likely are designed to minimize both wait times for riders and idle times for drivers. The system is highly transient and sensitive to the number of drivers. By using surge or dynamic pricing, Uber can incentivize drivers to enter the active pool and better achieve equilibrium in the flow process.
• Amazon—Order Picking Warehouse: Streamline flow— Items are organized in the warehouse in fixed locations and optimized by order frequency. Incoming orders are processed in a batch mode, or daily cycle, and orders are picked to minimize picker travel time and reduce warehouse labor costs. Brownian Flow—Amazon uses explosive storage and items are stored in multiple locations, and the locations are changing continuously. An infinite number of storage arrangements are possible. Incoming orders are highly uncertain but are processed immediately and the goal is to minimize fulfillment time. High-speed algorithms process the very large stocking location database to generate picklists that allow orders to be fulfilled within hours. This is extreme Brownian flow; you could watch the warehouse for months and never see the same picker travel ever again.
• Zenni—Prescriptions Eyeglasses: Streamline flow—Manufacturers of prescription eyeglasses partner with a network of opticians. The optician is an order aggregator with some customer value-adding services. Customers visit the optician and a streamlined flow of orders are transmitted to the manufacturers and then shipped to the optician for customer delivery. Variances and uncertainty at the customer level are accommodated by the optician. Brownian Flow—Zenni communicates directly with the customer and offers a very large portfolio of frame choices. The order flow is highly uncertain with a very large set of order variables including frame size and selection, customer value-added options, delivery address, and delivery speed. Customer order frequency is very low, and all shipments are unique.
• Venmo—Peer to Peer Money Flow: Streamline flow—Most consumer level financial transactions occur through checks or credit cards. Financial institutions have streamlined these flows such that these occur instantly and with perfect reliability between millions of pay points. Brownian Flow— Venmo provides a money flow process that lets immediate transactions between any pair of users in any amount with full validity. These money flows are highly uncertain in three dimensions, the originator and recipient, the frequency, and the amount.
Brownian Organization—The Why Nots?
A streamlined flow machine will have a higher operational efficiency than a Brownian flow machine but is going to become slower in the face of Brownian demand behavior. The design-build challenge is to minimize the efficiency gap and maximize the speed difference. Let’s identify the speed bumps and friction that cause this inefficiency.
1. Higher Costs—It’s the economies of scale mantra. Brownian flows involve many more process setups, slower processes, greater inventories, and a larger number of points of sale, all leading to higher costs.
2. Decision Complexity—When you transition from hundreds to tens of thousands of possible flow paths, there will be a correspondingly huge increase in the number of decisions you need to make. For every online order, you need to decide which warehouse to ship from, which bin to pick the item from, and which transport to use. The common response is, it’s too complex we cannot do that. Most likely, both the physical and information technology infrastructure is unable to process the required decision complexity.
3. Process Complexity—An increase in the number of product/service options will require a production process that can handle all the possible options. Consider a jeans manufacturer that currently offers jeans in seven waist sizes (28, 30, 32, etc.) and two inseam lengths (short, long). If they need to expand to 28 waist sizes (28, 28.5, 29, etc.) and six inseam lengths, there will be a multifold increase in the process complexity. BOFS fulfillment of grocery products is an inherently complex process. Twenty customers order red onions, and each of them a different quantity ranging from 1 lb. to 4 lbs., you need to weigh each separately and track it through the fulfillment process.
4. Delivery Complexity—More product/service options, plus more delivery locations, plus random behavior requires response flexibility outside the capabilities of the streamlined supply chain. Multiple speed bumps slow down the flow, these include issues with intermediate inventories and transport vehicles.
The previous four causes are the source of reasons that an innovative idea or concept is likely to be put aside. To counter with a why not do it response, start the argument from one or more of these causes.
Brownian Multipliers
To investigate whether these changes in online demand behavior affect a specific business, one needs to look first at the drivers that are forcing this change. I label these drivers as Brownian multipliers and identify four types that are defined in Figure 8.2: What, When, Where, and Like. Each of these originates from the market, not the company, and are related to customer events, or product/service transactions that are descriptive of the business’s products and services. In a streamlined flow, the multipliers are limited and in the extreme case will have a single answer or outcome. As the number of configurations (what), locations (where), frequency (where), and customer evaluation attributes (like) increase and change with time, the more likely a streamlined flow approach will not be successful. Inability to recognize and/or accept the growing effect of the multipliers has been a key reason many established company brands have faltered—it is the Kraft Heinz paralysis. The company produces a single popular brand of cream cheese: Philadelphia cream cheese. Yes, there are several flavors and channel-driven packaging options, but in a Brownian economy, these choices are insignificant when compared with the very broad range of demand tastes in the online customer.
For most products and services, demand is either growing or shifting. If the business has a great product but still losing market share, it could be because it is unable to fulfill the Brownian demand. In February 2019, Kraft Heinz lost a quarter of its market value due to slow sales. Famed investor Warren Buffet observed, “I was wrong in a couple of ways on Kraft Heinz. The ability to price has changed, and that’s huge.”77 The growing power of private brands has been a headwind for Kraft Heinz, he said, citing the rising power of Amazon as a brand and the Kirkland brand of Costco. What he did not say, though, and I argue was a critical factor, was that the consumer had many more choices provided by innovative producers with a fast fulfillment machine.
Figure 8.2 The Brownian wants and demand multipliers
Consider the case of a start-up company looking to disrupt the business of distributing wines to restaurants in a regional market. This is big business in Europe, and during my sabbatical in Italy, this is what I learned from some new upstarts. The company has a set of direct customers, restaurants, and a very large number of indirect customers, the restaurant patrons. The streamlined flow strategy would be to compete on price and service efficiency.
Wine Distributor supplying to restaurants—Brownian Multipliers
WHAT: (i) Catalog of wines unlikely to be stocked in the local store; (ii) Seasonal catalog changes; (iii) Supply in units, not cases; (iv) Half/Full bottle sizes.
WHEN: (i) Twice a week not once a month; (ii) Minimum quantity of one, not a case.
WHERE: (i) Ship/deliver directly to restaurants; (ii) All restaurants with no monthly minimums.
LIKE: (i) AI-driven wine selection to match restaurant menu and prices; (ii) Sales analytics for optimal price/quality selection; (iii) Custom online page linked to restaurant web page describing wine selection.
The upstarts succeeded, but they made it clear to me, they had to work long hours, develop a new fulfillment infrastructure, and cleverly use information technology to satisfy the multipliers.
Digitizing Brownian Flows
The most effective way to build efficiency in a Brownian organization is digitizing the information and decision flows. At first glance, this seems elementary. Enterprise computing has been the underlying infrastructure of all successful corporations. These systems digitized all business processes, created centralized data repositories, and automated a wide range of decision-making tasks. Most companies already have a significant IT department led by a Chief Information Officer (CIO), and any recommendations to digitize business flows would seem strange and archaic. Smart teams have successfully built and implemented large-scale enterprisewide cost and efficiency optimization models, and the productivity gains are obvious. So, what’s the big deal with Brownian flows? Brownian organizations require a much larger number of daily decisions relative to a streamlined organization. Each of these decisions, many of which are relatively minor, has an associated transfer function. Both the inputs and desired outputs from the transfer functions are highly random, which in effect eliminates the use of rule-based decision solutions. It is therefore virtually impossible to manage the decision and process of a Brownian flow model using traditional enterprise flow information technology solutions. Here is an example rule-based decision solution widely used in ERP systems.
Inventory Stocking Rule: Close to 80 percent of the inventory in large retails chains (Walmart, HomeDepot, Kroger, etc.) are commonly controlled by a Base Stock reorder policy. Putting it simply, when the current inventory in a store reaches a predetermined level (R) a replenishment order for a predetermined quantity (Q) is placed by the ERP system. Once every couple of years someone specifies Q and R. Now we transition to Brownian flow and decide that Q and R will be dynamic and derived every day, and for every item in every store. First, you need a transfer function model to derive Q and R in real time, then a system to collect and analyze the transfer function input data, customer data to update the desired likes and dislikes, an IT system to operationalize these functions and models, and finally a logistics chain to quickly restock the inventory. Furthermore, every so often the transfer functions need to be updated with new market knowledge.
In an Amazon fulfillment center, few decisions, if any, are being made by the warehouse employees. But the warehouse is not organized like an ERP style standardized flow process. The warehouse serves a constant stream of highly random customer orders and receives a constant stream of incoming stock. All activities have been digitized, and hundreds of model-based data-driven decisions are continuously being made by the cloud computers running the center.
The digital organization allows for physical disorganization. An Structured Query Language (SQL) relationship between the physical entities lets the disorganization be digitally formalized in the computing view. To digitize a Brownian flow organization, the design-build team must adopt and pursue the following three process digitization strategies:
1. Explosive Decision Process Mapping—Start with the assumption that all decisions will be made by a computer and generated dynamically in real time. The fulfillment machine is driven by an explosion in the number of decisions since every action that could potentially affect performance must have an associated decision. The decision process map must reflect this view. Design the network of objectives that will achieve optimal performance, starting from the task level up to the system level. For example, consider the sequence of objectives that link the flow of items through a fulfillment center: Next day order delivery → Prioritized order release → Efficient item picklists → Higher probability of neighboring picks → Prioritizing item restocking → Higher inventory turnover ratios. If the map is dominated by a few strategic decisions, it’s a warning sign that managers assume a deterministic streamlined flow will characterize the business.
2. Artificial Intelligence (AI) Freedom—Digitizing the flows will greatly expand the decision space in a fulfillment machine. What does this mean operationally? The decision space is not limited by any physical constraints or memory capability and hence unbounded. Decisions are not limited to a few choices but a very large, possibly infinite, number of possibilities. To exploit these choices, decisions must be made and implemented and cannot be just theoretical recommendations. Compared to an enterprise system, decisions are made more frequently, and the level of control is more specific and detailed. A fulfillment machine cannot be driven by today’s state of the art enterprise systems, they are just too slow and dependent on human intervention. The solution is AI, it’s real, doable, and profitable. It’s not Mars project AI, but rather small AI solutions that automatically make small decisions for the larger collective goal. Digitized flows set the stage for AI freedom—receive data, make any decision immediately, implement it now.
3. Transfer Function Modeling—In a fulfillment machine, autonomous AI programs make decisions to optimize the current situation. Building these programs requires quantitative knowledge of the relationship between decisions and objectives, and most importantly the uncertainty in the system. This knowledge transforms into a transfer function model at the heart of the AI programs. Multiple parameters that describe the current situation must be accurately identified. If the parameters are incorrect or inaccurate the decisions are likely not going to be optimal. An amazing feature of the fulfillment machine is decreasing decision granularity. By controlling the flow of the smallest transaction units, one can design systems and processes that were infeasible or even unthinkable just a decade ago. In the normal model, we tend to design for human cognition and memory. In a fulfillment machine, we must transfer more decision responsibility to the computer, why, because it has unbounded memory and processing capabilities.
To build a fast fulfillment machine, the design-build team must investigate how Brownian demand affects either the current business or generates new customers. The Brown Multiplier to Idea (BM-Idea) generation process is introduced as a tool that can be used to generate physio-digital ideas that can be integrated into the fulfillment machine. The BM-Idea process pushes the team into an exploration of changing customer trends and then sequentially helps create physio-digital innovations ides to support the fulfillment of these trends. The key steps are given as follows:
• Customer Exploration: Describe the attributes of the new customer who expects a range of driven services. Four news segments are described: (i) Average-New, (ii) ±3σ or very different, (iii) ±5σ or highly different, and (iv) ±7σ or extremely different customers. Starts with a benchmark description of the current average customer, which is derived from the history data. Next, the team must study how the marketplace is transforming. Internal market study reports are of little value, and very likely past trends will not continue in the Brownian future. Read blogs, online reviews, and insightful trend reports from the big-name experts (Gartner, McKinsey, and others).
• Identify and Describe Brownian Multipliers in Your Business Context: Building on the customer exploration step, this step identifies the specifics: the what, when, where, and like drivers of the multiplication process. The team investigates the multipliers that will allow the product/service to successfully fulfill the wide range of new customers.
• Identify the Why Nots: Good multipliers will not be easy to implement and should severely challenge the existing fulfillment process. Being aware of why not prepares the team for rigorous BM-Idea meetings and promotes risk transparency.
• Identify a Digital and a Physical Idea to Serve the Multipliers: This is a classical ideation step. The multipliers and why not are already known, the team is tasked with formulating ideas that can seed the development process. The task is not a full design process but only the generation of ideas that confirm the possibility of a doable solution. The ideas are innovation solution statements that later seed the design-build process.
• What Is the Critical Intelligence? Identify the required intelligence for optimal decision making in the context of digital and physical ideas. A key issue in systems with many decision points is the risk of performance deterioration due to suboptimal decisions at several points. Several approximate decisions or decisions with weak objective function linkages will lead to performance drift.
The innovation toolbox provides detailed steps and supporting worksheets to complete the BM-Idea generation process. The resulting fulfillment ideas should integrate one or more of the three process digitization strategies. The BM-Idea process and outputs highlight the opportunities and possibilities of disrupting the market, through new fast fulfillment processes.
Chapter Summary
• Brownian process flows are evolving into the new normal, and companies need to organize themselves to efficiently and rapidly respond to the varying demand trends in an online economy.
• Four sequential catalysts are driving the Brownian dispersion of online customer wants and their shifting demands: an unbounded selection, crowdsourced information, end of the normal distribution, and random is the new normal.
• The disruptive innovators are already servicing browning customer flows, thus validating that the trend is real and happening.
• Objectors to the Brownian flow model will identify four speed bumps and frictions: higher costs, decision complexity, process complexity, and delivery complexity. Prepare a counter response at the get go.
• The Brown Multiplier to Idea (BM-Idea) generation process is introduced as a tool to generate physio-digital ideas that can be integrated into the fulfillment machine.