Appendix

Innovation Toolbox

Here are three readily and easily applicable tools that facilitate the implementation of one more methods presented in the chapters.

For completed examples and detailed presentations on the toolbox worksheets, please visit the accompanying website www.fastfulfill.org

A. Process Speed Analyzer (PS-Analyzer)

What Information Does It Provide?

Focus is on a specific fulfillment process. The PS-Analyzer evaluates whether the process is fast enough to survive in an Internet-driven world, and identifies the slow causes that are reducing process speed. A quick investigation of the causes projects whether the team can come up with a set of innovative ideas that can eliminate or accelerate the causes. The final decision is whether to Fix-It: Reengineer the process or Design-Build-It: Build a fresh new process.

Definitions

Figure A.1 Process speed calibration

The machine speed range must be based on market information and customer behavior trends. The speed T^B is a critical speed and must accurately reflect the customer pivot factor.

Figure A.2 Speed effect calibration

PS-Analyzer Steps:

When innovative disruptions are successful, they commonly follow one of four performance themes: (i) Time Efficiency—Activities are done faster or quicker, (ii) Cost Efficiency—Fewer or cheaper resources are used, (iii) Quality Enhancement—Products satisfy higher levels of customer wants, or (iv) Provider Expansion.

PS-Analyzer—Header

Table A.1 PS-Analyzer

No.	SLOW CAUSE	TYPE	SOURCE	SPEED EFFECT
1		Friction	□  Skill Inefficiency □  Data Search □  Process Design □  Time Waste
		Speed Bump	□  Batch Process □  Low Priority □  Resource Short □  Decision Optimize
2		Friction	□  Skill Inefficiency □  Data Search □  Process Design □  Time Waste
		Speed Bump	□  Batch Process □  Low Priority □  Resource Short □  Decision Optimize
3		Friction	□  Skill Inefficiency □  Data Search □  Process Design □  Time Waste
		Speed Bump	□  Batch Process □  Low Priority □  Resource Short □  Decision Optimize
4		Friction	□  Skill Inefficiency □  Data Search □  Process Design □  Time Waste
		Speed Bump	□  Batch Process □  Low Priority □  Resource Short □  Decision Optimize
5		Friction	□  Skill Inefficiency □  Data Search □  Process Design □  Time Waste
		Speed Bump	□  Batch Process □  Low Priority □  Resource Short □  Decision Optimize

Table A.2 PS-Analyzer

No.	INNOVATION IDEA	LINK SLOW CAUSES	IDEA DRIVER	SPEED EFFECT
1			□  Data Streaming □  Decision Model □  Process Time □  Resource Cloud □  Automation □  Uncertainty Analytics	□  +10% □  +20% □  +40% □  +60%
2			□  Data Streaming □  Decision Model □  Process Time □  Resource Cloud □  Automation □  Uncertainty Analytics	□  +10% □  +20% □  +40% □  +60%
3			□  Data Streaming □  Decision Model □  Process Time □  Resource Cloud □  Automation □  Uncertainty Analytics	□  +10% □  +20% □  +40% □  +60%
4			□  Data Streaming □  Decision Model □  Process Time □  Resource Cloud □  Automation □  Uncertainty Analytics	□  +10% □  +20% □  +40% □  +60%

B. Lite-AI Builder

What Does it Do?

AI (Artificial Intelligence) adds complexity to decision problems by increasing the number of dynamic input variables and the controllable decision variables. The Lite-AI Builder captures and operationalizes the intelligence linking these inputs and decisions to a target performance metric. It prepares the team for the next stage where computer and data scientists get involved in creating an operational model.

Definitions

Specific Fulfillment Activity: A specific activity within the fulfillment process for which an AI model is being developed.

Performance Metric: The activity outcome you are focused on achieving, or literally, the performance target or goal. It is critical to ensure that the right goal is identified, else you will be efficiently progressing in the wrong direction. In a fulfillment machine, this metric will likely be defined in terms of activity speed and could be associated with listed friction and speed bump causes from an associated PS-Analyzer worksheet.

Process Variables (Dependent Inputs): Physical and digital attributes of the process activities that are changing or changeable. These inputs are known determinants of the performance metric, which implies that as they change the process performance changes. Variables should be described at the functional level with as much precision as possible.

Decisions or Controllable Variables: Decisions could be either stationary (design) or dynamic (operational) in nature. Stationary decisions are made occasionally and have a much lower frequency, while dynamic decisions are real-time and have a higher frequency. Take a disruptive approach to identify the decisions, often we assume something is fixed or cannot be changed, challenge this assumption.

Performance Transfer Relationships: This is the guts of the analysis and comprises the intelligence driving the AI solution. The intelligence team projects the effect of each input on the performance target. Described by two effect measures: (i) Steady-State Effect (Linear assumption)—For each variable what is the likely percentage increase in the performance metric for a unit/category change if all other variables are fixed and (ii) Paired Effect—Identify variable pairs that have an additive/multiplicative effect on performance, then estimate the likely combined effect of the pair for a unit/category change.

Data Curation: Starts with the ROC (Recorded, Optimized, and Controlled) principle, the variables specify the datasets that need to be collected and stored. These datasets will be used to learn from the data and then train your decision-making models. Over time they will validate the performance transfer relationships. All data is time-variant and readily recorded in a transactional database. Curation is the process of cleaning up and referencing this data so that it will be useful in creating decision intelligence.

Feature Engineering Variables: Selecting the variables that are most critical to performance. Set an effect threshold and only focus on modeling those variables with either a steady-state or paired effect above the threshold.

Supervised Learning Notes: In many decision problems the performance transfer relationships are not exact or readily understood. AI can apply the previous information to a curated dataset, to learn and propose a model for better decision making. Record here what are the strategies or tactics in the AI learning process. These will be used by the computer/data scientist to develop an applicable model.

Lite-AI Builder Steps:

Lite-AI Builder—Header

Table B.1 Lite-AI Builder

No.	INPUT PROCESS VARIABLE	DATA MEASURE	DATA SOURCE	UNCERTIANITY RANGE
V1			□  Flow Entity □  System State
V2			□  Flow Entity □  System State
V3			□  Flow Entity □  System State

Table B.2 Lite-AI Builder

No.	DECISION VARIABLES	MADE BY	FREQUENCY
D1		□  Human □  Computer Rule □  Analytical Program	□  Time - Interval: ______________ □  Batch - Size: ______________
D2		□  Human □  Computer Rule □  Analytical Program	□  Time - Interval: ______________ □  Batch - Size: ______________

Table B.3 Lite-AI Builder

	WHY - EFFECT INTELLIGENCE	STEADY STATE/PAIRED EFFECT	DATA CURATION	FEATURE SELECTION
No.	Single Variable Effect		Threshold % =
V1				□  Yes □  No
V2				□  Yes □  No
V3				□  Yes □  No
D1				□  Yes □  No
D2				□  Yes □  No
Pair	Paired Vairable Effect		Threshold % =
				□  Yes □  No
				□  Yes □  No
				□  Yes □  No

C. Brown Multiplier to Idea Generator (BM-Idea)

What Information Does it Provide?

In the Internet-of-Things (IoT) world, fulfillment processes need to adapt efficiently to a wide range of product/service level uncertainties and a broad range of customer preferences. The BM-Idea generator identifies what range of uncertainties and how it affects the fulfillment process. This information is used to generate physio-digital ideas that can be integrated into the fulfillment machine.

Definitions

Specific Product/Service: The specific product or service that is going through a design innovation update to meet the new customer market. Customer Exploration—Segment and Attributes: A customer segment is defined by an aggregate view of the specific product/service. Each segment is described by a set of attributes or product/service features. Five segments are studied: (i) Average-Now, (ii) Average-New, (iii) ±3σ or very different, (iv) ±5σ or highly different, and (v) ±7σ or extremely different customers. Average-Now is the default and represents the current market and serves as a benchmark. Most likely the ±3σ to ±5σ segments are not served in the current situation. Note the ±3σ to ±5σ segments are relative to the Average-New.

Customer Exploration—Research: Investigating how the product/service is being transformed in the marketplace, with a specific focus on digitization and the IoT. Internal market study reports are of little value since many past trends are unlikely to continue in the digitized future. Online blogs, customer reviews, and insightful trend reports from big-name experts (Gartner, McKinsey, and others) are valuable sources. An exploratory investigation of what the Internet allows the customer to do in the context of existing/future products is also recommended. The market knowledge will describe the Average-New segment and the ±3σ to ±5σ segments. These should together be disruptively different from the Average-Now.

Brownian Multipliers: Four types of market of driven multipliers will affect the fulfillment process of the new customer: What, When, Where, and Like (see definitions in Chapter 8). There can be several multipliers of each type. A multiplier is described by specific fulfillment requirements to meet the customer order range. A good multiplier should be easily associated with the attributes of the product/service.

Why Nots: There are four common reasons that an innovative idea or thought is put aside. The why not counter-response links each multiplier to one of the four reasons. Identify which of these applies to each of the multipliers. To counter with a why not do it response, start the argument from one or more of the previous causes. Good multipliers will not be easy to implement and should severely challenge the existing fulfillment process. Being aware of the why nots prepares the team for rigorous BM-Idea meetings and promotes risk transparency.

Digital Idea: A primarily digital solution that will let the fulfillment process rapidly and efficiently respond to the multipliers. This is a classical ideation step. The multipliers and why nots are already known, the team is tasked with formulating ideas that can seed the development process. The task is not a full design but seed ideas that confirm the possibility of a doable solution.

Physical Idea: A primarily digital solution that will let the fulfillment process rapidly and efficiently respond to the multipliers. The digital and physical ideas could be combined into a physio-digital innovation. Integration of the ROC principle is encouraged.

Critical Intelligence: Required intelligence for optimal decision making in the context of the digital and physical ideas. This intelligence will be in the form of data that will be required by associated models or algorithms.

BM-Idea Generator Steps:

BM-Idea—Header

Table C.1 BM-Idea

SEGMENT	ATTRIBUTES
AVERAGE - NOW
AVERAGE - NEW
± 3 Sigma
±5 Sigma
±7 Sigma

Table C.2 BM-Idea

BROWNIAN MULTIPLIERS		WHY NOT?	DIGITAL IDEA	PHYSICAL IDEA
WHAT
1.1		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
1.2		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
WHEN
2.1		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
2.2		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
WHERE
3.1		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
3.2		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
LIKE
4.1		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity
4.2		Higher Costs   Decision Complexity   Process Complexity   Delivery Complexity

Table C.3 BM-Idea

BROWNIAN MULTIPLIERS		DIGITIZATION STRATEGY	CRITICAL INTELLIGENCE
WHAT
1.1		Explosive Decision Map   Artificial Intelligence   Transfer Functions
1.2		Explosive Decision Map   Artificial Intelligence   Transfer Functions
WHEN
2.1		Explosive Decision Map   Artificial Intelligence   Transfer Functions
2.2		Explosive Decision Map   Artificial Intelligence   Transfer Functions
WHERE
3.1		Explosive Decision Map   Artificial Intelligence   Transfer Functions
3.2		Explosive Decision Map   Artificial Intelligence   Transfer Functions
LIKE
4.1		Explosive Decision Map   Artificial Intelligence   Transfer Functions
4.2		Explosive Decision Map   Artificial Intelligence   Transfer Functions

D. Automation Challenge Analysis (AC-Analysis)

What Information Does it Provide?

Focus is on a set of tasks or a process, within the fulfillment machine, that are to be automated. Successful development and implementation of the automation project is a dependent on several attributes that describe the task. The AC-Analysis helps the team to investigate the challenges associated with an innovation automation idea. The analysis then relates these challenges to target performance and the resource constraints. The analytical results will help determine how to prepare for the project and increase success probability.

Definitions

Automation Tasks or Process: The set of tasks or an entire process with defined start and end. Example: (1) Start—Pick order released to the storage area; End—Picked item in tote (2) Start—Ride request confirmed in the system; End—Ride assigned to a driver. Multiple entities (e.g., customer orders) will flow through the process over a period of time.

Input Factor: A task/process variable that determines the difficulty by which the task can be automated. The factor value/state varies between each incidence of the task and often determines the specificity of the task and differentiates it from a fixed automation machine. In a fixed automation program there are no input variables, the machine repeats the same task.

Situation Attributes: These describe the behavior of the input factor in the context of the automation task as presented sequentially to the machine.

Machine Capability: The required machine capability to complete the automation task with given situation attributes. Describe the associated machine building challenge, and identify the related input factors. Estimate the building effort required to overcome the challenge, ranging from minimal to very difficult. Consider the situation score of the input factors in estimating the building effort.

Capability Challenge Score—Derived by summing the three highest capability score and then dividing by 3. Indicates on a 0 to 10 scale the capability challenge.

Performance: Identify the required performance level or targets for the automation to be deemed a business success, and describe in the notes column. Three performance metrics are evaluated. Define both the minimum to succeed and the target performance levels. Estimate the probability that each performance level will be achieved.

Constraints: Explain why reasonable and/or justifiable resources may constraint the success of the automation project. Three constraint categories should be evaluated. Back up the explanation with the likely effect of the constraint on the target performance levels. The explanations and effects make it easier for management to either assign additional resources or explore alternative automation pathways.

AC-Analysis Steps:

AC-Analysis—Header

Table D.1 AC-Analysis

No.	INPUT FACTOR	DESCRIPTION	SITUATION ATTRIBUTES	SITUATION SCORE
1
2
3

Table D.2 AC-Analysis

No.	CAPABILITY	CHALLENGES	CRITICAL FACTORS	BUILDING EFFORT
1	SENSE		□  Input Factor #1 □  Input Factor #2 □  Input Factor #3	□  Minimal (Score=1) □  Normal (Score=3) □  Difficult (Score=7) □  Very Difficult (Score=10)
2	VISION		□  Input Factor #1 □  Input Factor #2 □  Input Factor #3	□  Minimal (Score=1) □  Normal (Score=3) □  Difficult (Score=7) □  Very Difficult (Score=10)
3	MOTION		□  Input Factor #1 □  Input Factor #2 □  Input Factor #3	□  Minimal (Score=1) □  Normal (Score=3) □  Difficult (Score=7) □  Very Difficult (Score=10)
4	MANIPULATION		□  Input Factor #1 □  Input Factor #2 □  Input Factor #3	□  Minimal (Score=1) □  Normal (Score=3) □  Difficult (Score=7) □  Very Difficult (Score=10)
5	MEMORY		□  Input Factor #1 □  Input Factor #2 □  Input Factor #3	□  Minimal (Score=1) □  Normal (Score=3) □  Difficult (Score=7) □  Very Difficult (Score=10)
CAPABILITY CHALLENGE SCORE =

Table D.3 AC-Analysis

No.	PERFORMANCE	NOTES	TARGET LEVELS
1	SPEED		Minimum to Succeed → ___________ Likelihood its Met: □  50% □  80% □  90%+	Disruption Tartget → ___________ Likelihood its Met: □  50% □  80% □  90%+
2	ACCURACY		Minimum to Succeed → ___________ Likelihood its Met: □  50% □  80% □  90%+	Disruption Tartget → ___________ Likelihood its Met: □  50% □  80% □  90%+
3	QUALITY		Minimum to Succeed → ___________ Likelihood its Met: □  50% □  80% □  90%+	Disruption Tartget → ___________ Likelihood its Met: □  50% □  80% □  90%+

Table D.4 AC-Analysis

No.	CONSTRAINT	EXPLAIN WHY	EFFECT ON THE PROJECT
1	COST		Likelihood the constraint will effect: SPEED: □ Low □ Average □ High ACCURACY: □ Low □ Average □ High QUALITY: □ Low □ Average □ High
2	TECHNOLOGY		Likelihood the constraint will effect: SPEED: □ Low □ Average □ High ACCURACY: □ Low □ Average □ High QUALITY: □ Low □ Average □ High
3	TALENT		Likelihood the constraint will effect: SPEED: □ Low □ Average □ High ACCURACY: □ Low □ Average □ High QUALITY: □ Low □ Average □ High