Supply Chain Strategy The Logistics of Supply Chain Management

1. 1Define the clock boundaries first. Logistics cycle time runs from order placement to final delivery, so write down the start event (order capture) and the stop event (delivery at the customer's location). Then list the activities between them using the passages' structure: order capture and entry, order processing, warehousing (receiving, putaway, storage, order picking, shipping), and transportation. Assign an owner to each segment.
2. 2Set the requirement from the customer service policy (CSP). Take the response-time requirement (arrive before), the time windows (arrive after / depart before), and the shipment frequency from the CSP. Cycle time is only good or bad against these numbers, so record them per customer segment and lane before you measure anything.
3. 3Measure the transportation segment with the passages' cycle time indicators. For each driver, carrier, lane, and location, capture in-transit time (ITT, point to point), in-transit time variability, vehicle load/unload time at each pickup and delivery location, and detention time (waiting at a dock because of congestion or delays).
4. 4Measure variability next to the average. Report the average and the deviant travel times side by side for every lane and carrier. The source says a carrier that always delivers in 48 hours can beat one that averages 24 hours but occasionally takes 96.
5. 5Pair every speed metric with a quality metric. Track on-time arrival percentage (OTAP), on-time departure percentage (OTDP), damage-free and claims-free shipment percentages, and perfect delivery percentage (PDP) for each carrier, lane, and driver. A fast delivery to the wrong place or with damage counts as a failure.
6. 6Mine the activity profiles for where time is lost. Build order-completion profiles (items against orders) and transportation activity profiles (shipment weight, cube, frequency, due time, lane transit times, time windows, on-time %). Look for small groups of items that complete large shares of orders, and for lanes or locations with high detention or variability.
7. 7Redesign the segment where the profile shows the loss. Move the small high-completion item group into a dedicated order-completion zone near the front of the facility, as in the media distributor. Where detention or load/unload time is the problem, work the dock or location. Where one carrier or lane drives variability, move volume or renegotiate.
8. 8Close the loop on a fixed cadence. Re-profile after each change and compare cycle time, variability, and the quality indicators against the CSP requirement. Confirm that cost (labor, space, equipment, transportation) did not rise faster than the service gain.

The source describes a large media distributor of compact discs, cassettes, and videos. An item-order completion profile of its roughly 4,000 SKUs showed that 5 percent of the SKUs (about 200) could complete 35 percent of the orders. The distributor assigned those SKUs to carton flow rack pods, with three flow rack bays per pod and one operator per pod, at the front of the distribution center. Operators could pick-pack from the pods at nearly six times the overall rate of the distribution center. That is a large cut in the order picking and shipping portion of the cycle for over a third of orders. The distribution center has won its industry's productivity award for the last two years. The same logic applies to transportation: the source's 48-hour carrier versus a 24-hour-average carrier with occasional 96-hour deliveries shows that lane-level variability can matter more than the average.

• If a carrier's average in-transit time is faster but its in-transit time variability is high, then prefer the slower, consistent carrier whenever the CSP is built on an arrive-before commitment.
• If a small share of SKUs completes a large share of orders (the 5 percent / 35 percent pattern), then build a dedicated order-completion zone for them instead of speeding up the whole warehouse.
• If detention time or vehicle load/unload time is high at specific locations, then fix dock congestion or scheduling at those locations before changing carriers.
• If a cycle time gain lowers OTAP, damage-free %, or PDP, then treat it as a loss, because a fast delivery that is late, wrong, or damaged is of no use to the shipper or the customer.
• If a transportation change shortens transit time, then check inventory carrying and warehousing costs in the same analysis, since the transportation solution affects both.

• Tracking only average transit time. The tell: customers complain about late deliveries while the dashboard shows a good average. Add variability and the deviant travel times by lane and carrier.
• Measuring speed without quality. The tell: a fast cycle time alongside rising claims, damage, or misdeliveries. Report OTAP, OTDP, damage-free %, claims-free %, and PDP next to every cycle time figure.
• Optimizing one segment in isolation. The tell: faster picking or shipping that doesn't change order-to-delivery time. Time the full chain from order placement to delivery and assign owners to each segment.
• Setting targets without the CSP. The tell: nobody can say whether a 48-hour lane is good or bad. Record the response-time requirement and time windows per customer segment first.
• Speeding up the whole operation instead of mining the data. The tell: broad, expensive speed-up projects with small results. Use activity profiles to find the small item groups or lanes where the time is actually lost.

• ▸Lane / customer segment: origin-destination pair or customer group being scored.
• ▸CSP requirement: required arrive-before time, delivery time window, and shipment frequency.
• ▸Carrier / driver / location: the entity measured on this line.
• ▸In-transit time (ITT): point-to-point average, and the deviant (longest) observed time.
• ▸In-transit time variability: spread of transit times for this lane and carrier.
• ▸Vehicle load/unload time: average at each pickup and delivery location.
• ▸Detention time: average wait at the dock from congestion or delays, by location.
• ▸Quality check: OTAP, OTDP, damage-free %, claims-free %, and PDP for the same lane and carrier.
• ▸Action and owner: segment with the largest time loss, the planned change, and who owns it.

1. 1Fix the boundary first. TLC is the expense and capital cost across all five logistics activities: customer response, inventory planning and management, supply, transportation, and warehousing. Write the five as rows in a ledger. Anything outside them stays out, and anything inside must appear, even if it sits in another department's budget.
2. 2Collect expense and capital together. Operating spend (labor, freight, purchased services) and capital (inventory, facilities, equipment, systems) both count. Put each activity's expense and capital in two columns so a cost like inventory carrying cost is not dropped because it never appears as a payment.
3. 3Compute TLC as a ratio to sales. Most U.S. corporations spend 8 to 15 percent of sales revenue on logistics, so divide total TLC by sales revenue and compare against that band. At the national level, logistics runs about 10 percent of GDP.
4. 4Split the total into the three decision components used in the book's optimization model: inventory carrying costs, response time costs (warehousing and transportation), and lost sales cost. Lost sales cost is the one most often missing from ledgers, so estimate it deliberately.
5. 5State the objective and constraints before you compare options. Objective: minimize TLC = inventory carrying costs + response time costs + lost sales cost. Constraints: inventory availability ≥ the customer service inventory target, and response time ≤ the customer service response time target. An option that breaches a constraint is not a cheaper option, it is infeasible.
6. 6Evaluate each design alternative on total cost, not on one activity's cost. Moving cost from one component to another (for example, more inventory to cut transportation or lost sales) is only a gain if the three-part sum falls while the constraints hold.
7. 7Use TLC as the basis for network and flow-path decisions. The book's implementation plan determines DSD, cross-dock (XD), and DC flow paths based on total logistics cost and its implications. Run each candidate path through the same ledger.
8. 8Track TLC over time as TLC divided by sales (or GDP at the macro level). The Cass series plots total logistics cost against GDP over 1989 to 1999 beside inventory carrying cost against GDP, so you can watch the total and its largest component together.

The book's own use of TLC is customer service policy optimization. The objective is to minimize TLC, made up of inventory carrying costs, response time costs (warehousing and transportation), and lost sales cost. The constraints are that inventory availability meets the customer service inventory target and that response time meets the response time target. Every candidate service policy is therefore priced three ways. Higher availability raises carrying cost and lowers lost sales. Faster response raises warehousing and transportation cost and also lowers lost sales. The lowest-TLC policy that still satisfies both constraints wins. Illustrative scale check (the revenue figure is hypothetical; the 8 to 15 percent band is from the source): a company with $100 million in sales and a logistics-to-sales ratio inside that band carries a TLC of $8 million to $15 million. A policy change that moved carrying cost down but pushed lost sales up by a larger amount would raise TLC and fail the test, even though the inventory line looked better.

• If a proposal lowers one activity's cost (say transportation), then require the full three-part TLC sum to fall before approving it; otherwise you are shifting cost, not saving it.
• If a design option breaches the inventory availability or response time constraint, then discard it regardless of its cost, because the constraints define the customer service policy.
• If your logistics-to-sales ratio sits outside 8 to 15 percent, then check the cost boundary first (missing activities or capital) before you conclude performance is exceptional or poor.
• If you choose between flow paths (DSD, cross-dock, DC) or network configurations, then rank them on total logistics cost, not on the cost of the single facility or mode you manage.
• If lost sales cost is unknown, then do not set it to zero; estimate it, because omitting it biases the optimum toward under-serving customers.

• Counting only expense and omitting capital. Tell: the inventory and facility lines are blank or sit in finance, and TLC looks lower than the 8 to 15 percent band suggests.
• Optimizing one activity in isolation, typically minimizing freight. Tell: transportation cost falls while inventory carrying cost or lost sales rises, and nobody sums them.
• Leaving out lost sales cost because it is not on a ledger. Tell: the model always recommends lower inventory and slower response, and the service constraints are the only thing stopping it.
• Treating the constraints as soft targets. Tell: the 'optimal' option has availability or response time worse than the customer service policy requires.
• Mixing scopes when benchmarking, such as comparing a cost-per-hundredweight figure or a corporate-only cost to a supply-chain-wide ratio. Tell: your ratio differs sharply from a benchmark, yet you cannot name which activities each side includes.

• ▸Period and sales revenue: state the period covered and total sales revenue for it.
• ▸Customer response cost (expense / capital): enter both columns for order capture, processing, invoicing and collections.
• ▸Inventory planning and management cost (expense / capital): include carrying cost as its own sub-line.
• ▸Supply cost (expense / capital): include sourcing, purchase order processing, buying and payment.
• ▸Transportation cost (expense / capital): freight, fleet, carrier and document management.
• ▸Warehousing cost (expense / capital): receiving, putaway, storage, picking, shipping.
• ▸Lost sales cost (estimate and basis): state the method used; never leave blank.
• ▸TLC total and TLC / sales: sum all rows and divide by sales; compare to the 8 to 15 percent band.
• ▸Constraint check: record availability vs. target and response time vs. target for each option; mark feasible or infeasible.
• ▸Option comparison: for each alternative, list inventory carrying + response time + lost sales and choose the lowest feasible total.

1. 1Make the order your unit of measure. The source's premise is that logistics exists to fill orders, so customer value is measured per order, not per activity. Set the standard at perfection. Anything lower, the source argues, will not produce the order-of-magnitude improvements logistics needs.
2. 2Define the nine elements of a perfect order in the customer's terms: (1) entered exactly as wanted, by the customer's chosen means, in a single entry; (2) fillable with the exact quantity of every item inside the customer-specified delivery window; (3) picked with correct items and quantities; (4) packaged and labelled as the customer designates; (5) shipped without damage; (6) delivered in the designated time window and to the designated location; (7) communicated, with order status available 24 hours a day; (8) billed correctly, with on-time payment; (9) documented by the customer-specified means (paper, fax, EDI, Internet).
3. 3Assign an owner and a data source to each element. The source warns that you may not even track some of the nine. It also warns that data is scarce because suppliers, manufacturers, wholesalers, inventory planners, carriers and third-party logistics providers all touch the order. Name the party who holds each element's data before you measure.
4. 4Measure each element as a percentage of orders handled correctly, by location. Use the Figure 3-9 layout: rows for % orders captured without error, % completely fillable at entry, % without status communication errors, % shipped without picking quantity errors, % shipped without SKU errors, % delivered without damage, % delivered with perfect documentation, % collected on time. Use one column per facility (the source's example has Atlanta, New York, Chicago, Dallas, Los Angeles).
5. 5Compute Perfect Order Percentage as the product of the element performances: POP = P1 × P2 × … × Pn. Compute it per location, then overall. Do not average the elements. Averaging hides the compounding the source is built around.
6. 6Read the weakest elements first. POP cannot exceed its lowest component. The source's example data shows order fillability (inventory availability) far below the other elements. Rank elements by distance from 100% and attack the lowest. For availability, that points to the inventory chapters (forecasting, fill rate planning, inventory control policy).
7. 7Test the target with the compounding table. At 90% per element, more than 60% of orders are imperfect. At 95%, about 40%. At 99%, about 10%. At 99.95%, about 0.5%. Decide which per-element level your customers' tolerance for imperfect orders requires, then set element targets from that, not from what feels good for a single activity.
8. 8Close the loop with customer-facing policy and monitoring. The source's customer response chapter pairs customer service policy design with customer satisfaction monitoring. Publish the policy (response times, availability promises) and track whether satisfaction moves as POP moves.

The source's own arithmetic shows why customers stay dissatisfied. Suppose each of the nine perfect-order activities is done correctly 95% of the time. Every activity looks strong, yet about 40% of orders are imperfect. At 90% per activity, more than 60% are imperfect. Only at 99.95% per activity does the imperfect rate fall to 0.5%. The source reports that most of its clients have a POP below 50%, and that this is 'the reason' customers seem dissatisfied. Now apply the Figure 3-9 style to one location, using values legible in the figure (the column assignments are partly illegible, so treat this as illustrative). Take 92% captured without error, 72% completely fillable, 92% free of status errors, 99% free of picking quantity errors and 99% free of SKU errors. The product is 0.92 × 0.72 × 0.92 × 0.99 × 0.99 ≈ 0.60, so only about 60% of orders are perfect before damage, documentation and collection are even counted. Fillability at 72% drags the result down most. Inventory availability is therefore the first lever, not picking accuracy, which is already at 99%.

• If you cannot state your performance on one of the nine elements, then stop optimizing and instrument that element first; the source says it is difficult to improve what you don't measure.
• If POP is low, then raise the lowest-scoring element before polishing high ones. The product is capped by its weakest component, and a 99% element has little room to lift a 72% one's drag.
• If the lowest element is fillability (exact quantities available within the delivery window), then treat it as an inventory problem. Work through forecasting, fill rate planning and replenishment policy rather than adding order-entry or picking effort.
• If an element's data sits with a supplier, carrier or third-party provider, then build the measurement into the relationship. Perfect-order performance spans all of them, and the lack of data comes from that spread of parties.
• If management reports individual activity scores near 95%, then convert them to POP before judging service. Nine activities at 95% means about 40% imperfect orders.

• Celebrating activity-level percentages while ignoring the compounded result. The tell: every dashboard line is green at 95–99%, yet customers still complain.
• Tracking fewer than the nine elements, typically dropping status communication, billing or documentation. The tell: POP looks respectable but customers dispute invoices or cannot get order status.
• Averaging the elements instead of multiplying them. The tell: the reported 'perfect order' figure is higher than the best-performing single element could support, or higher than the lowest element.
• Defining 'correct' by internal convention rather than the customer's designated window, location, packaging and documentation means. The tell: orders count as on time internally but are late against the customer's specified delivery window.
• Reporting one overall number with no per-location breakdown. The tell: no one can say which facility or element is pulling POP down, so improvement efforts spread evenly and move nothing.

• ▸Location / customer segment: name the facility or segment this column covers.
• ▸% orders captured without error: share entered exactly as wanted, via the customer's chosen means, in a single entry.
• ▸% orders completely fillable at entry: share where the exact quantity of every item can be delivered within the customer's window.
• ▸% orders without status communication errors: share where status was accurate and available around the clock.
• ▸% orders shipped without picking quantity or SKU errors: share with correct items and correct quantities.
• ▸% orders packaged and shipped/delivered without damage: share arriving undamaged, in the designated window and at the designated location.
• ▸% orders delivered with perfect documentation: share documented by the customer's specified means (paper, fax, EDI, Internet).
• ▸% orders billed correctly and collected on time: share with accurate invoices and on-time payment.
• ▸POP = product of all rows (as decimals): multiply, do not average; record the result per location.
• ▸Lowest element and owner: name the weakest row, the accountable party (including any supplier, carrier or 3PL), and the next action.

1. 1Define the metric before you measure it. LVA = P − LCC: after-tax profitability minus logistics capital charges, built the way a corporation computes EVA. Write down which P and which capital-charge rate your finance team will use, so every initiative is scored on the same basis.
2. 2Name the logistics capital. LCC charges the capital that logistics ties up. The passages name inventory and systems capitalization as drivers, and the North American fix they list is to 'reduce capital assets' through supply chain integration and logistics information systems. List inventory, facilities, equipment, and systems as separate lines. The passages give no formula for ROLA, so define it in-house as a return measured against that same logistics asset base, and agree the definition with finance.
3. 3Baseline cost ratios, but treat them as context only. Compute logistics cost as a percentage of sales. Most U.S. corporations spend 8 to 15 percent of sales revenue on logistics. The source warns that cost ratios can swing without any real change in practice, and that they ignore inventory availability and response time. Never report them alone.
4. 4Pair every financial figure with productivity and service indicators. Put LVA beside the Perfect Order Percentage (POP) and the response-time and availability measures. A cost or capital improvement that lowers service is not shareholder value.
5. 5Split the drivers by what your logistics managers control. They have limited control over wage rates, fuel costs, occupancy cost, inventory carrying rates, and systems capitalization rates. They directly control the amount of inventory, working hours, occupied space, and transportation miles. Assign accountability for LVA movement to the controllable four.
6. 6Score each initiative on all three LVA components: revenue impact, expense impact, and capital charge impact. The source calls LVA a good indicator of overall initiative value because it captures all three. Reject cases that count only the cost saving.
7. 7Optimize inside the service constraint. Frame every decision as minimize total logistics cost, subject to the customer service policy. Customer response is planned first because it defines the constraint, so LVA gains must not breach it.
8. 8Review LVA on a fixed cycle alongside the logistics scoreboard. Use it to rank projects and to show the board logistics' contribution to future shareholder value. EVA is cited (Stern-Stewart) as the most reliable predictor of future shareholder value, and LVA is its logistics counterpart.

The Coca-Cola case in the source shows where the lever sits. Coke leads in marketing, its route drivers reach nearly every customer location nearly every day, and its century-old production recipe is approaching theoretical limits on quality and efficiency. The source concludes that the link between those three processes, logistics, is the next great frontier for Coca-Cola. Here is how you would test a logistics initiative there with LVA. First, take after-tax profit with the initiative (P1) and without it (P0). Second, take the logistics capital charge with and without it (LCC1, LCC0). Third, compute LVA1 = P1 − LCC1 and LVA0 = P0 − LCC0. Only a positive LVA1 − LVA0 counts as shareholder value. A project that cuts transportation cost but needs capital whose charge exceeds the profit gain fails. A project that holds less inventory lowers the capital charge, but it must still clear the customer service policy. Check POP and response time before you book the gain. The source gives no Coke LVA figures, so populate P and LCC from your own finance data.

• If a cost ratio improves but POP, inventory availability, or response time worsens, then reject the gain as shareholder value and investigate the trade-off.
• If a project's profit gain is smaller than its added logistics capital charge, then LVA falls and you do not fund it, whatever its cost saving.
• If a cost driver is wage rate, fuel cost, occupancy cost, inventory carrying rate, or systems capitalization rate, then do not hold logistics managers to it. Hold them to inventory, hours, space, and miles.
• If you cannot show revenue, expense, and capital charge effects for an initiative, then it is not ready for LVA ranking. Go back and quantify all three.
• If management focus is a short-term shareholder return, as the source says of North America, then use the LVA and capital-reduction view to justify integration and information-system investments.

• Judging logistics by cost-to-sales alone. The tell: ratios move with no change in practices, and no one can state the service level behind the saving.
• Counting profit while ignoring the capital charge. The tell: a project reports savings, yet inventory, systems, or facility capital rose and LVA never gets computed.
• Holding managers accountable for uncontrollable costs such as fuel or wage rates. The tell: variance reviews blame fuel prices, and no one owns inventory, hours, space, or miles.
• Optimizing cost without the service constraint. The tell: cost falls while customers complain. Most clients' POP is below 50 percent, and POP is only as high as its weakest component.
• Not measuring at all. The tell: you cannot say what your POP or logistics cost percentage is. The source's line applies: it is difficult to improve what you do not measure.

• ▸Initiative name and owner: one line describing the change and who is accountable.
• ▸After-tax profit without initiative (P0): baseline figure agreed with finance.
• ▸After-tax profit with initiative (P1): include revenue and expense effects.
• ▸Logistics capital charge without initiative (LCC0): charge on inventory, facilities, equipment, and systems.
• ▸Logistics capital charge with initiative (LCC1): include any new systems or inventory capital.
• ▸LVA change = (P1 − LCC1) − (P0 − LCC0): positive means value added.
• ▸Controllable levers moved: inventory, working hours, occupied space, transportation miles, with the size of each change.
• ▸Service check: POP, inventory availability, response time before and after; flag any decline.
• ▸Logistics cost-to-sales ratio before and after: context only, never the verdict.
• ▸Decision and review date: fund, revise, or reject, and when you will re-measure LVA.

1. 1Write one definition of logistics before any design work starts: the flow of material, information, and money between consumers and suppliers. Get every function in the room (marketing, sales, materials, manufacturing, warehousing, transportation) to agree to it. Without this, each person argues from a different professional definition.
2. 2Set the scope as the five interdependent logistics activities: customer response, inventory, supply, transportation, and warehousing. Mark which ones the project touches. Because the activities are interdependent, a change in one has to be traced into the others.
3. 3Start in Investigate, not Innovate. Run logistics activity profiling (logistics data mining), the systematic analysis of item and order activity or of supply and demand activity. The goal is the root cause of material and information flow problems, the biggest improvement opportunities, and an objective basis for team decisions. Do this while a design change still costs only 'tearing up a piece of paper or erasing a white board.'
4. 4Have the affected groups supply the data, verify and rationalize it, and help interpret the results. This follows Hugh Kinney's line, 'People will only successfully implement what they design themselves.' It also gives you a baseline for justifying new investment.
5. 5Build the performance baseline with the four measure families: financial, productivity, quality, and cycle time. Then run a performance gap analysis, which is the last step of the Investigate section, to show where current performance falls short of world-class.
6. 6Move to Innovate only with the profile and gap analysis in hand. Choose world-class practices for each affected activity (customer response, inventory, supply, transportation, warehousing). Drop options the profile shows are not worth considering. Combine optimization techniques, common sense, best practices, and political savvy. Most organizations already have the last three, so the usual gap is analytical resources for modeling.
7. 7Implement through the information systems and the organization. Use logistics information systems and web-based logistics to substitute information for inventory and work content. Then design and develop the logistics organization to run the new processes.
8. 8Before adding advanced stages (collaborative or virtual logistics), confirm the basics are executed consistently. Each stage of logistics development is a prerequisite to the next.

A large food manufacturer asked Frazelle to help build a formal logistics organization. At the kickoff, participants spent the first 2 hours arguing about who belonged in it, and the meeting nearly adjourned by default. The cause was that attendees came from marketing, sales, material management, manufacturing, warehousing, and transportation, plus the chairman's nephew. None had a formal logistics degree, so each held a different definition of logistics. Applying LMP means fixing step one first: agree on a single definition (flow of material, information, and money between consumers and suppliers) and name the five activities before anyone designs an organization. Next comes the Investigate work: profile item and order activity, with people from the affected groups supplying and checking the data, so the organization is designed from evidence rather than from whoever argues hardest. The passage reports the failed meeting and the diagnosis. It does not report the client's results, so none are claimed here.

• If participants cannot state the same definition of logistics, then stop design work and align the definition first, because you cannot build an organization or a process that is not defined.
• If a project is about to specify hardware, software, or a solution before profiling, then pause and profile first, because design-change cost is lowest and improvement opportunity highest at the start and both worsen as commitments are made.
• If an option is proposed and the profile gives it no support, then eliminate it, because projects fail when they work on concepts that never had a chance.
• If the team is strong on common sense and best practices but cannot model the cross-functional problem, then add analytical or optimization resources rather than more opinion.
• If the organization wants to move to a more advanced logistics stage (collaborative, virtual, e-commerce), then first verify it executes the basics consistently.

• Prescribing before diagnosing: the project begins with a solution in hand. The tell is a team that cannot name the root cause of the problem it is fixing, like the doctor who writes the prescription before examining you.
• Rushing the investigation because of media hype, pace of change, and competition. The tell is that no activity profile or baseline exists when vendor or system selection starts.
• Treating logistics as one function's concern. The tell is that the kickoff turns into an argument about whose domain it is, with several competing definitions in the room.
• Running profiling as an analyst-only exercise. The tell is that the affected groups did not supply or verify the data, so they reject the results and do not own the design.
• Skipping the basics in pursuit of advanced models. The tell is that fundamentals such as customer response, inventory, and warehousing are inconsistent while the plan calls for e-commerce or collaborative logistics.

• ▸Logistics definition: write the agreed definition (flow of material, information, money between consumers and suppliers) and list who signed off.
• ▸Activities in scope: mark which of customer response, inventory, supply, transportation, and warehousing the project touches.
• ▸Problem statement: state the flow problem you were asked to fix, in one sentence.
• ▸Profiling data sources: list the item and order activity data to be analyzed and the owner who will supply and verify each set.
• ▸Root-cause findings: record what the profile shows is driving the material and information flow problems.
• ▸Baseline measures: record current financial, productivity, quality, and cycle-time values for each activity in scope.
• ▸Performance gaps: list each gap against world-class practice, with the activity it belongs to.
• ▸Options eliminated: list the ideas the profile ruled out, with the evidence for each.
• ▸Analytical resources needed: name who will model the problem, or flag the shortfall.
• ▸Basics check: confirm the fundamentals are executed consistently before any advanced-stage move.

1. 1Fix the definition before anything else. Write one sentence for everyone involved: logistics is the flow of material, information, and money between consumers and suppliers. Get marketing, sales, materials, manufacturing, warehousing, and transportation to sign it. Without a shared definition, nothing downstream gets built.
2. 2Name the five interdependent activities as your scope: customer response, inventory management, supply, transportation, and warehousing. Assign one accountable owner to each, and put in writing where the activities hand off to one another. Excellence in one activity does not count if it breaks another.
3. 3Profile each activity before judging it. Frazelle's sequence puts activity profiling and data mining ahead of measurement. Collect the activity's own data (for customer response, order and customer profiles; the same pattern holds for inventory, supply, transportation, and warehouse) so that opinions are replaced by an objective picture.
4. 4Build a scoreboard with four kinds of measure for every activity: financial, productivity, quality, and cycle time. Add the end-to-end quality measure, perfect order performance (POP). Treat it as a chain-wide measure, because it depends on many parties.
5. 5Measure POP even though it is hard. Little data exists because so many parties are involved. Start with a rough count of orders delivered right versus total orders, and expect an uncomfortable result. Most of the author's clients score below 50 percent.
6. 6Run a performance gap analysis against the world-class standard for each activity. Use the book's chapter structure as the checklist: customer service policy and satisfaction monitoring, order capture through collections, forecasting and fill-rate planning, sourcing and supplier integration, shipment planning and carrier management, and receiving, putaway, picking, and shipping.
7. 7Use data mining on the largest gaps to find the root cause of the material or information flow problem. Do not choose a project until the root cause is documented, and make the project-team decision rest on that data.
8. 8Sequence the work as investigate, innovate, implement, the order of Logistics Master Planning. Treat software as an implementation tool for a process you already understand. Software projects succeed less than 15 percent of the time, against less than 30 percent for logistics projects overall.

The passage's own case is a large food manufacturer that asked Frazelle to help build a formal logistics organization. The kickoff attendees came from marketing, sales, materials management, manufacturing, warehousing, and transportation, plus the chairman's nephew. They spent the first two hours arguing about who belonged in the organization, and the meeting was about to adjourn with no progress. The cause was not politics. Each person held a different definition of logistics. The fix was to adopt the single definition (flow of material, information, and money between consumers and suppliers) and to name the five activities as the scope. The same logic applies to Coca-Cola. Its marketing, route-driver customer service, and production are already strong, and production is nearing its theoretical capacity. The author calls the linking of those processes, logistics, the next frontier. Once scope and language are agreed, the next move is to profile each activity and compute POP. If your POP comes in below 50 percent, as most of his clients' do, you have a measured reason why customers are dissatisfied, and the gap analysis can start from that number.

• If participants define logistics differently or come from different disciplines, stop and agree on the definition before designing an organization or project.
• If a proposed project is mainly a software purchase, then require a documented root cause and a redesigned process first, because software-based projects succeed less than 15 percent of the time.
• If you cannot report perfect order performance, then treat that as a finding and build the data trail across the parties involved. Assume the real figure is below 50 percent until shown otherwise.
• If an improvement in one activity (for example transportation cost) is proposed, then check its effect on the other four activities before approving it, because they are interdependent.
• If you are choosing between projects, then fund the one whose root cause is best supported by data mining, not the one with the most vendor enthusiasm.

• Starting without a shared definition. The tell is long meetings with no decisions, each function describing logistics in its own terms.
• Treating logistics as one function's job. The tell is that marketing, sales, manufacturing, and transport each optimize their own piece and no one owns the five activities together.
• Buying systems or hiring vendors before diagnosing. The tell is a project built around a vendor pitch with no profile or root-cause data. The marketplace has over a thousand logistics software vendors, which makes this easy to do.
• Avoiding end-to-end quality measurement because it is hard. The tell is that departmental metrics look healthy while customers still complain and nobody can state the share of orders delivered right.
• Optimizing one activity in isolation. The tell is that a gain in one area (such as lower freight cost) shows up as a loss in another (such as service or inventory).

• ▸Activity (customer response, inventory, supply, transportation, or warehousing): one sheet per activity.
• ▸Accountable owner: the single named person for this activity and its handoffs.
• ▸Activity profile data: list the data pulled and the period it covers, so the baseline is objective.
• ▸Financial measure: current value of the cost or value metric you track.
• ▸Productivity measure: current value of the output-per-resource metric you track.
• ▸Quality measure (include perfect order performance where this activity touches it): current value and how it was counted.
• ▸Cycle time measure: current elapsed time for the activity's core flow.
• ▸World-class standard and gap: the best-practice benchmark you compare against, and the difference from your current value.
• ▸Root cause from data mining: the evidence-backed reason for the largest gap, and the project it justifies.

1. 1Name the inventory or work content you want to remove before you pick any system. The source's candidates are finished-goods inventory held in large warehouses, in-house sortation, receiving staging, putaway, pick-face replenishment, order picking, and packing. Breakthroughs come from substituting information for these, so each system requirement must trace back to one of them.
2. 2For each target, list the specific information that replaces it. In the source's cross-docking case this is daily shipping requirements from the DC, online links to suppliers, automatic identification of unit loads, real-time dock-door assignment for inbound loads, and scheduling of inbound receipts. If you cannot name the data element, you do not yet have a substitution.
3. 3Check access before design. The source states that it is difficult to substitute information for inventory and work content if you do not have access to the information. For every data element, record who holds it (supplier, carrier, customer, your own floor), the channel (Internet, EDI, fax, RF tag, smart card, optical memory card) and when it arrives relative to the physical load.
4. 4Map each data element to the LIS function that handles it: capture, analyze, or communicate. Then place it in one of the five LIS subsystems. The customer response system is named as one; the inventory, supply, transportation (TMS) and warehouse management systems each have their own systems section in the book's structure. Gaps appear where a data element has no home subsystem.
5. 5Move information capture upstream of the physical event. Have the vendor send load contents at shipment by Internet, EDI, or fax, so location assignment and product identification no longer force material to be held at the dock. Use ASN information, and schedule receipts proactively to shift time-consuming receipts to offpeak hours.
6. 6Put the information to work in decision rules: assign inbound trailers to the dock closest to the centroid of their putaway locations; assign outbound trailers to the door closest to the load's contents; slot items by popularity, cube, and turnover to cut warehouse travel; prepackage in popular issue increments identified through order-quantity analysis.
7. 7Align the organization to the system. Each logistics activity chapter ends with an organization design and development section, and Chapter 10 covers corporate logistics organization alignment. Assign an owner for each substitution, and for the data it depends on, in the function that runs the physical process. Manage workforce performance too: the source notes that warehouse managers who weed out operators impeding the whole workforce's effectiveness run much lower operating costs and higher shipping accuracy.
8. 8Verify with outcome measures tied to the substitution: on-hand inventory days (Dell and Micron operate under seven days), steps eliminated from receiving and shipping, shipping accuracy, and operating cost. If the work content is still there, the information is not yet being used.

Dell and Micron use the merge-in-transit model. Online information on production schedules and capacities lets them avoid anticipating and housing finished goods in large warehouses. It also lets them avoid in-house sortation, because the carrier's sorting center already does that work. Two things are substituted: finished-goods inventory and sortation labor. The enabling infrastructure has two parts. The LIS provides online visibility into production schedules and capacities. The organization is aligned so that producers and the carrier act on that information rather than on a forecast. The measured result, per the source, is operating with less than seven days of on-hand inventory and significantly higher financial returns than competitors. The same logic applies to a warehouse. Daily shipping requirements, supplier links, unit-load identification, real-time dock-door assignment, and receipt scheduling together allow cross-docking. The DC ships each day what it receives that day and bypasses receiving staging, putaway, pick-face replenishment, order picking, and packing.

• If a physical activity exists only to hold material while information is gathered (for example, staging at the receiving dock for location assignment and product identification), then capture that information at shipment through ASN, EDI, smart card, or RF tag and remove the hold.
• If the data needed for a substitution is not accessible yet, then fix access (supplier links, load identification, schedule visibility) before buying analytical or optimization tools.
• If an inbound or outbound trailer needs a door, then assign it by proximity: inbound to the dock closest to the centroid of its putaway locations, outbound to the door closest to its load contents.
• If the DC can know daily shipping requirements, link online to suppliers, identify unit loads automatically, assign doors in real time, and schedule receipts, then pursue cross-docking. If it cannot, use popularity, cube, and turnover slotting to cut travel time.
• If system capability is in place but cost and accuracy do not improve, then examine the organization and workforce: ownership of the process, and operators who impede overall effectiveness.

• Buying systems before securing information access. The tell: the software is live but the key inputs (inbound load contents, supplier schedules, shipping requirements) still arrive late or by manual re-entry.
• Capturing information but leaving the physical process unchanged. The tell: ASN or EDI data is received, yet freight is still staged at the dock for identification and location assignment.
• Treating one subsystem as the whole LIS, such as equating the LIS with CRM/customer response software. The tell: the customer response system is mature, but inventory, supply, transportation, and warehouse systems cannot share data with it. The source describes the CRS as one of five subsystems.
• Designing the system without the organization. The tell: no owner is named for a substitution, and each activity's organization design section was never done, so decision rules like dock assignment and slotting are not followed on the floor.
• Ignoring workforce effectiveness. The tell: high operating cost and low shipping accuracy persist despite good systems, with a few operators impeding the whole workforce and no one addressing it.

• ▸Target inventory or work content: the specific item to eliminate (e.g., receiving staging, in-house sortation, finished-goods warehouse stock).
• ▸Replacing information: the exact data elements that make the target unnecessary (e.g., load contents, daily shipping requirements, production schedule).
• ▸Information source and channel: who holds it and how it reaches you (Internet, EDI, fax, RF tag, smart card, optical memory card).
• ▸Access status and timing: available now / not yet, and whether it arrives before the physical event.
• ▸LIS function and subsystem: capture, analyze, or communicate; customer response, inventory, supply, transportation, or warehouse management system.
• ▸Decision rule applied: the rule the information drives (e.g., dock nearest putaway centroid, slot by popularity/cube/turnover, prepack in issue increments).
• ▸Accountable owner and organization link: the role responsible for the process and its data.
• ▸Workforce check: any operators or practices impeding overall effectiveness, and the action planned.
• ▸Outcome measure and target: the metric that proves the substitution (on-hand days, steps removed, shipping accuracy, operating cost).

1. 1Define the productivity ratio per logistics activity before you measure anything. The source organizes logistics into five activities (customer response, inventory planning and management, supply management, transportation, warehousing). Assign each one a ratio: inventory turns for inventory, units shipped per person-hour for warehousing. Write the numerator, denominator, and time period in one line each.
2. 2Pull transaction-level data and profile it before you set targets. The source treats data mining as the way to find root causes and to give project teams an objective basis for decisions. Build item-order completion profiles (% of items vs. % of orders completed) for full-case, broken-case, and overall orders. A flat average across the building hides where productivity is won or lost.
3. 3Find the small group of items that completes a large share of orders. Rank SKUs by how many orders they can complete, then locate the break point where a small % of items completes a large % of orders. Test candidate groups with demand correlation analysis (pairs of items ranked by how often they appear together on orders) to see which items belong in the same zone.
4. 4Re-slot that group into a dedicated order-completion zone and give it its own productivity measure. The source reports such zones running two to five times better than the general warehouse on productivity, processing rate, and quality. Measure the zone's units per person-hour separately from the building's, so the gain is visible and the baseline is clean.
5. 5Report the productivity ratio next to financial, quality, and cycle-time measures. The source frames measures as a scoreboard: financial, productivity, quality, and cycle time. A productivity gain that drops shipping accuracy or lengthens response time is a shifted cost, not an improvement.
6. 6Compare your ratios against benchmarks and name the gap. Use gap analysis (the source lists it in Chapter 3) to rank improvement projects by size of gap, so effort goes to the largest shortfall, not the loudest complaint.
7. 7Put each productivity project into the logistics master plan with an owner and a metric. The source's implementation plan includes items such as implementing productivity metrics for the supply partners and inventory performance metrics. Each project gets one metric, one baseline, and one review date.

The source describes a large media distributor of compact discs, cassettes, and videos with about 4,000 SKUs. Profiling the item-order completion data showed that 5 percent of the SKUs (about 200 items) could complete 35 percent of the orders. The team assigned those SKUs to carton flow rack pods at the front of the distribution center, each pod having three flow rack bays and one operator. Operators pick-packed from the pods at nearly six times the overall rate of the distribution center. The measure was units picked and packed per operator-hour, compared pod vs. building. The distribution center has won its industry's productivity award for the last two years. The lesson: the gain came from the data profile that found the 5 percent, not from working harder or buying equipment.

• If a small percentage of items completes a disproportionate share of orders in the item-order completion profile, then build a dedicated order-completion zone for them and measure it separately.
• If an average building-wide units-per-person-hour looks acceptable, then still split it by full-case vs. broken-case orders, because the aggregate masks the zones where the rate differs.
• If a productivity gain arrives with a drop in accuracy or a longer cycle time, then count it as unproven until the quality and cycle-time measures are checked.
• If the productivity metric has no owner, baseline, or reviewer, then do not launch the project; the source notes fewer than 30 percent of logistics projects achieve their goals.
• If a proposed fix is mainly software, then require a manual or analytical baseline first, because the source reports software-involved projects succeed less than 15 percent of the time.

• Measuring with a single building-wide average. The tell: every area reports roughly the same rate, and nobody can name which items or zones drive the result.
• Choosing metrics by what the system already reports. The tell: the ratios are easy to produce but do not map to any of the five logistics activities or to cost.
• Optimizing one ratio in isolation, such as pushing units per person-hour while accuracy or response time worsens. The tell: productivity rises while complaints and re-picks rise.
• Buying technology before profiling the data. The tell: a system is selected before anyone has built an item-order completion or demand correlation profile.
• Treating one-time gains as a trend. The tell: no recurring review, so the ratio drifts back, consistent with the source's observation that US logistics productivity has stayed flat.

• ▸Activity and ratio: name the logistics activity and its ratio (e.g., units shipped per person-hour, inventory turns).
• ▸Numerator, denominator, period: state exactly what is counted and over what time window.
• ▸Baseline value (overall): record the current building-wide figure.
• ▸Baseline by order type: record the figure separately for full-case and broken-case orders.
• ▸Candidate item group: list the % of SKUs and the % of orders they complete from the item-order completion profile.
• ▸Correlated pairs: note item pairs that frequently appear together on orders and should share a zone.
• ▸Zone design: describe the layout, staffing (e.g., operators per pod), and location.
• ▸Zone productivity vs. baseline: record the zone's ratio and the multiple against the building figure.
• ▸Quality and cycle-time check: record accuracy and response time before and after to confirm no shifted cost.

1. 1Accept the premise first: no industry standard for logistics quality exists, and so many measures are available that many managers give up. Choose the Perfect Order Percentage (POP) as your single headline indicator, because it ties together the quality indices of every logistics activity.
2. 2Define 'accurate' before you measure. The source uses the dictionary sense: deviating only slightly or within acceptable limits from a standard. Take the standard from your customer service policy, which stipulates availability, response time and quality requirements, rather than from internal preference.
3. 3Break POP into its five interdependent activities: customer response, inventory planning, supply (manufacturing and procurement), transportation, and warehousing. Assign one or two quality indicators to each. Anchor warehousing on shipping accuracy, which the source names as a customer-service-policy requirement.
4. 4Name every party who can spoil an order: suppliers, manufacturers, wholesalers, inventory planners, carriers, and third-party logistics providers. A perfect order needs integrated, coordinated performance across all of them, so record which party owns each indicator.
5. 5Set the baseline by measuring your current POP. Ask the source's questions: do you measure it, what is your performance, and if not, why not? Do not fix anything until you have a number.
6. 6Read POP as a composite and look beneath it. When it is low, find which activity's index drags it down, then use logistics data mining to trace the root cause in the material and information flow.
7. 7Redesign the process around the cause. For picking quality, use item-order completion profiling: find the small group of items that completes a large share of orders and give them a dedicated order-completion zone, where the source reports processing quality two to five times better than in a general warehouse.
8. 8Re-measure POP on the same definition and the same indicators so the change is attributable. Report it as a shared metric across the parties in step 4, since it only improves through their coordination.

The source's closest operational case is a large media distributor of compact discs, cassettes and videos with 4,000 SKUs. Item-order completion profiling showed that 5 percent of the SKUs (about 200) could complete 35 percent of the orders. The distributor put those SKUs in carton flow rack pods at the front of the distribution center, with three flow rack bays per pod and one operator per pod. Operators picked and packed from the pods at nearly six times the distribution center's overall rate. The source states generally that such small order-completion zones deliver productivity, processing rate and processing quality two to five times better than a general warehouse. The distributor has won its industry's productivity award for the last two years. For a quality program, the lesson is where to aim: shrink the number of SKUs, handoffs and people an order touches, and the warehouse contribution to POP improves. The source gives no POP figures for this case, so measure your own before and after.

• If you cannot state your current POP, then measure it before starting any improvement project. The first step to improving something is to measure it.
• If POP is weak, then locate the activity with the lowest quality index among the five before assigning blame or buying software. Software projects succeed less than 15 percent of the time.
• If an order failure crosses organizations (supplier, carrier, 3PL), then fix it at the interface with a shared measure. A perfect order requires integrated, coordinated performance by and across all parties.
• If a small set of items fills a large share of orders, then isolate it in a dedicated order-completion zone, since processing quality there runs two to five times better than in a general warehouse.
• If a proposed quality indicator has no counterpart in the customer service policy, then drop it or tie it to a stated requirement. Accuracy means staying within acceptable limits of a standard, and the policy supplies the standard.

• Measure sprawl: you track dozens of accuracy indicators and report none as a whole. The tell is that nobody can state a single quality number for the business. Collapse them into POP.
• Siloed quality: each function reports good local accuracy while customers still get imperfect orders. The tell is strong departmental scorecards beside weak end-to-end results. Measure across all five activities and all parties.
• No defined standard: 'accurate' is never tied to the customer service policy. The tell is that teams argue about whether an order was acceptable. Fix the limits first.
• Measurement without root cause: POP is reported but never traced to its source. The tell is a flat trend despite repeated projects. Use data mining to find the cause.
• Treating the warehouse as the culprit: the warehouse portrays the efficiency or inefficiency of the entire supply chain, so errors that originate upstream surface there. The tell is warehouse fixes that leave POP unchanged.

• ▸Perfect order definition: write the conditions an order must meet to count as perfect, taken from your customer service policy (availability, response time, quality).
• ▸Measurement period and order count: state the time window and total orders reviewed.
• ▸Customer response quality indicator(s): list the measure(s) you chose and the current result.
• ▸Inventory planning quality indicator(s): list the measure(s) you chose and the current result.
• ▸Supply quality indicator(s): list the measure(s) for manufacturing and procurement and the current result.
• ▸Transportation quality indicator(s): list the measure(s) you chose and the current result.
• ▸Warehousing quality indicator(s): include shipping accuracy; list the current result.
• ▸Composite POP: state the baseline value and how you combined the five activity results.
• ▸Weakest activity and owning party: name the lowest index and who owns it (supplier, carrier, 3PL, internal).
• ▸Root cause and next action: record the data-mining finding and the process change, then the re-measure date.