Net Working Capital and S&OP
4 likes3,795 views
The document outlines the importance of net working capital (NWC) in financial management, emphasizing strategies to optimize it through effective sales and operations planning (S&OP) and inventory management. It details various approaches to improving key metrics like days sales outstanding (DSO), days payable outstanding (DPO), and days inventory outstanding (DIO) to free up capital for strategic investments. The text also presents a structured framework for NWC project implementation, including phases for solution design, execution, and sustainability.
Net Working Capital and S&OP
- 1. MALMÖ 10 NOVEMBER 2016 Net Working Capital and S&OP A million is always a million
- 2. 2 Content Main speech 1: Net Working Capital – p. 3 Main speech 2: NWC and S&OP – p. 11 Café 1: Implement’s approach to Net Working Capital projects – p. 39 Café 2: Visual Management – Implement’s tool Stock Monitor – p. 46 Café 3: Reference case – Stock Killer project & One Arla project – p. 49 Café 4: Inventory – the definition & calculations behind Stock Monitor – p. 56 Café 6: How to involve Sales in project & Efficient scenario planning – p. 72 Café 5: Implement’s view on Simple forecasting – p. 63
- 3. Main Speech 1: Net Working Capital By Jan Lythcke-Jørgensen
- 4. 4 How many of you know Donald Duck? Scrooge McDuck has a radical view on capital …
- 5. 5 Net working capital (NWC) is a measure of a company’s financial strength Net working capital = value of assets – liabilities Why focus on net working capital and why optimise it? Watch video What is net working capital? Freed up capital can be used to develop the business or reduce debt!!! Why optimise net working capital? Invest in operations Improve service Reduce debt Fund other investments
- 6. 6 Focus areas when optimising net working capital (DSO, DPO and DIO) Usually, a significant amount of capital is tied up in running the business,which is something that the CFO already know … Improving the DSO (-), DPO (+) and DIO (-) will free up $, i.e. change the capital balance, and secondlyand even more importantly, the focus will be on operating the business efficiently.i.e. the cash conversioncycle! Days Inventory Outstanding (DIO) Days Sales Outstanding (DSO) Days Payable Outstanding (DPO) Focus area How to improve » Negotiate creditterms (reduce credit period)with customers » Make sure the customers obeypayment terms, pay on time » Negotiate betterpayment terms (longer payment period)with suppliers » Always use the full payment period,do not pay early » Reduce raw material inventory » Reduce WIP » Reduce FG inventory Greatest potential
- 7. 7 Common pitfalls that typically drive up working due to other drivers of business improvements 1. More convenient to purchase in large batches 2. Strong focus on unit cost/volume discount 3. No differentiated planning and low differentiated inventory management 4. High unmonitored stock service level towards customers 5. Consider capital reduction projects to be CFO projects 6. Not a lean production set-up/long throughput times 7. Operating model does not differentiate between products with different characteristics (one-model-fits-all) Working capital pitfalls
- 8. 8 Working capital is tied up in numerous places in the value chain Minimise lead time and minimise production batch sizes Sales Finished goods inventory Assembly, finished goods Pre- assembly, sub-parts Manufacturing processes Raw material inventory Purchasing Material planning Reduce inventory Negotiate longer credit period towards suppliers • Faster billing process • Negotiate shorter credit period • Make customers pay on time Move order production point upstream in the value chain Value chain – production company (main control areas, i.e. BOM) Working capital improvementareas can be identified by analysing the value chain of a company. Free up capitalby reducing stock/remove unnecessary inventory: » Reducing throughput lead time and improve agility (JIT) » Understanding variance in demand, supply and productionbetter » Pushing order productionpoint upstream (make-to-ordervs make-to-stock) Differentiate products as late as possible in the production
- 9. 9 When improving the DIO and reducing stock, we see a number of solution hypotheses that involve all aspects of the supply chain Flow and productioncontrol principles Supply chain integration Improve sales forecasting and planning Productpruning and SKU reductions » Improving sales forecasting » Improving S&OP system and governance » Improving planning and inventory control » Clear definitionof push/pull, segmented flowin productionand differentiated planning principles » Balance through produce-to-stockon low-risk items » Reducing the number of SKUs and/or raw materials » Introducing standards or configurable items,allowing increased postponement(i.e. short lead times) » Supplierintegration to reduce lead times and variance and increase operating flexibility » Optimising the size of consignmentstock(i.e. zero lead times) KPIs and visual management focus » Increased mutual managementfocus on the topics » Effective KPIs with meaningful targets driving the desired behaviour balancing “trade-offs” Reduce inventory
- 10. 10 Improving the NWC is only achieved through full engagement from the entire business. The resultscan be tremendous and enable future growth As-Is To-Be » 100% P&L focus,cash consumptionneeded to optimise EBIT » Uncoordinated pricing strategy towards customer/productsegments » An extremely high service level always trumps inventory optimisation » Strong focus on low unit costinstead of “one-piece flow” requirements » Responsibilityand management of NWC are anchored to the CFO » Inventory costand write-downs are not crucial as we will use these parts at some point later on » Working capital as a strategic target to change “the way of working” » Clear governance and control set-up » Adjust logistics footprint » Enable lean set-up and flow (internally and externally to strategic partners) » Less “standalone” focus onP&Land unit cost » Reduced lead time to create flow and responsiveness » Joint NWC business effortand maybe bonus (not a CFO project)
- 11. Main Speech 2: NWC and S&OP By Thomas Holm
- 12. 12 When improving DIO and reducing stock, we see a number of solution hypotheses that involve all aspects of the supply chain Flow and production control principles Supply chain integration Improve sales forecasting and planning Product pruning and SKU reductions • Improve sales forecasting • Improve S&OP • Improve planning and inventory control • Clear definitionof push/pull, segmented flowin productionand differentiated planning principles • Achieve workload balance through productionto stock on low risk/value items • Reduce numberof SKUs and/or raw materials • Introduce standards or configurable items allowing increased postponement • Supplierintegration to reduce lead times, variance and increase flexibility • Optimising size of consignmentstocks KPIs and Visual management focus • Increased managementfocus on the topic • Effective KPIs with meaningful targets driving the desired behaviour Reduce inventory
- 13. 13 Agenda Sales forecast and NWC The best way to sales forecast Sales & Operations Planning and NWC Scenario planning Conclusion
- 14. 14 Medium-term sales forecast bias drives higher inventories & costs Over sales forecasting (2-24 months) Excess capacity because the capacity decisions are based on the too high sales forecast • Excess capacity & manning • Excess sourcing of raw materials, components consumptionmaterials, etc. Under sales forecasting (2-24 months) Lack of capacity because the capacity decisionsare based on the too low sales forecast • Lack of capacity & manning • Shortage of raw materials, components consumptionmaterials, etc. Increased inventories& obsoletestock due to over productionand purchasing Increased costs due to idle capacity Delivery performance issues Increased inventories& costs due to firefighting & overreaction
- 15. 15 Medium-term sales forecast has huge business impact Business Impact Behavioural Impact Reduce under sales forecasting Reduce over sales forecasting Improve financial predictability Reduce costs Sales & Operations Planning decisions: • Production capacity • Inventory targets for finished goods, components & raw materials • Sourcing of external capacity, components and raw materials Reduce inventories & obsolete stock Increase sales forecasting stability Reduce variability from statistical forecasting Reduce tamperingReduce human bias
- 16. Sales forecasting & statistical facts
- 17. 17 Sales Forecasting & Statistical facts Company Category, sales org. Product grp, sales org. Material, sales org. Material, customer type, sales org. Material, location, customer, sales org. 1. Law of large numbers: The relative variability is less much for aggregated sales history than on the lower levels 2. Not possible to accurately forecast sales on the lowest level due to high unpredictable demand e.g. due to few random lines per week or high order size variance 3. Tampering is waste of time i.e. try to adjust forecast within the variability of the sales.
- 18. Two ways of Sales Forecasting: • Advanced Statistical • Manually
- 19. 19 Advanced Statistical Sales Forecasting 1. Find the right level to statistically forecast on with appropriate variability 2. Use the forecast model that has the best forecast accuracy for the different materials 3. Disaggregate sales forecast to lower levels 4. Aggregated in other dimensions like sales organisation or customer type to support the data needed by key stakeholders 1 3 4 1 2 3 Company Category, sales org. Product grp, sales org. Material, sales org. Material, customer type, sales org. Material, location, customer, sales org.
- 20. 20 Manual Sales Forecasting 1. Find the appropriate level to enter the manual forecast on 2. Disaggregate sales forecast to lower levels 3. Aggregated in other dimensions like sales organisation or customer type to support the data needed by key stakeholders 1 1 2 2 3 Company Category, sales org. Product grp, sales org. Material, sales org. Material, customer type, sales org. Material, location, customer, sales org.
- 21. Is there a better way to sales forecast?
- 22. Is there a better scientific way?
- 23. 23 Very simple statistical sales forecasting is scientifically better than both advanced statistical & manual sales forecasting 1. Advanced statistical forecasting methods gives lower forecast accuracy than simple ones as shown by Professor J. Scott Armstrong at Wharton University: “If you nevertheless use forecasts from complex methods to help you make decisions, expect to be confused about how the forecasts were made and an accuracy penalty of more than one quarter 25%)”; see www.simple-forecasting.com 2. Aggregating the very simple statistical forecasting on the lowest level – gives the same result as if we had forecasted in the same way on aggregated level 3. Only manually sales forecast leads to forecast bias and is very time consuming 4. Disaggregation gives bad results – if there is not a very simple forecast on lowest level Company Category, sales org. Product grp, sales org. Material, sales org. Material, customer type, sales org. Material, location, customer, sales org.
- 24. 24 Very simple statistical sales forecasting combined with focused insights from sales & marketing is scientifically the best way 1. Aggregated seasonality index on e.g. combination of product group and sales org. 2. Very simple statistical forecasting on the lowest level that takes seasonality into account. The forecast is never used on this level! 3. Focus insights from sales & marketing 4. Aggregated in other dimensions like sales organisation or customer type to support the data needed by key stakeholders 1 2 Company Category, sales org. Product grp, sales org. Material, sales org. Material, customer type, sales org. Material, location, customer, sales org. 2 3 1 4 3
- 25. When is it ok to use advanced statistical forecasting?
- 26. 26 WHAT is a good sales forecast to support medium-term business processes – and HOW to obtain this 1. Unbiased 2. Stable 3. Transparent 4. Market & customer insights 5. Minimize work load for sales & marketing A good medium-term sales forecast The six HOWs: 1. How to structure a clearsalesforecastingprocess. 2. How to build a stable and transparentstatisticalforecast – that is easy to understand. 3. How to incorporateinsights from Sales and Marketing with minimum workload. 4. How to handlesales forecastswith high uncertaintyand impact. 5. How to handlenew productintroductionswith high impact. 6. How to continuously improvethe sales forecast.
- 27. 27 Sales & supply chain impact segmentation helps to focus sales forecasting efforts where it creates the largest impact High Low SalesforecastImpact Demand variability / unpredictability Low High • Trends • Significant step changes • Review total sales forecast • Trends • Scenarios • Review total sales forecast No sales forecast focus No sales forecast focus High Low Sales forecast impact on sales Low High Sales forecast impact on supply chain Product group sales forecast has high impact on sales in country, region Product group sales forecast has high impact on both sales & supply chain No sales forecast focus Product group sales forecast has high impact on supply chain decisions
- 28. Sales & Operations Planning and NWC
- 29. 29 The S&OP process is a cross organisational process that involves various stakeholders on many levels in the organisation Financial forecast DivisionCategory Sales region Product group Product Customer SKU Account Manager ‒ Financial sales forecast per customer to reallocate promotions and sales activities and resources BU VP: ‒ Financial forecast to overall resource allocation Strategic purchasing - External capacity, component & raw material requirements to renew & adjust sourcing CxO’s: - Financial predictability Production/Supply Chain - Capacity load on key resources to adjust capacity Category Manager: ‒ Category forecast to adjust marketing activities & resources Master scheduling & Purchasing - Mix forecast to plan production and purchasing SalesDirector ‒ Financial sales forecast per category to reallocate marketing and sales activities and resources
- 30. 30 Keys in achieving the best forecast possible are by tailoring according to needs and involving stakeholders in the process In tailoring the forecastsetup for the need,planninghorizonsand granularity are ofvital importance • The medium term forecastis used when taking capacity decisions,negotiating supplier agreements and purchasing materials with very long lead times • The short term, operational forecastis used for master planning, purchasing, setting inventory levels etc. Value Drivers Strategic planning 5 years • Effective budgeting • Ability to forecast expected top line and profit • Better capacity investments in machinery and plants Sales & Operations Planning 2-18 months • Timely adjustments of capacity incl. hiring and dismissal of manpower • Efficient strategic sourcing • Optimize balance between capacity, inventories and service Master Scheduling 1-3 months • Better prices from sourcing partners • Lower inventory levels to handle uncertainty Operational planning 0-6 weeks • Limited amount of self-induced rush orders • Less scrapping or lost sales and penalties from customers Focus level of the forecast must be aligned with the purpose of the planning activities taking place. Long Term ShortTerm Medium Term
- 31. 31 PRODUCT REVIEW SUPPLYPLANNING INVENTORY PLANNING DEMAND PLANNING BALANCING & DECISIONS The S&OP process is a cross organisational process to take and execute the decisions that are best for the company as a whole Create (statistical) forecast Review unconstrained demand plan Sales approval Sales input Interface to NPD process Prepare decision proposal Executive decision meeting Update revenue & cost estimate Pre-meeting based on reconciliation Analyse scenarios, incl. financial impact AllSales & Finance SCM & Finance EXECUTION “Handover” from decision meetings Communication & decision execution Update of latest estimate (quarterly) Week 1 Week 2 Week 3 Week 4 Identify demand gaps & supply issues Develop initial supply chain plan Review critical resources & assess scalability Finance Review target stock Review Service Level Agreements Decisions Demand gaps & supply issues Unconstrained forecast The S&OP process consists of 4 general steps: Demand planning, Supply planning, Balancing & Decision Making and Execution Decisions are executed
- 32. 32 Stock increase is fast but inventory decrease takes long time when right sizing stocks Right sized inventory based on differentiated target service levels Increased targets Decrease targets Increased inventories Slowly decrease inventories
- 33. 33 Fine-tuning the balance between service level and inventory is a continuous process Plan: SL improvement Do: Adjust inventory Check: Balance between service level & inventories Act: If necessary Need to improve SL Determine additional adjustments with differentiate service levels Calculate necessary inventory adjustments to stepwise improve SL Implement inventory adjustments Monitor whether service level improvement is sufficient
- 34. Scenario planning is a vital part of Sales & Operations Planning
- 35. 35 The game board is a simple and powerful tool to support scenario planning The financial, NWC, sales, operations consequence of each combination of scenario & choice Best case Base line Worst case Scenarios No change choice 1 choice 2 Choices
- 36. Conclusion
- 37. Planning problems can be complicated!
- 38. Planning problems can be complicated solutions cannot
- 39. Café 1: Implement’s approach to Net Working Capital projects By Jonas Sjögren
- 40. 40 The overall approach to an NWC project, is divided into three overall phases, where we later will deep dive into Phase I Phase I: Individual solution catalogue and action plan for the customer Phase II: Implementation and benefit realisation Phase III: Sustain results Design future state and solution elements across processes, planning and steering principles Plan implementation Purpose 6-12 weeks, depending on scope / size 2-12 months 12+ monthsDuration Implementation of plan and realisation of identified potential for the customer Bringing the management team together in this joint effort, to reduce the NWC Sustain results by using the S&OP process for focus Continuous monitoring and follow up through Stock Monitor tool ICG role and involvement ICG to drive project, conduct analysis, prepare workshops and document solution catalogue Team usually consists of 2-4 consultants in 6-12 weeks depending on scope / size Depending on customer resources, it could range from full scale implementation support to occasional reviews Participation on steering team meetings Usually no planned ICG activities, but not seldom customers come back on additional areas to explore, for help on realising certain activities or just for some stochastic advisory OVERALL APPROACH TO AN IMPLEMENT NWC PROJECT 4 overall phases, being INSIGHTS, EVALUATION, SOLUTION and MONITORING Content Implementation of established plan and realisation of identified potential in phase I Impact tracked in Stock Monitor and focus is secured through the S&OP process Content is designed to fit the customer’s situation and ability to maintain and further develop the setup Solution catalogue Solution success criteria High level road map and target setting Deliverables All selected prioritized initiatives are implemented Stock Monitor fully functional and is used Stock Monitor follows the development of the stock situation, and both corrective actions and newly identified actions are realised
- 41. 41 The solution catalogue is developed with a high degree for key stakeholder involvement in order to secure impact Solution catalogue • Prioritized catalogue ofstrategic levers to reduce NWC • High level description offocus areas, strategic levers and components ofthe solutions. • Easy-to-communicate documentwith reasoning,solutions and impact Solution success criteria • Simple levers • Focus on the mosteffective measures to reduce the NWC Fact pack • Stock analysis,accounts receivable, accounts payable,Customer lead time requirements,supplier performance, mapping ofplanning logic etc. B 1 High level road map and target setting • Indication targets and timeline for impact and NWC reductions ofindividual initiatives 2 INSIGHTS INTO WORKING CAPITAL EVALUATION SOLUTION CATALOGUE & PLAN Solution hypothesis development • Hypotheses developmentbased on the value stream,currentstate fact pack and the build up of data in the Stock Monitor tool C Evaluation of strategic levers and prioritization • Evaluation of impacton cost and customer service. • Prioritization based on impact and ease of implementation D MONITORING & MEASURING Value stream • Product and production task overview • Currentflows and stocking points in the network • Utilization of network • Performance metrics A Stock monitor • Integrated solution securing stock transparency,allowing deep dive analysis,setting stock targets and following up • Developed to fit, by Implement E
- 42. 42 Examples of deliverables; value stream map with flow and steering principles, initial fact pack and solution hypothesis Value stream maps Total set of identified solution hypotheses A B C Stocking points overview related to planning logic A Demand patternB Stock level overview Focus areas, made easyC
- 43. 43 Examples of deliverables; further analysis for hypotheses testing, evaluation, road map and project charters D D 1,2Further analysis related to testing of hypothesis. Evaluation of potential Evaluation of strategic levers and prioritization Road map; overview of strategic levers, plan, idea of initiative D Potential evaluation of specific initiative D Map of solution element in impact and ease of implementation 1,2 One pager of initiative, idea, solution description, targets, milestones
- 44. 44 The project delivers also a Stock monitor, both for the initial analysis but also for continuous transparency, target setting and follow up The Stock monitor is a software, easy to set. It features: • Robust tool developed in MS Access with reports in Excel for easy distribution and subsequent analysis • Integrates easily to various ERP systems • Monitors stock level aggregated by group, business units, sites etc. • Drill down through all level of details to individual material number (pivot table, with total stock data repository) • Target setting by business unit and type • Follow up on stock movements and comparison to targets • View dead stock and monitor progress on campaign actions Easy to add modules for • Optimal calculation of “how much to stock?”, i.e. inventory parameter calculation, safety stock and economic order quantity, based on automatically identified demand patterns (normal, lumpy), lead time and target service level • Global stocking policy for “what are where to stock?”. Based on decision tree logic taking transport cost, stocking cost, risk cost, service requirements, criticality into account Stock development and age profile Stock overview by Group, BU, etc.Easy user interface Tool is simple and adjustable Dead stock overview and developmentEasy to change import files from ERP E E E E E E
- 45. 45 Implement normally foresees three major challenges in the project and has a clear strategy for how to succeed in solving them There are conflicting incentives It’s not a quick fix! One project across several business units 1 2 3 THE LARGEST PROJECT CHALLENGES Run entire project with large involvement from business unit organizations and co-develop hypothesis as well as solutions. Solutions shall be owned by the people who bring them to life Seek solutions of structural nature and set the NWC agenda on top of mind. The Stock Monitor should be a integrated part of daily business. Use S&OP process keep focus. Management support is vital. Run project with multiple work streams identifying only business unit relevant levers to NWC reductions. Stock Monitor will ensure a full group overview.
- 46. Café 2: Visual Management – Implement’s tool Stock Monitor By Peter Bundgaard and Adam Lewestam
- 47. 47 An operational inventory model in MS Excel to track development of demand and inventory levels.
- 48. 48 The inventory model is built on five different key areas and is designed to get operational with from “Day 1”. Demand Input Demand & Variability Calculations Segmentation Matrix Simulated Inventory Profile Stock Value and Potentials Calculation
- 49. Café 3: Reference case – Stock Killer project & One Arla project By Henrik Hahn Sørensen from Arla
- 50. 10 November 2016 50 12,600+ OWNERS THE 5TH LARGEST DAIRY COMPANY MILK INTAKE 14+ BILLION KILO 19,000+ COLLEAGUES 10+ BILLION EU REVENU PRODUCTS SOLD IN 100+ COUNTRIES Goodness comes from within
- 51. 51 We will deliver our mission by following Good Gro 2020 EXCEL in 8 categories & 3 global brands Our identity: Healthy, Natural, Responsible & Cooperative FOCUS on 6 regions WIN as ONE Arla Our vision: Create the future of dairy to bring health and inspiration to the world, naturally Our mission: To secure the highest value for our farmers’ milk while creating opportunities for their growth
- 52. More milk – more opportunities 52
- 53. A change of mindset among key stakeholders was needed to solve the inventory challenge… 53 • Batch sizes • Cost optimization • Capacity utilization • Max EBIT • High delivery service • Trustworthy available volumes • Balancing milk intake • High delivery service • New product mix • Low DIO The key challenge is to balance inventory levels, delivery service level, unit cost and production capability and capacity assets… ...in an environment with where “each stakeholder seek to optimize their business within their area of responsibilities” Supply Chain Planning Trading Sales BG’s Finance • P&L • Investments • Market requirements • Balance sheet Production Capacity Unit cost FG stock levels Service Level 100% milk utilization • High delivery service • Stocks 2nd priority Inv. Optimisa tion
- 54. The Stock Killer Journey 54
- 55. How do Arla perform in terms of working capital? 55 ”Reducing Arla’s inventory is very much like being on an ascending escalator. The natural motion is up, but we are doing everything we can to climb down. And that is hard work…”
- 56. Café 4: Inventory – the definition & calculations behind Stock Monitor By Elin Aalders Hemmingsen
- 57. 57 Aligning the inventory level to the target stock manually often results in compromising the service level Inventory level Target service level Service level Order size Lead time Variance Target stocklevel Inventory level = target level Targetservicelevel NOT achieved A manually maintainedstock leveloften results in a compromisedservice level With an inventory model,there is a fixed link betweenthe targetservice level and the resulting inventory leveland service
- 58. 58 The inventory optimization model Segmenta- tion Stable/ Lumpy items Inventory Calculation Safety Stocks/ ROP Impact and reporting Potential Inventory Segmentation Model Tuning of parameters Diff. scenarios ICG Inventory tool System Defining service level targets for different groups of products Possibility to simulate different scenarios e.g. shorter lead time, higher target or similar Projected potential of the proposed inventory parameters
- 59. 59 The Implement Segmentation Matrix is an excellent tool for finding the best model for computing a re order point Low High Low High Lines/lead time CV(demandduringleadtime) Erratic Intermittent Stable/ predictable 60% 8 lines/lead time The cut-off values are based on empirical experience The Implement Segmentation Matrix uses frequency and variance to segment items into groups and find the best possible model to compute the re-order point. Stable/predictable – This category contains items characterised as having low variance and frequent demand incidences. Items in this category do not give rise to any forecasting or inventory control issues. Use normal distribution. Erratic – This category contains items characterised as having high variance and frequent demand incidences. Use normal distribution with spike order procedures and reduce possible bias. In special cases, gamma distribution is used. Intermittent – This category contains items characterised as having low variance and low and infrequent volumes. Use normal distribution. Lumpy – This category contains items characterised as having low and infrequent volumes and high variance when demand occurs. Items that belong in this category are the most difficult to manage. Use Compound Poisson distribution with spike handling procedures. For lumpy items, we can further segment the items into subgroups with the same characteristics. This is done by finding the ratio between the average and the median order size for each item.
- 60. 60 This leads to the following decision tree for segmenting items into the groups of distribution Product age* < 4 months Lines in LT > 8 No Erratic (Stable) Yes Lumpy No Stable Yes Product age < 6 months New product Yes 𝐶𝑉𝐿𝑇 <= 60% No Mean/ median <1,1 Slightly lumpy Mean /median <1,7 Mod. lumpy Mean /median <3 Highly lumpy Mean /median >=3 Manual review
- 61. 61 Understanding demand and supply variability is important to estimate safety stock – and to reduce it … Demand and supply variability: 1. Demand variability during lead time. The safety stock must cover the demand variability during lead time. A longer lead time leads to higher safety stocks. 2. Lead time variability. The total replenishment lead time for supply may vary due to delays in production, transport, quality control etc. Higher lead time variability leads to higher safety stocks. The lead time variability is measured for a group of materials. A good way to understand what drives lead time variability is to measure plan adherence and register why changes occur. 3. Supply variability (order size). The supply may vary due to component stock-out, capacity issues, production variability, quality issues etc. A higher supply variability leads to higher safety stocks if the supply order size is lower than demand during lead time plus safety stock. LT Time Quantity Safety stock ROP 2. Lead time variability 3. Supply variability (order size) 1. Demand variability during lead time
- 62. 62 For Lumpy materials the re-order point is found by simulating the connection between re-order point and service level Finding the re-order point: For the lumpy materials the safety stock and re-order point resulting in certain service level cannot be found via well known formulas. In order to find the link between the re-order point and the corresponding service level we need simulations that link the re- order point and the resulting service level Simulation: The service level is simulated for combinations of the rate of usage and a number of potential re-order points. This simulation is done once in Anylogic and the outcome is a fixed table that can reside in an Excel document. The re-order points are found via lookups in Excel. No further simulations are needed going forward. 0 1 2 3 4 1 2 3 4 5 6 7 Poisson Distribution # Orders per period #Periods Usages during leadtime Re-order point Target service level 0,5 6 99% 1,0 7 99% 2,0 9 99% 3,0 10 99% 4,0 12 99% 5,0 13 99% 6,0 15 99% 7,0 16 99% 8,0 18 99% 9,0 19 99% 10,0 20 99% Usages during leadtime Re-order point Target service level 0,5 6 99% 1,0 7 99% 2,0 9 99% 3,0 10 99% 4,0 12 99% 5,0 13 99% 6,0 15 99% 7,0 16 99% 8,0 18 99% 9,0 19 99% 10,0 20 99%
- 63. Café 5: Implement’s view on Simple forecasting By Andreas Kloow
- 64. 64 WHAT is a good sales forecast to support medium-term business processes – and HOW to obtain this 1. Unbiased 2. Stable 3. Transparent 4. Market & customer insights 5. Minimize work load for sales & marketing A good medium-term sales forecast The six HOWs: 1. How to structure a clear sales forecasting process. 2. How to build a stable and transparent statistical forecast – that is easy to understand. 3. How to incorporate insights from Sales and Marketing with minimum workload. 4. How to handle sales forecasts with high uncertainty and impact. 5. How to handle new product introductions with high impact. 6. How to continuously improve the sales forecast.
- 65. 65 Don’t overcomplicate things – these are the few elements we need in order to control the statistical forecast. Not More, Not Less. Constant Forecast Group Seasonality Step Changes Baseline Forecast History Cleaning The biggestchallenge with statistical forecasting is complexity,leading to lack of transparency and a lot of frustration on fitting various parameters trying to understand the outcome. As scientific literature shows it doesn’tcreate any betterresult to use complexalgorithms, so why bother? Based on our experience,we actually only need a few simple elements in order to get a solid baseline statistical forecastas illustrated below. Long Term Trend
- 66. 66 “Relying only on statistical forecast is like driving only looking in the rear-mirror” 5 issues with statistical forecasting: 1. Lumpy and sporadic demand can’t be statistical forecastedwith a reasonable forecast error (<60%): • This is typically due to few unpredictable sales orders per month and / or high variability of ordersizes 2. Optimizing of statistical forecasting models & parameters (“bestfit”) is based on wrong assumptions: • Step change or trend in the past always gives step change in the future (this might have beenthe reality in the 1960’s whenthis method was invented, but not any more) 3. Statistical forecasting of seasonality and trend does not work when the demand variability is higher than 10% 4. Complexstatistical forecasting methods give lower forecastaccuracy than simple ones as shown by Professor J. ScottArmstrong at Wharton University: “If you neverthelessuse forecastsfrom complexmethods to help you make decisions,expectto be confusedabouthow the forecasts were made and an accuracy penalty ofmore than one quarter25%)”; see www.simple-forecasting.com 5. All statistical forecasting methodswill create bias if: • The sales has a trend • The sales had a step change “Relying only on statistical forecast is like driving only looking in the rear-mirror”
- 67. 67 Proper data quality is a prerequisite for accurate forecasts • We need to remove any outliers and event from the sales history, such as promotion, stock-outs and exceptional sales • We should not clean random noise or natural variance, because these will be handled via a simple smoothing forecast model. We should only clean for significant extremes like promotions or outliers • We need to establish an easy, simple and non-time consuming approach for cleaning history, supported by alerts and warnings • Statistical models will not be able to forecast these events, and thus promotions, tenders and other uplifts/drops needs to be added on top of the baseline forecast based on input from Sales and Marketing If we are to achieve an accurate and reliablestatisticalbaseline forecast,we need to ensure that the input - in the form of historical sales - are cleansedfrom significantoutliersand events,which otherwise will lead to a biased and inaccurate forecast Time Volume Uncleaned History Time Normal Sales Volume Outliers / Events The sales history used for the statistical forecast must consist of “normal sales”
- 68. 68 Simple exponential smoothing (SES) is robust, simple and does the job The SES is an excellentforecastmodel because of its smoothing factor alpha parameter (α). It is basically a weighted moving average weighing the mostreason observation more.This means that we get the stable level of the forecast from the average, and the reactivenessfrom the weighted alpha factor. *Makridakis’ forecasting competitions (M1, M2, M3) • Creates a constant future forecast, with the alpha value controlling the “reactiveness” of the model • Simple Exponential Smoothing is better than moving average because exponential smoothing reacts faster on trend and step change than a moving average with same aging and it has almost similar forecast variability; a 12-month moving average has the same aging as alpha = 2 / (12 + 1) = 0.15 • Often, optimisation of statistical models and parameters makes the planning non-transparent and, worst of all, it increases forecast variability. Forecast variability is noise and is amplified in planning and gives more unstable plans • Simple Exponential Smoothing has scientifically been proven to perform at least as good as more complex models* • We recommend having a low alpha value to create a stable forecast, and handle changes in demand with the step change functionality concept Time Volatile demand 0 100 200 300 400 500 600 700 800 900 1 000 1 100 Demand Forecast Volume Stable Forecast The SES has smoothing factors (α) and is an excellent forecast model
- 69. 69 Seasonality cannot be ignored We cannotignore seasonality since it can have a high impact on the decisionand thus forecast - However, using traditional methods forcontrolling seasonality might only lead more complexity,less trust and thereby not achieving the intended forecastaccuracy Dealing with seasonality can be very tricky • The “traditional” statistical models e.g. Holt Winters model, can be complex and the output can be difficult to understand due to its various input variables (Alpha, Beta, Gamma) • In traditional models, products needs to have at least 1 year of history in order to accept a seasonal forecast - This challenges New Product Introductions (NPI) • Seasonality pattern can be very difficult to spot on a detailed SKU level, since noise and variance can obscure the pattern We recommend using group seasonality logics! • Groups seasonality is simple and easy to understand since it is basically just an index to add on top of the constant baseline forecast • We can use this for ALL products, even NPI with few or non periods of sales • We accumulate sales history across a range of products, which, due to the law of large numbers, results in a much more clear and smooth seasonal pattern with less noise and variance. Group Seasonality overcome these challenges Calculate the seasonality indices based on groupings, and apply it to the constants statistical forecast 0 0,2 0,4 0,6 0,8 1 1,2 % 0 0 0 1 1 1 1 Group seasonal index Constant FC (SES) Baseline FC w/ season.
- 70. 70 We need to incorporate market & customer insights on significant step changes Significantstep changes of demands create huge challenges for the statistical forecast,and can be a source to a lot of manual effortto manipulate history or adjusting future sales.We need to handle this in a simple manor. Time Volume Reactive Period Step Change Today Reactive step change – step changes in the past Demand Forecastw/o step change Forecastw/ step change • Since statistical forecast always is reactive, it can never foresee step/level changes caused by e.g. new listings, customers, etc. Sales needs to provide these information. • Statistical forecast needs some periods of observations to “catch up” – a reaction period. The step change logics resets the forecast at the right level and thus achieve a better and more accurate forecast. Proactive step change – step changes in the future Time Volume Step Change Today • Expected step/level changes should be provide by Sales or based on POS data, and added to the forecast. • When the step change period is over in the past – then it resets the forecast at the right level. • This is added on the aggregation level which makes sense e.g. customer, category or material, customer type. Demand Forecastw/ step change
- 71. 71 Long-term trend must be handled by a combination of statistical forecasting and input from Sales and Marketing Small monthly trends in the market – growth or decline – have a significant impact on our planning, and thus we need to included these expectations to our forecast. We cannot rely on past statistical trends, because then we’re already too late. We need support from Category and Marketing. Include long term trends on aggregate level Time Volume Trend Expectations Today Demand Forecastw/ trend • Add long term trends on a high aggregation level – it is impossible to catch the trends in the details • Short term market changes should be managed by event planning, uplifts and adjustments – not market trends • Include Marketing in the trends discussions, they are the ones with the long term expectations • Use historical trend patterns as input to the discussion with Marketing, don’t rely on them solely Future trends will rarely mirror past trends • Due to different stages in the Product Life Cycle (PLC) we cannot expect past trends to mirror to the future • Trend has a great impact on the long-term forecast, but not so much on the short-term forecast • Trying to catch short term trends is time-consuming and difficult du to the random variation in demand. • Optimizing trend parameter algorithms via the traditional 𝛽 value (Beta), is another element to increasing complexity and instability.
- 72. Café 6: How to involve Sales in project & Efficient scenario planning By Thomas Holm
- 73. 73 The forecast is used by various stakeholders for different purposes on multiple aggregation levels Financial forecast DivisionCategory Sales region Product group Product Customer SKU Account Manager ‒ Financial sales forecast per customer to reallocate promotions and sales activities and resources BU VP: ‒ Financial forecast to overall resource allocation Strategic purchasing - External capacity, component & raw material requirements to renew & adjust sourcing CxO’s: - Financial predictability Production/Supply Chain - Capacity load on key resources to adjust capacity Category Manager: ‒ Category forecast to adjust marketing activities & resources Master scheduling & Purchasing - Mix forecast to plan production and purchasing SalesDirector ‒ Financial sales forecast per category to reallocate marketing and sales activities and resources
- 74. 74 How to define on which level in the product hierarchy & time horizon for sales & marketing to focus on to minimize their work load Business area Category Product line Product Group Material Decision horizonsbasedon supply chain scalability and flexibility Decision horizonsfor sales & marketing activities Promotion x x Marketing campaign x Sales meetings x Change product grp focus x Time Plan the Volume Manage the Mix Suicide Quadrant Time AggregatedDetailed Plan as aggregated aspossible to supportthe business decisions Avoid the suicide quadrant AggregatedDetailed
- 75. 75 WHAT is a good sales forecast to support medium-term business processes – and HOW to obtain this 1. Unbiased 2. Stable 3. Transparent 4. Market & customer insights 5. Minimize work load for sales & marketing A good medium-term sales forecast The six HOWs: 1. How to structure a clearsalesforecastingprocess. 2. How to build a stable and transparentstatisticalforecast – that is easy to understand. 3. How to incorporateinsights from Sales and Marketing with minimum workload. 4. How to handlesales forecastswith high uncertaintyand impact. 5. How to handlenew productintroductionswith high impact. 6. How to continuously improvethe sales forecast.
- 76. 76 Sales & supply chain impact segmentation helps to focus sales forecasting efforts where it creates the largest impact High Low SalesforecastImpact Demand variability / unpredictability Low High • Trends • Significant step changes • Review total sales forecast • Trends • Scenarios • Review total sales forecast No sales forecast focus No sales forecast focus High Low Sales forecast impact on sales Low High Sales forecast impact on supply chain Product group sales forecast has high impact on sales in country, region Product group sales forecast has high impact on both sales & supply chain No sales forecast focus Product group sales forecast has high impact on supply chain decisions
- 77. 77 The process for scenario planning consists of 5 steps • Identify critical uncertainties that drive change/assumptions • Monitor and examine the current environment to determine which are the most important factors that will decide the nature of the future environment within which the organisation operates • Link these drivers together to provide a meaningful framework usually with 5-10 logical groupings of drivers 1. Identify drivers for scenarios 2. Produce scenarios 3. Describe possible choices 4. Assess risk and impact 5. Create decision proposal • Identify and describe scenarios based on different assumptions, and understand if they are interlinked. What does each assumption represent? • Reduce 2-3 realistic core scenarios and define probability for each • Identify and describe the choices for 2-3 core scenarios • Design the “game board” with combinations of scenarios and choices • If scenarios are not interlinked, a game board for each group of scenarios can be designed • List and assess risks (probability X consequence) for each combination of scenario and choice • Analyse impact on key metrics such as profit, cost, networking capital, utilisation, lost sales, gained sales/opportunities, lead times, service levels etc. • Create a one-pager with supporting appendices • Include recommendations, plan for monitoring and risk mitigation proposal
- 78. 78 The game board is a simple and powerful tool to support scenario planning The financial, sales , operations, supply chain consequence of each combination of scenario & choice Best case Base line Worst case Scenarios No change choice 1 choice 2 Choices
- 79. Implementconsultinggroup.com Implement Consulting Group Implement Consulting Group is a leading Scandinavia based management consultancy, specialised in driving strategic transformations with a strong differentiator on “making change happen” – delivering documented Change with Impact. Stalk us on: