Best practices for food retail supply chain management (2023)

1. Grocery retail strategies fail if not supported by the right supply chains

Food retail is a tough and turbulent market. Grocery has never been easy, but the current business transformation is more dramatic than anything we've seen in decades, especially with COVID-19 further accelerating many of the trends shaping the market.

Grocers have to deal with several important trends simultaneously:

• Innovative development of online ordering and order fulfillment capabilities.From online ordering to home delivery and curbside pickup, the speed of omnichannel development has made operational efficiency particularly challenging for food retailers due to the low-value product mix and high cost of handling fragile products of various sizes that often require temperature control. But while many struggle to make e-commerce profitable, few food retailers can afford not to go online.
• Discounts have shown the power of operational efficiency.Since the financial crisis of 2008, discounters have increased their market share, demonstrating that customers value low-cost private label products and well-curated assortments, even at the expense of a rich selection. Controlling costs always provides a competitive advantage, especially when consumers are reluctant to spend, so streamlining operations to improve efficiency should be a key part of any grocer's future strategy.
• Grocers also have to compete with the food service industry.Restaurants, meal delivery services and takeout take up an ever-increasing share of consumers' wallets. As a result, food retailers are turning to ready meals to increase relevance and high-value packaged products to improve margins in a tough environment. Some even open restaurants in stores. However, poorly implemented, these initiatives also introduce new opportunities to waste money on costly food waste.
• The focus is on provocatively fresh products with a short shelf life.Healthy eating trends have even forced discounters such as Aldi and Lidl to improve their fresh offerings to include organic meats and freshly baked bread. In their quest for growth, these formerly hard discounters are abandoning the highly efficient comfort zones previously built on simplicity, standardization and high volumes. Increasing complexity as they experiment with fresher produce, smaller store formats and local varieties will test their grocery supply chains.
• The drive towards sustainability.Consumers' concerns about the environment are only growing and they expect their grocers to develop sustainable practices as well. Many leading retailers have already committed to reducing their carbon emissions, but very soon sustainability will move from 'positive messages' to roles at the table. Retailers who do not take significant steps to reduce waste and emissions risk their reputation. The good news is that sustainability, efficiency and profitability go hand in hand in many cases.

All these trends represent challenges and opportunities, but it is clear thatSupply chain management will be at the heart of the successes - and failures - we see in the future of grocery retail.All food retailers today must make difficult decisions about where to place their business bets. Whatever strategy a retailer chooses, it has little chance of success unless it develops its grocery supply chain accordingly.

Successful food retailers must master both the lean, high-performance grocery supply chains and the flexible and responsive supply chains required for fresh produce. Additionally, many of them will need to manage the complexity of working with multiple store formats while offering multiple fulfillment options.

To achieve this, marketers need the right planning tools at their disposal. In addition, they must understand how to implement them.

• Fresh productstend to have a high risk of shrinkage and wastage, so it is very important to accurately forecast demand and replenish accordingly. For fresh produce, the planning process must be granular enough to capture even the smallest changes in demand, and the supply chain must be flexible enough to adapt to changing demand.
• Central trade and other products with a longer shelf lifeis the key to efficient goods management and inventory flow optimization. While accurate forecasting is still necessary, replenishment for long-duration products does not need to be exactly aligned with demand at all times, allowing for a smooth flow of inventory through the supply chain for efficient capacity utilization. Setting up in-store delivery to enable one-touch replenishment or truck-to-shelf delivery is key to increasing profitability.
• Multichannelit has grown to the point where it can no longer be used as an "add-on" service exempt from standard performance requirements. Online fulfillment actually emphasizes the need for high quality and freshness because end consumers cannot evaluate the products themselves. Additionally, with many grocers today experiencing significant losses in online ordering, the continued growth of omnichannel underscores the need for significant improvements in operational efficiency.

In this best practice guide, we will highlight key approaches to increase responsiveness and efficiency in food supply chains. You'd be hard-pressed to find a single retailer that implements all of these best practices. Instead, we encourage you to prioritize the most feasible and effective areas of growth from your perspective.

2. Harness the power of AI to optimize your grocery supply chain

Today's grocers collect vast amounts of data about transactions and interactions with consumers, online and offline. This is precisely why food retail is a perfect fit for artificial intelligence (AI), which enables the use of this data for faster and more accurate decisions. This is an invaluable asset in an industry where retailers must control millions of merchandise flows and accurately match supply with demand across hundreds or even thousands of locations on a daily basis.

Tech companies can eagerly position their AI algorithms as “smart” by making them as human-like as possible—even giving them human names like Siri, Alexa, Einstein, or Watson. However, keep in mind that AI is not a person. AI isn't even "it".

We are still a long way from a general artificial intelligence capable of creatively solving ill-posed problems. But we are making great progress in specialized AI that solves well-defined problems (such as image recognition algorithms) and combinations of several types of specialized AI (such as self-driving cars).

Figure 2:What is the best robot for washing clothes? Specialized artificial intelligence is becoming more common and is often used to run applications that don't seem particularly smart at first glance. (Illustration inspired by excellentblog post by Ben Evans.)

Retailers don't need "AI" - they need to use enough AI algorithms in their analytics toolbox to complement "legacy" technologies like statistical analysis and rule-based heuristics. Machine learning algorithms, for example, take into account hundreds of factors that potentially affect demand when forecasting retail sales (Section 3.2), something a human demand planner could never achieve.

The application of artificial intelligence in food retail is not limited to demand forecasting, although. Retailers can reap even greater benefits by leveraging AI to optimize their entire range of operations – from order optimization (see section 4.1.1) to workforce optimization (readwhite paper) and optimized reductions (transwhite paper).

3. Demand forecasting is the engine that drives your grocery supply chain

Demand forecasting is the engine that drives your grocery supply chain. However, despite the technology available, many big names in grocery have yet to truly begin their data-driven forecasting journey.

3.1. Detailed data-driven forecasting is essential for grocers

UNorth America Grocery Survey 2020, only 52% of respondents said they could forecast at the daily store SKU (SKU = Stock Keeping Unit) level. We wanted to make predictions on a daily basis, but it just wasn't possible. Few retailers today can individually forecast the online orders they receive in their stores, which inevitably leads to capacity management issues, especially after the significant increase in online demand fueled by the COVID-19 crisis.

Detailed forecasting isn't just a best practice—it's essential in today's grocery retail business. Without detailed forecasting, it is impossible to properly place inventory in the supply chain to increase sales and reduce waste. Detailed forecasts are also the basis of planning for resource and capacity management, and should therefore be considered a prerequisite for profitable operations.

3.2. Machine learning has great value in forecasting grocery demand

What started with a few predecessors like RELEX became mainstream over the years:machine learning to predict retail demand;.Currently, retail technology vendors are either implementing machine learning or rushing to update legacy systems to offer it.

Machine learning enables a forecasting system to automatically learn and improve its predictions using only data, without the need for additional programming. As marketers generate massive amounts of data, machine learning technology is quickly proving its worth.

Of course, machine learning algorithms aren't new - they've been around for decades. But never before have they been able to access as much data or data processing power as is available today. While grocers may have had trouble updating their forecasts quickly in the past,big data processing and in-memory technologynow activate millions of budget forecasts within a minute.

Machine learning makes it possible to incorporate a wide range of factors and relationships that affect demand into your grocery sales forecasts. This is extremely valuable because the weather data itself can consist of hundreds of different factors that can affect demand.

However, here's the caveat: it really does make a differencehow machine learning is performed;. Although grocers can collect vast amounts of data, their data is often quite limited at the store and SKU level. Slow moving products may not provide enough sales transactions to study. master data for past promotions and product impressions may be missing. and as products fluctuate in and out, the data may not be able to do so for the exact SKU they are trying to predict. Machine learning prediction must be fine-tuned to be less vulnerable to data problems, which would cause a "garbage in, garbage out" scenario.

Applied in the right way, machine learning effectively solves common challenges with retail data to provide benefits in all aspects of grocery demand forecasting: 1) capturing recurring demand patterns caused by daily and seasonal 2) predicting the impact of offers , of price changes and other internal business decisions, 3) predicting the impact of local traffic, events, weather and other external factors, and 4) even detecting when unknown factors (such as an unrecorded change in the way a product appears) may affect demand.

In simple scenarios, such as forecasting predictable recurring demand fluctuations, machine learning provides only equal or slightly better accuracy than traditional time series demand forecasting. However, when dealing with complex situations such as overlapping promotions or sales cannibalization, machine learning-based forecasting clearly outperforms traditional forecasting approaches.

3.3. Typical challenges of forecasting demand for supermarkets, discount stores and convenience stores

Then we'll discuss how you can overcome some of the typical forecasting challenges faced by supermarkets, discounters and convenience stores.

3.3.1. Forecast demand for new products and stores

Because machine learning relies on finding patterns in historical sales data, new products without historical sales data can prove challenging. Fortunately, there are additional routines to improve the management of new product imports.

When introducing a new product, the most common approach is to assign it a reference product to use as a blueprint for its sales pattern until the new product has accumulated enough historical data of its own. However, in grocery retail, the number of new products per year can be overwhelming. Because of this, manual identification and placement of reference products is impractical or at least very inefficient.

It is much more efficient to usea system that can automatically compare product features(eg product group, brand, pack size, color or price) to assign the most relevant reference product. The same approach can, of course, be applied to finding suitable reference stores for opening new stores.

3.3.2. Predicting the impact of promotions and price and display changes

Your own business decisions as a retailer are also a major source of variation in demand, from promotions and price changes to adjustments to how products are displayed in your stores. Despite marketers themselves planning and controlling these changes, many in the industry are still unable to accurately predict their impact.

uNorth American Grocery Study 2020, 70% of respondents said they cannot consider all relevant aspects of a promotion—such as price, type of promotion, or in-store exposure—when predicting promotional increases. But I wish they could.

Machine learning allows retailers to accurately model a product's price elasticity, or how strongly a change in price will affect demand for that product. However, price elasticity alone does not capture the full effect of a price change.

Pricing a product relative to other products in its category also often has a big impact. In many categories, the lowest-priced product captures a disproportionately large share of demand. Machine learning based demand forecasting makes it very easy to consider a product's pricing position, as shown in Figure 8 below.

However, machine learning does more than just exploit price data. With predictive machine learning, grocers can accurately predict how promotions will perform by considering factors that include, but are by no means limited to:

• The type of promotion, such as a price reduction or multiple purchases
• Marketing activities such as circulars or in-store signage
• Reduction in product prices
• In-store display, such as displaying a promoted product in a header or table

3.3.3. Accounting for cannibalism and halo effects in forecasting

It is quite common for a promotional increase for one product to actually decrease sales for another. For example, if a supermarket that sells "HappyCow" and "GreenBeef" brands of lean organic ground beef puts the HappyCow product on sale, more people will buy it — but it's possible that some of the underlying demand for GreenBeef will shift to HappyCow . If they do not reduce the demand forecast for the GreenBeef product, planners run a high risk of stockpiling, which leads to waste.

For most products in stores, such as canned foods or cereals, cannibalism is not a big problem. If demand temporarily drops, the replenishment order for the cannibalized product will simply be placed later than usual. However, when working with fresh products, and especially products that have a limited number of direct substitutes, it is necessary to consider the forecaststhe impact of cannibalismto avoid overstocking and spoilage.

Manually adjusting forecasts for all potentially cannibalized items is simply not feasible in most retail environments because the number of products to adjust is simply too large. Typically, the patterns are quite specific to the variety of individual stores as well as buying patterns. The ability of machine learning algorithms to automatically recognize patterns and adjust forecasts accordingly adds tremendous value when it comes to cannibalizing sales.

The other side of cannibalism is, of course, the halo effect, or when the promotion of a HappyCow product also drives sales of related products outside the "ground beef" category. Hamburger buns, for example, have an obvious and predictable association with ground beef.

Unfortunately, the halo effect can be so widespread across a variety that identifying each affected product becomes more or less impossible, even with machine learning. Think onions, chips, beer, watermelon, taco meal kits, salad dressings, oyster crackers, corn, Worcestershire sauce, soy sauce, or any other item that customers can associate with dishes with minced meat. But even if prediction systems cannot identify all possible halo relationships, they should make it easier for planners to adjust predictions for relationships they know exist.

3.3.4. Assessment of the effect of weather and other external factors on the request

External factors such as weather, local concerts and races, and competitor price changes can have a significant impact on demand.

It often seems intuitively easy to understand how something like the weather, for example, can affect sales: hot temperatures increase ice cream sales, rain increases demand for umbrellas, and so on. However, when you look at a retailer's overall offering, it gets more complicated.

Using local weather data and forecasts to increase the accuracy of demand forecasting is a great example of the power of machine learning. Machine learning algorithms can automatically detect the relationship between local weather variables and sales for individual products in individual stores.

Machine learning algorithms can similarly be used to take advantage of a wide range of data sources over time, helping marketers identify relationships between external variables such as local sporting events or concerts and local sales for specific products.

In grocery retail, the most useful external data sources include:

• Local weather data and weather forecast (reada case study)
• Number of passengers and forecasts at traffic centers (transa case study)
• Information about past and future local events, such as football matches or concerts
• Competitor pricing information

3.3.5. Dealing with unexplained changes in demand

In brick-and-mortar retail, local conditions—such as a direct competitor opening or closing a nearby store—can cause demand to shift. Unfortunately, information about the agent causing this change may not be recorded in any system. Sometimes even the internal decisions of the retailers themselves are not recorded, such as adding products to a special display area in the store.

Fortunately, machine learning can help in such situations. Machine learning algorithms can tentatively set a "tipping point" in a prediction model and then track subsequent data to disprove or confirm a hypothesis. This allows forecasts to quickly and automatically adjust to new levels of demand.

Consider the example in Figure 12 below, which shows the impact on sales when store staff created a tabletop display in addition to regular product shelf space. Although no one recorded this change in the master data, the system could easily track the impact of demand as a factor in how the product was displayed in the store.

3.3.6. Incorporating the designer's expertise into demand forecasting

If you want to stay competitive in food retail, machine learning is something you need to embrace, but you also need to understand its limitations. Automating much of demand forecasting is not only desirable—it's actually quite possible with machine learning. But the COVID-19 crisis has made it clear that there will always be circumstances in which system-generated forecasts will be wrong (althoughsome systems can retrieve morefaster than others).

It was a pandemica particularly extreme shock to the system, but in a business as dynamic as retail, there is always the risk that predictions based on how things were in the past may not accurately reflect how things are now or will be in the future. No demand planning solution, no matter how advanced, can completely avoid forecasting errors.

That's why it's so important that experts in retail demand planning teams can fully understand forecast errors. If their system provides transparency in how it forms the forecast, retail professionals can quickly understand and correct any errors they see in it.

Too many retailers rely on "black box" forecasting systems that can collect all kinds of data to make accurate forecasts, but lack transparency. Black box systems can destroy your business - or at least the effectiveness of your programming - for a number of reasons. First, occasional extreme prediction errors can cause much more damage to performance than smaller, more frequent errors. Second, when demand planners cannot account for forecast errors, it erodes their confidence in all forecast calculations, leading to increased double-checking and manual forecasting, and undermining the entire goal of leveraging computing power to automate your forecasting .

That's why best practice marketers understand the value of transparency. Even when the system does the heavy lifting, planners must always be able to understand and control how their predictions are generated.

4. Improve food replenishment for better availability, wastage and efficiency

The quality of a grocer's replenishment process has a direct impact on its top and bottom price.

Predicted high-quality grocery replenishment consistently translates into the following benefits:

• Increased revenue from better shelf availability increasing sales by up to several percent
• Up to 40% lower markdown and spoilage costs because supply more accurately matches demand
• Optimized inventory flows enabling up to 30% reduction in the cost of moving goods in distribution centers and stores
• Much more efficient use of capacity at the transport, storage and manual stages of the entire supply chain

Nevertheless, in oneNorth America Grocery Survey 2020, only 24% of respondents implemented some level of predictive store replenishment and only 7% implemented it extensively. Store replenishment is certainly one area where the operations of many supermarkets, discounters and convenience stores are currently well below best practice.

4.1. Replenishment with fresh food requires detailed planning and execution

For fresh produce,well managed trade replenishmentIt is vital to find the optimal balance between the risk of lost sales margins caused by inventory shortages and the risk of wastage or write-offs reaching already thin margins.

Although traditional supermarkets have decades of experience in the fresh produce business, many still do not excel in this area. Their supply chains are reactive enough to support frequent deliveries, but their replenishment planning is not up to par.

According to grocers surveyed in North America, the annual value of spoilage averaged over $70 million, and for the largest companies offering a wide variety of fresh produce, up to several hundred million dollars annually. A 10-40% reduction would mean annual savings of between US$7-28 million. It's not only possible, it's what modern food retailers are expected to do to reduce their carbon footprint and make their business more sustainable. (Read more about reducing the carbon footprint of supermarkets, food outlets and online storeshere).

4.1.1. Reconciliation of waste and lost sales

For so-called ultra-fresh products, i.e. short-life items that must be sold the same day, 100% availability means that there will always be waste or depreciation unless the forecast is consistently flawless at the day, store and product level. This means that very detailed control is required to find the optimal balance between the risk of shortages and the risk of waste. Other fresh produce faces a similar challenge, only slightly less severe.

The demand for a product in a particular store usually varies between different business days. For some stores and products thisweekday variation in fresh replenishmentit can be very dramatic. This means that the same safety stock does not correspond to all working days when it comes to products with a short shelf life.

Roast beef, for example, tends to sell a lot more before the weekend than after. For roast beef – even when the daily forecast is accurate – a static safety stock level leads to 1) excess stock after the weekend, with increased risk of wastage, and 2) dangerously low safety stocks during the weekend, with increased stock risk .

To find the right balance between waste risk and inventory risk, safety stocks must move up and down according to expected sales volumes and forecast errors for different days of the week.Good retail planning systems automate this type of precise safety stock optimization.

Actually,the best retail planning systemsimprove optimization not only by enabling dynamic safety stocks, but by optimizing each order based on a cost-benefit calculation that balances the risk of wastage and the risk of stock-out. Such machine learning algorithms minimize overall lost sales margin and waste costs.

The cost function should be adjustable in the weight it places on shelf availability versus scrap to allow consideration of the strategic roles of key categories and items and whether there are many or limited opportunities for substitution within product categories.

Figure 14:Best-in-class retail planning systems optimize each order based on a cost-effectiveness budget that balances waste risk with out-of-stock risk.

When managing the replenishment of stores with fresh products, it is very important that all calculations and optimizations are done automatically. It is impossible for any human to keep track of all the factors that affect demand, such as day-of-the-week variations (eg seasons, weather and promotions), as well as all the factors that affect replenishment (eg delivery schedules , lot sizes and probabilities at the daily level of waste and inventory) for hundreds or thousands of items per day in one store, let alone hundreds of stores.

However, it is equally important thatforecasting and replenishment systemit doesn't turn it into a black box. Active analytics enable supply planners to easily identify and correct exceptions such as historical or forecasted waste or poor availability.

Examples of typical exceptions to the fresh food refund are:

• Too many order lots cause waste in stores.Sometimes batch orders, such as packing boxes, are so large relative to store demand that each product delivery will result in waste. To effectively address this problem, your supply planners must be able to determine whether it is a problem in a few or many stores, what the financial implications are, and whether the problem can be mitigated by targeting replenishment on specific days of the week . such as ordering product only for the weekend .
• Excessive allocated shelf space causes waste in stores.Sometimes minimal visuals, designed to keep windows attractively stocked, encourage overstocking and waste of fresh produce. Supply planners must be able to determine if the problem is isolated to a few stores with low demand or if there is a more widespread problem with the planograms in use.
• Systematic poor availability or high wastage on certain business days.Systematic patterns of poor performance on certain days of the week, such as higher-than-average waste on Mondays, are not unusual. To solve this problem, supply planners need to understand the root cause of this problem. For example, there may be process issues, such as controlling sales staff's sales deadlines and recording waste on certain days of the week, that need to be considered in replenishment planning.

Automation radically reduces the time spent on routine tasks in store replenishment planning. At the same time, it multiplies the influence of your most experienced special processes. If store replenishment is not automated, your best supply chain analysts are of limited power. They can review successes and failures in the mirror and try to translate some of their findings into action in stores with the help of a field training team.

When store replenishment is automated and replenishment planning is centralized for an experienced team, your designers can make a visible difference to hundreds of stores, almost instantly, just by adjusting replenishment settings.

4.1.2. Shops were turned into kitchens

As consumers increasingly seek convenience, takeout and takeout solutions are on the rise. Many stores are converted into kitchens where sandwiches, hot dogs and salads are made.

Traditionally, on-site products were considered separate items that had to be handled manually in stores. However, with the growing demand for ready-to-eat meals, the importance of on-site production has become much more pronounced and critical to the profitability of food retailing.

The process of replenishing ready meals is not so different from replenishing other products sold in the store. It's just a bit more complicated. The demand for final products – meals – must be translated into ingredients used to produce the final products. Replenishment calculations must be made for each component taking into account each component's lead time and available inventory.

Basically, the process is as follows:

1. Demand forecast for the final product.
2. Translate the estimated demand for finished products into the estimated demand for ingredients needed to produce the finished product. This requires knowledge of the recipe (sometimes called a bill of materials, a term borrowed from the manufacturing industry), as well as the yields of the various ingredients. If a sandwich calls for 1.3 ounces of lettuce, calculations may need to be made at 1.7 ounces of lettuce per sandwich to account for the portion of lettuce not suitable for use in the sandwich.
3. Calculate the estimated demand for each component. The total demand for an ingredient often reflects its use in many finished products.
4. Calculate the quantity required to replenish each component based on lead time, on-hand inventory, potential incoming orders, estimated demand, and safety stock target.

Sometimes the ingredients included in the recipe are made up of other ingredients, such as special mayonnaise or mustard produced on site. In these cases, it is necessary to make similar calculations for several recipe levels. A daunting task for any human, but quite doable for a computer.

4.1.3. High frequency replenishment for super fresh products

For super fresh produce, many retailers choose to deliver to stores multiple times a day to guarantee freshness. Similarly, products produced on site are usually made in several batches throughout the day. This is especially true for the growing category of baked goods in stores, which ideally should still be warm when the customer receives them. In addition, the new trend of food retailers to open small stores in urban areas has made several replenishments per day mandatory due to lack of storage space in stores.

Placing a large number of orders per day or making an optimal baking plan per day requires taking into account the fluctuations in demand associated with the working day and within the day. For some products, the intraday or so-called intraday demand pattern will follow the general movement of buyers for that day. for other products, such as lunch items, demand is more influenced by how the items are planned to be consumed.

Again, manually tracking daily and intraday demand patterns is a fairly complex and error-prone process. However, many retailers still rely on their store associates to figure it out on their own. This is a high-stakes bet, as ultra-fresh produce inevitably has a big impact on how consumers judge the quality of fresh produce in the store.

Best-in-class retail design systemsIt can determine the optimal allocation between multiple orders or production lots per day, as well as automatically adjust quantities as needed.

4.1.4. Adding science to the art of fruit and vegetable management

Fruits and vegetables are often last in line when grocery ordering is automated. Obviously, produce faces the same challenges posed by short shelf life and fluctuating demand as other fresh produce categories. In addition, the different supply and quality of fruits and vegetables requires additional flexibility of the planning system used.

The regions where fruits and vegetables come from are constantly changing as crops are harvested in different parts of the world at different times. Even growers in the same region could time their crops slightly differently. Furthermore, since there is always some uncertainty about the availability of good quality products, food retailers usually try to ensure that they always have multiple sellers for the same product.

From the consumer's perspective, a lemon is a lemon, but the supply chain may have to deal with dozens of different product codes for lemons, each associated with a different supplier. Effective fruit and vegetable management requires the scheduling system to be able to switch seamlessly between scheduling levels as needed:

1. Demand forecasting should be done at the product level, for example "domestic organic tomato", using historical sales data for all domestic organic tomatoes regardless of supplier.
2. The quantity of supply should also be determined based on available local organic tomato stocks as well as forecasted demand.
3. However, a replenishment order should be created for the current supplier of domestic organic tomatoes. Here, the planning system needs to move from working at the product level to the SKU level, i.e. from "domestic organic tomatoes" to "domestic organic tomatoes supplied by GreenGrowersCo."
4. Often, a replenishment order must be split between two or three suppliers to ensure availability in the event of supplier shortages, as well as to keep enough suppliers in the business. In this case, the scheduling system should also take into account the allocation of the order requirement to various suppliers—for example, 65% of GreenGrowersCo and 35% of OrganicFarmersCo.

The fruit and vegetable forecasting and replenishment process is too demanding to manage manually, but can be effectively automated. A key requirement is clear instructions on which products will be included in the store's collection and which suppliers will be used for supply at any given time. As in any automation process, high-quality master data is essential.

4.2. Optimized central store replenishment is critical to supply chain efficiency

Fresh produce must be delivered to stores in perfect harmony with demand. On the other hand, warehouse products and other products with a longer shelf life offer more opportunities for optimized inventory flow in the supply chain. Optimized product replenishment in central stores is key to reducing costs in stores and throughout the grocery supply chain.

Retailers who have mastered non-perishable replenishment benefit from a much more even flow of goods through their distribution centers, enabling a much faster return on investment in warehouse automation and reducing the risk of capacity bottlenecks that negatively impact shelf availability. In addition, as grocers spend a lot of time and effort putting products on the shelves, optimized replenishment helps retailers reduce operating costs in their stores.

4.2.1. Replenishment and synchronization of rack space for cost-effective operations

Grocery stores have traditionalit worked in a very mediocre waywith very little communication between the sales teams responsible for store planograms, the supply chain teams responsible for store replenishment, and the store operations teams responsible for store work processes. This has to change.

The space allocated to each product in the store has a significant impact on the results and costs of the store's replenishment process:

• If the allocated space is too large compared to the demand, the inventory required to ensure optimal shelf availability will not be sufficient to maintain a visually appealing, full display. For this purpose, it is necessary to define additional optical minima. Visual minimums indicate how many units of product should be on the shelf to ensure that the display is visually appealing. For slow sellers, visual minimums will always be higher than the inventory levels required for high shelf availability. For products with a long shelf life, this may not be a problem, but for fresh products, excessively minimal visual elements may cause unnecessary spoilage.
• If the allocated space is small compared to the demand, incoming deliveries will not fit on the shelf. At least part of the delivered quantity should be placed in a back room or other storage area. This significantly increases the cost of stacking shelves, as goods must be moved back and forth between the sales floor and the back room. In addition, the use of a back warehouse significantly increases the risk of empty shelves, as timely replenishment from the back area depends on the vigilance of store personnel.

Although surprisingly rare, full integration between space planning and compensation is an important best practice for increased operational efficiency:

• Access to planogram data makes it easy to automate the maintenance of store-level visual minimums for products based on the number of sides or total shelf space allocated to each product in each store.
• Access to planogram data facilitates automatic reduction of replenishment orders that would keep inbound shipments off the shelf. Usually, this rule should be balanced against the risk of shortages if the space available for certain products is too small compared to their demand.
• Access to floor plan information enables allocation of prime replenishment days based on where products appear in the store, with the goal of creating more focused deliveries that minimize the need for store staff to move around the store unnecessarily when shelves are stocked.
• Access to planogram information enables replenishment scheduling so that shelves are fully stocked each time a delivery arrives, minimizing store shelf work. This means that instead of receiving two batches at once, if there is room for delivery and a third the following week, the order is calculated to fill the allocated shelf space on arrival.

The space allocated to each product is crucial to how efficiently the replenishment process can work, so it's important to provide constant feedback to sales. Good analytics tools will help you identify products and stores where there is a mismatch between space and sales, i.e. products and stores for which incoming deliveries do not fit directly on the shelves, or products and stores where minimal visuals lead to waste or branding.

Ideally, spatial planning should always be based on detailed forecasts at store, product and day level, as well as information about replenishment cycles and main replenishment days available from replenishment planning:

• By using accurate forecasts rather than looking at historical sales data when optimizing how to allocate space to products, the space planning team can more easily account for seasonal movements and trends.
• Based on good forecasts of expected peak sales per delivery interval, shelf space can be optimized to be truly efficient for all products in the store on all business days. This type of optimization makes it possible to achieve fewer deliveries and direct-to-shelf flows for a much larger percentage of the product range.

We've seen predictive shelf space optimization translate to up to 30% lower in-store distribution and replenishment costs.

4.2.2. A smart supplement for efficient store operation and a more even flow of goods

Typically, every major grocer replenishes all or at least most of their stores daily from their distribution centers. This is because fresh produce requires frequent deliveries and because total inventory flows are significant enough to justify daily deliveries.

If all replenishment opportunities for all product groups are used indiscriminately, two problems will arise:

1. Store deliveries will consist of a random mix of products from different product categories displayed in different parts of the store. This means that store staff will spend significant time moving protective cages around the store to stock shelves (see Figure 17).

2. Delivery volumes on different days of the week will not be approximately equal, but will reflect daily fluctuations in sales volume, often with significant delivery peaks towards the end of the week in anticipation of weekend demand. This leads to fluctuating capacity needs in distribution centers and stores, which increases costs.

Instead of automatically using all available order or replenishment opportunities for all products, it is a best practice to set early replenishment days for products with a longer shelf life.This means that the replenishment of certain product groups in the main store is concentrated on certain days of the week rather than spread throughout the week. Replenishment planning, such as safety stock optimization and order quantity calculation, will be based on the delivery of goods on the specified main replenishment days. However, to ensure the highest possible availability, replenishment orders are also triggered on other available replenishment days to avoid stock shortages in case of unexpected peak loads.

In practice, this means that instead of ordering detergent every day, fast-acting detergents are replenished mainly on, for example, Mondays and Thursdays, and slow-acting detergents on Thursdays. For detergents, other supply days from the distribution center to the store are only used if there is a risk of the store running out of stock.

Using prime stock days allows for significantly more efficient in-store replenishment without disrupting shelf availability. More centralized deliveries make it more efficient for store staff to replenish store shelves, especially when early replenishment days are set based on product categories displayed in the same aisle or store zone.We have seen a 20% reduction in time spent stacking policies after introducing the main top-up day.

As with many other processes, the use of primary product replenishment days can be further optimized once the basics are in place. By using artificial intelligence to optimize the days of main replenishment across the store, the inflow of goods into the store can be balanced during weekdays. In many stores, weekends can be very busy, with many customers doing their weekly shopping while large quantities of fresh produce are delivered to stores. Setting the main stocking days for products in the main store on quieter weekdays balances the incoming flow of goods and facilitates the scheduling of store staff. (Read thisa case studyfor more details.)

4.2.3. Dynamic pack sizes to meet dynamic demand

A powerful tool for increasing store replenishment efficiency is optimizing the use of different pack sizes. In many cases, stores may choose to order bundles of boxes, pallet layers, or full pallets from a distribution center. It is more efficient to handle larger lots in both stores and distribution centers, but it is clear that deliveries must match available space and store demand. Otherwise, inventory will build up in stores and reduce rather than increase efficiency, congesting backrooms and causing multiple trips between backrooms and the shop floor to fill shelves.

Optimizing package refill sizes by product and store has a direct impact on handling costs, especially for retailers who operate stores of varying sizes. However, doing it just once as a concerted effort is not enough because demand changes over time and for some products with the seasons. During the season, a pallet may be more efficient, while off-peak, smaller boxes may be more efficient.

A retail planning system must be able to automatically optimize the size of packaging to be used by product, store and order.This means that every time an order is placed, the system always checks all available pack sizes—typically ranging from box packs to full pallets—and selects the most efficient pack size based on forecasted demand.

In order to achieve full efficiency, warehouse suppliers must be able to estimate the demand for different package sizes. Otherwise, they could end up using individual packages to assemble pallets for stores, rather than flowing full pallets through the distribution system. This is possible when the store projections (see Section 5) used as a basis for distribution planning reflect the expected use of different pack sizes in the stores.

5. An integrated supply chain driven by customer requirements

Traditionally, store replenishment and inventory management in regional distribution centers or central warehouses were separate processes, driven by separate demand forecasts.

Usurvey 2020,we found that 31% of large US grocers still base their distribution center forecasts on historical outbound shipment data from those distribution centers. It is similar to driving a car while looking in the mirror.

According to the same survey, only 29% of respondents chose a longer-term approach by basing distribution center forecasts on store demand forecasts. Admittedly, this is a better approach than looking only at outbound deliveries.

However, there are some significant drawbacks to using store demand forecasting to drive scheduling in distribution centers:

1. Goods must be delivered to the stores before the stores can sell them.This means that the distribution center forecast must be increased before the store demand forecast is increased and vice versa. The time difference depends on the sales prices and replenishment schedule of the stores, which means the time difference varies between stores and products, and sometimes on weekdays. The result is that it is nearly impossible to accurately estimate the difference in schedule, which is bad news for your forecasting accuracy in distribution centers.
2. When goods are pushed rather than pulled through the supply chain, there will be outbound delivery peaks at distribution centers that are not visible in store demand forecasts.A typical example is promotions, where anything between 30% and 100% of the expected promotional increase must be delivered to stores before the promotion begins. The promotion therefore causes a much higher peak demand in the distribution center than in the stores. This peak is fully controlled by the retailer itself, but still requires a lot of manual scheduling work or "guesswork" when supply planners in the distribution center try to predict when and in what quantities the store will receive the promoted products.

It is rather ironic that many of the situations considered the most difficult to manage in distribution centers, such as stocking stores for promotions or introducing new products, are situations that are entirely in the hands of the retailers themselves.

It is a best practice to base the distribution center forecast on forecasted store orders, which reflect pull-based demand as well as push-based planned inventory movements.In a 2020 survey, 40% of North American grocers implemented this.

To achieve a seamless integration of store planning and distribution, the planning system must be able to calculate forecast store orders by product, store and day several months or even a year into the future, reflecting current and known future replenishment parameters as and demand forecasts. These calculations, of course, require significantdata processing capability, which is probably one of the explanations for the surprisingly low adoption rates.

In practice, store order views consolidate data on their current inventory, safety stock and visual minimums, delivery schedules (including long replenishment days), and all planned inventory movements, including everything from storage to creating advertising displays to shifting orders to requirements for distribution capacity has been equalized.

Table 1 shows some examples of situations where the value of the distribution center forecast base on forecasted store orders is particularly important.

Table 1:Examples of situations where using forecasted store orders, rather than forecasting store demand, allows for much more accurate planning in distribution centers.

When order views are aggregated across stores, they form a highly accurate customer-driven forecast for distribution centers.

Additional benefits of supply chain integration include supply chain transparency that supports capacity planning, supplier collaboration (discussed in Section 7), and easy handling of cross-docking, pick-to-zero, and shortage situations.

5.1. Plan once and execute automatically across the supply chain

When planning in distribution centers is based on forecasted store orders, the effect of planned activities, such as promotions or pre-season allocations, is immediately visible throughout the supply chain. To take full advantage of this transparency, all planning data should be available in the planning system as soon as a promotion plan, variety change, price change or any other related decision is made.

A planning system that supports time-dependent master data is a key factor in proactive planning.Below are just a few examples of how time-sensitive master data allows you to capture valuable information as it becomes available. This in turn allows replenishment planners to rely on the scheduling system to automatically initiate the necessary actions at the right time with very little manual work.

• Replenishment timelines:When store replenishment schedules can be managed using dates, it becomes possible to update planned future replenishment schedules in your scheduling system as information becomes available. This allows replenishment planners to trust the planning system to automatically account for these changes in both replenishment planning and when calculating supply chain projections.
• Range activation and termination dates:When start and end dates are set for an active product line, product increases and decreases are much easier to manage. Routine planning tasks, such as filling the pipeline for new products or reducing inventory for products to be discontinued, can be automated. This automation reduces manual work, but also ensures optimal inventory levels at all stages of the product life cycle.
• Stock before promotion:Of course, promotions have start and end dates, but it is equally important to determine in advance how the stores will be supplied. It is usually ideal to define how many days before the promotion the promotional products should arrive in the stores, what quantities of inventory the store should have in order to be able to create the planned advertising displays, and what percentage of the forecasted advertising demand the first shipment cover. . Rules and standards make it possible to achieve accurate replenishment plans for each store and product without manual work.
• Temporary Supplier Delivery Restrictions:Suppliers may have temporary delivery restrictions. Chinese manufacturers may, for example, not ship during Chinese New Year. If this kind of information is available to the scheduling system, the system knows to issue commands for this period early enough to ensure high availability during the affected period while minimizing manual work and reliance on human memory.

Creating an integrated supply chain eliminates the need for duplicate planning.The effects of planned changes in store replenishment are automatically reflected in the forecasted store orders that form the demand forecast for the distribution centers. This means that once the necessary store stocks for promotions are planned, they will be visible in the distribution center forecast on the right dates and in the right quantities.

Of course, having the right functionality in your planning system is a key factor, but the real challenge is getting the whole organization to work more proactively.Ensuring that decisions are made early enough, but not too early to unnecessarily reduce flexibility in a dynamic market, requires everyone in the organization to have a basic understanding of how the supply chain works and what the relative times are for different types of decisions .

5.2. Multi-level optimization of goods flows

An integrated supply chain enables efficient management of multi-level inventory flows, with minimal waste and a high level of automation.When all data on demand forecasts, available inventory, delivery schedules, lead times and lot sizes for all levels of the supply chain are available in the same planning system, it enables seamless optimization of inventory flows throughout the chain supply.

Cross-docking is an inventory strategy that aims to maximize transportation efficiency while reducing handling costs.Cross-docking is often applied to bulky products, such as beverages, to achieve lower storage and handling costs. It can also be used to shorten delivery times for products with a short shelf life. In a cross-docking facility, goods are delivered from a supplier to a cross-docking facility where the goods are not stored, but transferred from an inbound truck to an outbound truck for distribution to stores.

There are certain requirements for cross-docking to work effectively: 1) Suppliers must be able to deliver full truckloads to cross-docking facilities, 2) delivery units such as pallets or crates must be ready for immediate movement to outbound trucks without additional handling and 3) outbound trucks must have a high occupancy rate to keep transportation costs low. Therefore, the planning system must optimize inbound and outbound flows to and from cross-docking facilities, as well as understand the total delivery time from supplier to store.

Another example of an inventory policy that requires comprehensive supply chain planning is pick-to-zero.In this inventory strategy, orders to suppliers are based on store replenishment needs. However, instead of determining the quantities delivered to each store, the supplier's delivery is redistributed to the stores upon receipt based on the latest inventory information and forecasts. This allows for adjustments to delivered quantities per store in the event that a supplier is unable to deliver in full or in response to possible unexpected spikes in store demand after the initial replenishment need has been calculated. As a result, supply matches demand more accurately than when using a traditional cross-docking approach. A pick-to-zero approach can be seen as a way to reduce the time from order to delivery to stores, as store-specific quantities are finalized not when ordering from suppliers, but when goods are prepared for distribution to the store.

When supply chain planning is fully integrated, exceptions can be handled in an optimal and automated way.Let's look at inventory shortages due to, for example, a late inbound shipment. Instead of fulfilling in-store orders on a first-come, first-served basis, available inventory can be automatically distributed across stores to maximize overall shelf availability or according to regular store priorities. At best, it doesn't even affect shelf availability. Likewise, lots of inventory nearing expiration dates can be proactively moved to stores that have the best chance of selling the product at full price.

6. Effective inventory management in distribution centers

Replenishment of central warehouses and distribution centers is sometimes considered more of an art than a science.It is true that longer delivery times, especially when ordering overseas, and a lack of control over external suppliers add complexity. However, at least in principle, the replenishment of central warehouses or distribution centers is not so different from the replenishment of stores.

When replenishing stores from their own distribution centers, retailers can optimize order fulfillment at their convenience. However, when ordering goods from suppliers, there may be complex restrictions on the minimum order value or quantity. In addition, there may be volume discounts or other discounts that, when used effectively, can have a significant impact on profit margins. Many retailers were unable to put this type of supplier contract or pricing information into their planning systems, requiring business buyers to invest significant time in double-checking orders.

When replenishing stores, the active merchandise flows (combinations of products and stores) for each major retailer are typically in the millions or tens of millions, which means automation is key. For central and regional warehouses, the number of order lines is much smaller and the value per order line is much higher, which makes the economic effect of each order line more pronounced. This allowed and encouraged a lower degree of automation in operational purchases compared to store replenishment.

We have found that setting up operational purchasing processes in a structured way with good system support can also lead to very high levels of automation in distribution centers.This does not necessarily mean, however, that best-practice retailers have significantly weaker buying teams. A key result of increased automation of routine tasks is that business buyers have more time to proactively address potential capacity, delivery or quality issues and analyze the performance of current assortment, vendor and supplier contracts for continuous improvements.

6.1. Total cost optimization of input streams

Since inbound flows of goods to distribution centers are more consolidated than outbound flows, there are more opportunities to optimize orders when replenishing distribution centers than when replenishing stores.

It is important that the planning system can optimize orders at multiple levels in order to achieve the most economical result.

Some examples of order optimization at different levels are:

• Calculation of Economic Order Quantity (EOQ) per product to reduce inventory and handling costs
• Choosing the optimal order lot size—such as case pack, pallet layer, or full pallet—when multiple order lot options are available, taking into account potential price differences between different lot options
• Create mixed pallets for efficient transport and handling of goods
• Manufacturing orders that are fulfilled by one or more freight carriers, such as trucks or containers, or that meet supplier order restrictions, such as minimum order value or minimum number of pallets

Although it seems simple, the process of consolidating orders for multiple products to fill a freight carrier or meeting supplier order limits can be a real test of the flexibility of your scheduling system.

To meet supplier requirements and take advantage of lower freight costs or supplier discounts without stocking excess inventory, you typically need to be able to:

• Flexibly define which products should be combined when planning the order. Products from the same supplier are often grouped together, but sometimes different production sites of the same supplier should be considered separately, or all products originating from the same region, regardless of supplier, should be considered as one group.
• Set goals and/or limits for a bulk order in multiple units, such as value, volume, number of pallets, weight, or combinations of these dimensions. For example, when loading a truck, you want the order to efficiently fill the available space to avoid airfreight costs while ensuring that the maximum weight allowance is not exceeded.
• Let the planning system decide which type of load carrier to try to fill with the order. With some suppliers, it may sometimes make sense to order a truck, sometimes a truck and a trailer, and sometimes two trucks with only one having a trailer, depending on forecasted demand.
• Set the correct order trigger level. When supplier order constraints are difficult to satisfy, it may make sense to require sufficient demand for, say, at least 30% of a truckload before the scheduling system begins to generate an order covering the entire truckload.

Rather than letting the planning system do the heavy lifting when it comes to supplier order requests, it's a best practice to continually evaluate these constraints and their impact on the flow of goods.Multi-year contracts in a dynamic market or fixed order limits for products with seasonal demand may prove expensive or unfeasible as demand changes.

To support this, an ideal scheduling system should flag any proposals to order more/less than needed as a result of these constraints, as well as show the difference from actual need. In addition, it should provide analytical support to help business buyers make rational decisions about whether the benefit, such as a discount, of meeting supplier restrictions outweighs the resulting increase in inventory carrying costs and obsolescence risk.

6.2. Smart shopping takes advantage of good prices

Retail costs are dominated by cost of goods sold.The purchasing team should take responsibility for effectively using discounts to improve gross margin.

In theory, smart buying when prices change is pretty simple:

• When you know the price of a product is going to go up, stock up just before the price goes up.
• When you know the price of a product will drop, order only the quantity you need most before the price changes, then stock up after the new price takes effect.
• When the price will drop temporarily, for example due to a supplier campaign, order less just before the price drop and get stock when the price drops.

To really take advantage of price changes, you also need to factor in the cost of carrying inventory, accurately time orders relative to the timing of price changes, and potentially split the investment purchase—the extra quantity you buy over the one you need to meet demand - in multiple orders.

To further complicate matters, there may be other factors affecting the optimal order quantity. For fresh produce, shelf life is always a factor. It makes absolutely no sense to build up inventory that will end up as waste or damage your reputation by placing products with an unattractive shelf life in stores. Additionally, in situations where storage space is scarce, inventory costs can suddenly jump to a whole new level if you exceed the capacity limits of your current warehouses. When your warehouse space is too full, you would have to find additional space outside of your current warehouses for additional goods, which would quickly turn your investment purchase into a very unprofitable move.

A best practice is to feed your planning system with time-dependent price data so that the system optimizes when and in what quantities to buy when prices change.In this way, you can effectively benefit from even small price changes, as business buyers do not have to spend time manually determining optimal order quantities. It is important to keep in mind that there are limitations to consider, such as shelf life of perishable products or storage capacity limitations. If your planning system cannot automatically handle such restrictions, the purchasing team will need to re-check the proposed investment purchases.

It is not uncommon for supplier contracts to include a discount that is triggered if the customer's annual order value exceeds a specified quota.Again, keeping track of supplier quotas, orders placed and forecasted orders is very difficult manually. Intelligent planning systems support intelligent purchasing decisions by recommending additional orders to obtain discounts when possible and not recommending additional orders that would lead to counterproductive stockpiling.

6.3. Inventory management of perishable goods at lot level

At present, it is impossible to know the exact expiry dates of stocks in stores. It can even be difficult to get a decent estimate if there are multiple batches of produce in the store at once, as some consumers work hard to find the freshest produce available.

However, distribution centers are a different story. In distribution centers, modern warehouse management systems ensure that inventory is shipped on a first-in, first-out basis. In addition, they track the exact expiration date of each lot in stock.

Good use of lot-level expiration data in inventory management reduces waste and improves service levels:

• When the planning system can calculate the predicted spoilage based on the predicted demand and the expiration date of the available inventory, it can order enough goods to replace the soon-to-be-expired inventory before the products run out. This significantly improves the service level of distribution centers in your stores.
• The best classroom planning systemstrigger early warnings when there is a risk of waste. This allows you to effectively push to those stores that are more likely to be able to sell the products at a good price, or gives you time to find other sales channels, such as discount stores, which may be willing to take a lot of the get your hands on a discount.
• Spoilage forecasts also indicate whether there is a systematic risk of scrap, for example due to products being backordered compared to demand.

6.4. Real-time data for buying fresh produce

Perishable products require very high inventory turnover in stores and supply distribution centers. This means that the supply chain is very sensitive to quality issues, delivery problems or unexpected spikes in demand. In situations where commercial demands exceed available inventory, quick responses are critical.

In many cases, suppliers of short-lived perishables make multiple daily deliveries to the same distribution centers. This is partly to guarantee freshness and partly to equalize volumes.

Several daily supplier deliveries allow the retailer to adjust to actual demand by placing orders as close as possible to different order deadlines using the latest demand and inventory data.

However, to be able to detect spikes in demand, the scheduling system must be tightly coupled to the underlying trading systems and have access to real-time data. Of course, the planning system must also be able to process the data fast enough to convert the latest data into orders as accurately as possible.

Such quick reactions and intraday calculations based on real-time data are valuable when fruits and vegetables, which are prone to supply and quality issues, are taken in the morning.Since actual inventory on hand may differ from planned inventory, it makes sense to redistribute inventory based on the latest forecast and inventory data from stores rather than filling store orders with arbitrary order.

7. Planning for optimal capacity and resource utilization throughout the food supply chain

In a dynamic business like retail, capacity bottlenecks can occur at almost any part of the supply chain in response to a variety of events, from holidays or unusual weather to promotions or updates to a wide variety of stores.

To proactively identify and address these bottlenecks, retailers need to understand how forecasted demand will affect inventory, capacity and resource requirements in their supply chains.

7.1. Retail and Executive Activities (S&OE)

The S&OE process aims to ensure that retailers can meet short-term demand for the next 0-3 months as cost-effectively as possible. The starting point is a very detailed demand forecast at the SKU-channel-day level (see Section 3.1). From there, planners can use supply chain views (see Section 5) to gain detailed insight into inventory, capacity, and resource requirements throughout the supply chain.

This end-to-end visibility of retail operations brings many S&OE benefits, including:

• Cross functional coordination:Visibility makes the impact of business decisions made in one function readily available to all functions that need to be considered in their planning. For example, the expected impact of a planned promotion is immediately reflected in all local demand forecasts, as well as inventory and resource forecasts throughout the supply chain. This means that business decisions such asOffers only need to be scheduled once to run automatically.
• Proactive exception handling:With an overview of current and future inventory, capacity and resource requirements across the supply chain, the system can automatically identify potential bottlenecks and help planners prevent or resolve them quickly. For example, if the system alerts planners that upcoming promotional inventory combined with seasonal allocations will create an extremely large delivery peak, they can proactively manage the problem before it starts to fill storage areas or exhaust picking capacity.
• Effective emergency planning:With the digital twin model, retailers can compare and better understand how different planning scenarios will affect their supply chain. For example, if a sales territory exceeds delivery capacity, the distribution planning team can easily model when and how to shift fulfillment for some of those regional requirements to another delivery center, effectively balancing capacity requirements in the distribution network.

When dealing with streams of millions of goods, tens of thousands of employees, hundreds of vendors, frequent promotions, and regular changes in price and assortment, there are bound to be exceptions to every retailer's plans. Many of these exceptions requireimmediate attention and quick adjustmentsto avoid or minimize any negative impact on S&OE.

However, by turning to artificial intelligence and advanced optimization, marketers can actually resolve most of these discrepancies without human intervention. Best-in-class stand-alone problem resolution increases the speed and accuracy with which a merchant can manage exceptions by:

• Autonomous inventory balancingin the supply chain: initiate optimized shortage allocation and culling according to the retailer's business priorities, for example, or re-optimize in-store orders on the fly according to a fulfillment-from-scratch approach.
• Ensuring performance and efficient use of capacity:preventivelysmoothing out fluctuations in commodity flows(see Figure 23), continuously optimizing replenishment and space forreduce in-store merchandise handling, Meoptimizing workforce developmentusing continuously updated workload forecasts.
• Recommending corrective or "better" actions: automatically suggests optimal discounts to remove excess inventory from the supply chain, for example.

Top-performing retailers are able to combine human expertise with technology, adapt quickly to new situations, and implement new business priorities at scale. Goodcase exampleamong them is German pharmacy retailer Rossmann, which took just two days to create brand-new design configurations that prioritized in-store delivery of essential products during 2020's shift in demand caused by the COVID-19 pandemic.

7.2. Retail and Operational Planning (S&OP)

If the goal of retail S&OE is to address unforeseen capacity and resource challenges in the short term, then retail S&OP, by contrast, looks further into the future. The goal of retail S&OP is to ensure sufficient capacity and resources to support future growth goals, planned distribution network changes, high seasons and more.

Preparing for major holidays is an important use case for retail S&OP, with the winter holiday season being the most important (and challenging) in most markets. The holiday season is generally characterized by higher than normal demand that builds up to the peak of the holiday season. After this peak, demand usually returns to normal - or even falls below normal levels for a while.

But the retail S&OP process is approxmore than just supply chain efficiency - it's about maximizing profitability.S&OP should result in:

1. Flows of goods that match and do not exceed capacity throughout the supply chain. This ensures reliable supply, which in turn prevents lost sales due to delivery issues.
2. Profitable business, with minimized costly overtime in all parts of the supply chain.
3. Informed decisions that continue to support company profitability when capacity constraints prevent guaranteed availability for all products.
4. Full transparency of resource requirements, enabling assurance that all resources, including in-store labor availability, are based on the same business plan.

Retail preparations for major holidays such as Christmas usually begin about six months before the start of the season. The first step is to agree on constraints: will there be changes to delivery schedules/delivery times or supplier capacity constraints due to the holiday season?

After agreeing to the constraints, the next step is to review your sales and delivery plans to identify potential bottlenecks that may occur anywhere in the supply chain during the holiday season. Potential bottlenecks may include oversized deliveries that increase store congestion on high-traffic days. days when warehouse staff are faced with more order lines than they can pick. days when the warehouse receives more frozen products than it can store; and so on.

Using spreadsheets, it would simply be impossible to model the supply chain – with all its complexities – accurately enough to proactively identify bottlenecks. Even building a simplified model would be time-consuming and error-prone.The only way to detect the types of mobile capacity bottlenecks described above with any degree of certainty is to use supply chain views.

After identifying potential bottlenecks,Marketers should use what-if scenario planning to examine and eliminate these. Bottom-up scenario planning allows retailers to see exactly how changes in lead times, replenishment schedules, or projected sales volumes will affect the flow of goods.

Typically, food retailers must deliver long-life products to stores earlier to free up capacity to efficiently manage fresh products in their peak season. Different strategies and scenarios for smoothing the flow of goods ahead of the holidays include:

• Filling store shelves:For many products in central stores, full shelves could meet demand for several weeks. Since waste is not a problem, it makes sense for retailers preparing for the Christmas holidays to stock those shelves in early November so fresh produce can be shipped in December.
• Allocation of products with a longer shelf life in stores:Spare parts respect the allocated shelf space in stores, but this is not always enough. Sometimes it makes sense to distribute demand for the next 2-4 weeks across the stores at once. Stores can put these items in the back and replenish them there as needed. Although not suitable for all products, this approach can be useful in special cases.
• Customize delivery schedule:Most food retailers deliver to stores on a regular basis, even outside of the holiday season, but if necessary, it may make sense to schedule additional deliveries to meet increased demand.

Marketers should use the scenario planning capability of their software to determine the scenario that best meets their goals and agree with their suppliers on it. This allows them to lock in their plan well in advance of the season so they can then focus on execution and corrective actions.

An effective S&OP process leads to a higher level of capacity utilization throughout the season.Additionally, planners know and can plan ahead for the remaining peaks, rather than trying to deal with expensive surprises as they arise.

7.3. Effective cooperation with suppliers

Supplier collaboration has been a topic of discussion for decades, but surprisingly few retailers have implemented it extensively.In order to create a fruitful partnership, both parties must put in the effort and reap measurable benefits from the process. Unfortunately, because this rarely happened, many collaborative initiatives failed.

While technology doesn't solve the challenge of partnering with suppliers, it can ease the pain. For example, most collaborative projects spend most of their effort simply collecting data from different sources, buta real programming systemcan minimize this work. Rather than trying to fix everything at once, we recommend building your vendor collaboration processes incrementally.

A good place to start is to share order forecasts with your suppliers, as this is an easy way to collaborate.If your planning system can calculate supply chain forecasts, then a purchase order forecast is already available, telling your supplier what you plan to buy from them in the coming weeks and months.Good systemmay send automated reports that share this information with your suppliers.

You can also share relevant information about planned promotions, upcoming events or other changes, helping your suppliers understand the reasoning behind your purchase order forecast. Retailers can also share demand forecasts or point-of-sale (POS) data with their suppliers, but the most important information is what you expect the supplier to deliver and when.

A more collaborative way of working requires that both parties recognize the value of investing their time and effort.While simply sharing a forecast is a one-way communication, collaborative planning, forecasting and replenishment (CPFR) is a true two-way communication.A good scheduling systemIt helps by providing reliable forecasts of future purchase orders, analytical tools to understand potential changes and issues, and a platform or portal for collaboration.

Ideally, suppliers can simply access a retail view of forecasted demand, order placement plans, and data on promotions, seasons, events, etc., and then add their own view. Combining the supplier's holistic view of their categories and products with the retailer's understanding of their business and marketing activities in this way ultimately leads to a more accurate overall plan.

Best-in-class scheduling systems canto support this type of collaborationproviding a platform that can accept multiple types of forecasts, alert users to any discrepancies, allow users to edit plans and finally analyze the agreed plan to whatever level of detail they need - be it trades, products or days - to support operational execution.

8. Conclusion: Work with the machines to win

Retail is in turmoil and it is unclear what the impact of different sales and delivery channels, store formats or even retail players will be. In 10-15 years, we will probably look back on this era in amazement and ask "How did we not see this?"

However, it is easy to make some predictions about the future of food retail:

1. Food retailers will fly less. It's a shame to waste so many resources on growing, transporting and handling food only to have it end up in a rubbish bin behind a supermarket. Grocers must and will take responsibility for significantly reducing food waste, and since waste leads to profits, their efforts will also be great for their business.
2. The retail food supply chain will become more efficient. Consumers are very price conscious and will not accept premium prices to keep inefficient supply chains in business. No one benefits from huge fluctuations in workload or capacity requirements caused by poor planning and management, so neither retail employees nor management should be sad to see old, inefficient practices go.
3. Technology and automation will play a big role in the transformation of retail. We have already seen this in other sectors that once relied heavily on routine manual work. There's no reason why retail shouldn't follow.

In short, retail supply chains need to be more responsive and better controlled than ever to meet the demand for fresh produce with minimal waste. At the same time, retail supply chains must become more efficient by optimizing inventory flows from multiple angles—store operations, distribution, picking, and warehousing—to respond to price pressures. This is only possible by working with intelligent machines.

The world of food retail is too complex to be managed by notebooks and intuition. This, of course, has been true for a long time. The breaking news is that not only are the simplest tasks being automated, but the significantly more advanced roles of designers are also being filled by machines. More importantly, intelligent automation will not only replace manual work, but will take programming to a level of sensitivity never seen before.

Will humans then have a role in this brave new world? Yes, there will be many. The three important roles are:

1. Master of Machines: We are making great progress in specialized artificial intelligence, a type of machine intelligence useful for solving very specific tasks. However, we still need to have talented people designing systems and determining when and how best to use the available machine intelligence.
2. A fellow machine: Machine learning algorithms are highly dependent on access to data. They struggle to apply common sense or find innovative solutions to new data-poor scenarios. This is where their human counterparts can provide invaluable insight.
3. Innovators, who think outside the box: Especially in companies experiencing creative destruction, there is a great need for new thinking, new business models and new ways of providing food and services to consumers. Innovation in retail is still far beyond the capabilities of artificial intelligence.

So don't hold your breath waiting for AI to revitalize your retail business or even solve your supply chain challenges. But start gradually using machine intelligence where it is most feasible and effective. This collection of best practices is a good place to start.

Food Supply Chain FAQs

1. What are the biggest challenges in the food supply chain?

Some of the biggest challenges in the grocery supply chain include managing the rapid growth of online ordering and delivery options, competing with discount stores and the food industry, handling fresh produce with a short shelf life, meeting growing consumer demand for sustainability and adaptation to changing consumer preferences and market conditions.

2. What strategies help companies optimize their grocery supply chain?

Strategies for optimizing the grocery supply chain include:

• Using artificial intelligence (AI) to forecast demand and manage inventory
• Improving the replenishment process for fresh and in-store products
• Integrating supply chain planning and execution
• Improving cooperation with suppliers
• Use advanced scheduling software to manage capacity and resources

Companies should focus on the most urgent areas of improvement first to address the most significant inefficiencies and achieve the fastest ROI from their optimization efforts.

3. What features should I look for in grocery supply chain planning software?

Companies need a robust scheduling solution to manage the many supply chain challenges they face. Some of the most important features of any capable platform include:

• Detailed demand forecasting and machine learning capabilities
• Integration of store and distribution center operations
• Tools for effective collaboration between stores and central teams
• Support for multi-level inventory management and automated replenishment
• Scenario planning and capacity optimization capabilities
• Effective collaboration tools with suppliers

Additionally, the software should provide transparency and allow for human input and expertise to complement automated processes.