Skip to Main Content

Generative AI: Using artificial intelligence to make human impact.Learn how

Shipping containers in the shape of a global map


Supply Chain Transformation

How retailers can improve e-commerce profitability and strengthen the customer experience through supply chain reinvention and optimization

Sudip Mazumder
Sudip Mazumder
Guy Elliott
Guy Elliott

Today’s digital shopper expects more from retailers: fulfillment solutions like click & collect; same-day delivery; free shipping; hassle-free returns. While it has been difficult for retailers to meet these expectations, the growing popularity of e-commerce has only compounded matters as many organizations lack the maturity to support the volume of customer activity and the variety of their fulfillment needs—all while maintaining profitability.

Unfortunately, the implications reach far beyond cost. Without proper planning and resources, the shift to digital also runs the risk of degrading the customer experience. For some retailers, a poorly executed digital strategy has triggered an erosion of a brand that took decades or longer to build.

As retailers operate in this increasingly digital world, many have a need to transform the supply chain to both increase e-commerce profitability and also enable a strong, consistent customer experience. In this paper, we discuss how and where organizations can focus their efforts across the value chain in order to eliminate waste and redundancy, build affinity and loyalty and drive growth.


Retailers’ digital maturity

Inventory Visibility

Only 17 percent of retailers have full inventory visibility – including in stores, in transit and in returns.¹


Cost of fulfillment accounts for 9-12 percent of a retailer’s revenue² and fulfillment costs have increased 12 percent over the last 12 months.³

Warehouse Automation

Only 15 percent of retailers use automated packing, while 19 percent use automated picking.⁴


Products ordered online are returned up to 3X as often as in brick & mortar stores.⁵

Focus on HOW
Supply chain reinvention and optimization

While nearly all retailers understand the need for supply chain transformation, there is less consensus as to where to focus efforts in order to have the most significant impact. Which levers are the most critical? How can they be identified? Once they are, what steps should be taken that have a cumulative positive impact across the value chain and customer experience that builds in flexibility and agility?

While every situation is unique, there are three key areas that nearly all retailers can benefit from as it relates to supply chain optimization:

1.	Demand Planning and Inventory Optimization: How to enable growth and optimize costs through more precise demand and inventory predictions  2.	Fulfilment Optimization: How to drive faster and more expansive fulfillment  3.	Returns Optimization: How to rethink models to deliver a circular economy

Figure 01

Publicis Sapient Supply Chain Optimization Framework

1. Demand Planning and Inventory Optimization: How to enable growth and optimize costs through more precise demand and inventory predictions

Demand planning—the ability to accurately plan and predict product needs across both digital and physical channels—is one of the most crucial components in a modern supply chain. Effective demand planning allows organizations to optimize service and inventory levels, thus reducing the need for excess inventory and minimizing costly markdowns. In addition, demand planning is instrumental in optimizing costs through tangential operations like workforce planning, transportation, warehouse allocation and vendor management.

Demand planning consists of three main areas:

  • Demand shaping: Creating demand through interventions such as promotions, events, marketing and other customer-facing tactics
  • Demand forecasting: Predicting future demand
  • Demand sensing: Identifying and reacting to short-term (hourly or daily) fluctuations
$1T: Total of annual lost sales worldwide due to out-of-stocks   $472B: Total of lost revenues due to overstock markdowns
Inventory Optimization: - Improve replenishment process and frequency  - Reduce inventory carrying cost by decreasing number of days of supply  - Reduce safety stock - Improvement in order fulfillment rates - Improved inventory turns.  Operations - Improve accuracy of budgets and financial planning - Implement purchase order automations to avoid stock issues - Better anticipate staffing needs - Measure progress towards business and sales objectives - Streamline production process - Plan advertising and marketing campaigns and budgets - Enhance the customer experience (avoid out-of-stocks, backorders, late shipments, etc.) Resourcing and project management. Agility and Efficiency - Automation of data workflows could greatly improve speed to insight for forecasting outcomes - Improved ability to respond quickly to large demand shifts

Figure 02

Implications and Benefits for Demand Planning and Inventory Optimization

In the modern retail landscape, demand planning is most accurate when a business considers both internal and external data to inform predictive models. Internal metrics may include historical sales numbers, ad spend, digital traffic and in-store footfall. The business may also consider external factors, such as consumer trends, weather, and holiday spending. The insights generated from these models can be used to develop sales predictions which can then be tracked in real time. While accurate demand planning takes some of the guesswork out of inventory, this capability also helps organizations improve agility and flexibility.

For example, an accurate demand planning tool will take seasonal weather and local climate into account when selecting inventory and determining product levels. However, if an area experiences unexpected extreme weather, such as heavy snow and ice, the system will detect upticks in related purchases, such as shovels and rock salt and restock inventory before it runs out. On the other hand, this system will identify sluggish sales on other items and enable the business to consider promotions, offers or other advertising efforts to help generate interest before drastic markdowns need to be taken.

One leading grocer had an 80 percent success rate in its current demand forecast, which meant that items were not selling as predicted 20 percent of the time. This resulted in unnecessary overstocks and costly out-of-stocks, squeezing the bottom line from both ends. An AI-based demand forecasting system with machine learning was able to produce a more accurate picture of inventory needs, reducing the error rate to just over 5 percent. This represented a 75 percent improvement over the current system and resulted in significant savings for the organization.

While accurate demand planning takes some of the guesswork out of inventory, this capability also helps organizations improve agility and flexibility

Figure 03

Global inventory visibility has three key pillars

When it comes to demand forecasting, specifically, there are two main ideas for how to improve accuracy. The first is to use data-enabled tools and technologies to ananlyze data on a granular level and formulate insights. This method helps improve the accuracy of short-term forecasting by assessing a wide range of variables, such as:

  • Location intelligence, including purchasing power, crime rates and traffic patterns;
  • Special events, such as Singles Day or Prime Day;
  • Dynamic pricing, including market place pricing and price match;
  • Personalization of ads, promotions and recommendations;
  • Social trends across major social media platforms;
  • e-commerce shopping trends, including search terms, basket trends, item abandonment and reviews; and
  • Commercial insights, including store manager feedback and category manager insights

Using this method, one of our clients was able to reduce the forecasting processing time from seven days to three hours through the use of cloud-based tools. In another engagement, we were able to reduce the time to upload forecasting data from 550 million records in .csv format from 10 minutes to 2.7 seconds through the use of big data techniques.

The second way to improve demand forecasting accuracy is through the use of fast data analytical tools to sense near-term (hourly, daily) demand. This approach may incorporate metrics such as:

  • Intraday activity data, including footfall, point of sale numbers and inventory levels;
  • Intraday competitor data, like dynamic pricing and promotions; and
  • Intraday external events, such as temperature changes and traffic congestion.

Given the growing importance of omnichannel fulfillment, it is critical for retailers to modernize their order management and inventory solutions. One way to do so is by leveraging microservices to provide a real-time view of inventory across the retailer’s supply chain. This helps optimize inventory while also reducing the need for markdowns, increasing customer conversion, and building retention.

For example, with global inventory visibility, if a customer walks into a store and finds an item out of stock, the manager can instantly refer to the inventory view and provide options for fulfillment based on in-transit goods, additional nearby stores, or online orders. Looking to digital channels, a near real-time inventory solution enables retailers to provide choices at the time of purchase and nudge customers towards more profitable “win-win” fulfillment options.

Provide choices at the time of purchase and nudge customers towards more profitable "win-win" fulfillment options.

Publicis Sapient global inventory visibility solution framework

Alongside global inventory visibility, multi-echelon inventory optimization enables retailers to optimize inventory levels against attributes like cost, shelf-life, profit margins and space. It enables companies to automatically calculate the right inventory levels at all tiers, including national, regional and local levels, thus enabling companies to have the optimal working capital investment.

In conclusion, demand planning and global inventory management are two of the most significant drivers of cost optimization within the supply chain. While there is a significant variety in the way retailers can plan—including demand forecasting, shaping and sensing—nearly all rely on the use of big data, predictive modeling and advanced technologies, such as AI and ML. So too is the foundation for a modern global inventory management system. As such, organizations must invest in these technologies and build their internal capabilities to ensure they have the resources to optimize demand planning efforts and tap into this powerful growth opportunity.

Demand planning and global inventory management are two of the most significant drivers of cost optimization within the supply chain.
2. Fulfilment Optimization: How to drive faster and more expansive fulfillment

Fulfillment – the process of receiving, packaging and shipping orders – has long been the cornerstone in every retail organization. However, as consumers take their purchases online, many organizations are struggling to adapt and expand traditional supply chains in order to serve customers when and where they are. This not only requires the rethinking of the fulfillment design but also finding ways in which fulfillment could be made more profitable for the retailer without compromising the customer experience.

In today’s digital age, retailers need to design their fulfillment strategy and channel mix to both optimize costs and increase speed, choice and convenience for customers. In many cases, this will include re-engineering traditional physical channels to enable true omni-channel fulfillment, including leveraging the existing store network to reduce the cost and increase speed of last mile fulfillment of digital orders. Retailers must also re-think how to increase and scale in-store pick up, which not only takes last mile costs out of the equation but also increase store footfalls and the ability to cross sell/up sell.

Omnichannel fulfillment strategies
Person with an iPad Pro

Enabling modern day fulfillment models requires more than just technology.

Figure 04

Publicis Sapient Fulfillment Optimization Business Framework

Data/AI plays a key role in fulfillment optimization throughout the customer journey. During the pre-order process, insights about customers, inventory availability and rate shopping should be leveraged in real time to provide the right fulfillment options to drive higher conversion in the buying process while minimizing cost. In the post order portion of the journey, data optimization models can be used to manage multiple complex parameters like customer proximity/last mile costs, store performance, split shipments, existing and predicted markdowns and customer service level agreements (SLAs). 

As shopping continues to shift to digital channels, retailers face significant pressure to optimize the fulfillment model, not just as it relates to individual sales and cost to serve, but the channels themselves. To do so, retailers must eliminate silos, literally and figuratively, to help redesign the organization in a way that supports a true omnichannel experience.

Figure 05

Publicis Sapient Fulfillent Optimization Solution Framework

3.Returns Optimization: How to rethink models to deliver a circular economy

Returns are the new normal. In the U.S, e-commerce returns have grown exponentially in the last five years, and e-commerce generally sees three times as many returns as the brick & mortar channel⁶. Based on the present growth rate for returns, by 2027, customers will return nearly $1 trillion in merchandise annually.⁷

From the customer’s point of view, returns should be seamless. However, for retailers, they are anything but. Returns have proven to be both timely and expensive, ultimately eroding margins. For fashion retailers, in particular, this is a challenge as items may not make their way back into inventory for weeks after they are initially shipped to the customer. Oftentimes, the items may also need to be reconditioned before they are resold. In total, this process can take up to 30 days, which is an incredibly lengthy period of time for an industry that capitalizes on fashion trends and seasonal needs.

The High Cost of Returns: 1 in 2 Returned items cannot be resold at their full retail price⁸ - 95% Increase in U.S. ecommerce returns in the last five years - 2X ecommerce return rate vs. stores - $1T Anticipated value of returned merchandise by 2027
By 2027, customers will return nearly $1 trillion in merchandise annually.

In order to reduce both the cost of returns to the business and the length or the return cycle, retailers must build a competitively advantaged return capability that optimizes products, the purchase experience and the cost of the return—all while maximizing customer value.

Figure 06

Differentiated customer service based on customer value

While it’s important to focus on improving returns process efficiencies and putting products through the re-commerce cycle more quickly, the use of data/AI to predict return and fundamentally address reasons for returns is equally important.

Figure 07

Publicis Sapient Product Return Rate Reduction Framework

Using AI and ML, retailers can identify which items are experiencing a high rate of returns and determine reasons why. For example, size/fit is amongst the largest contributor to apparel product returns. Investments in developing the right product recommendations for size/fit at the time of purchase could help reduce returns. Retailers can also cross reference returned and non-returned items, thus having a better sense of size/fit both as it relates to the product catalog and the personal profile of each customer. Retailers can use this information to make adjustments, such as improving the product description, adding a sizing guide or upgrading the imagery.

Data/AI-driven insights about their customers can also enable retailers to address some of the most challenging and complex drains on the returns process, such as compulsive and impulsive shoppers and net-loss customers.  By using technology, retailers can develop models that predict real-time return probability during customer checkout. The retailer can then decide whether to intervene based on transaction characteristics, the customer profile and customer lifetime value (CLV). This technology can also help eliminate fraud, as it can detect with great accuracy customers who abuse or circumvent the return policies.

Category	Lever	Reduce Costs	Drive Sales	Impact Policy	•	Increase the number of items that are non-returnable •	Cut off high volume returners that are unprofitable	High   High	Low   Low	Quick Win   Quick Win Loyalty	•	Reward high-value customers with favorable return policies  •	Reward high-value customers with or no shipping minimums •	Reward customers for keeping items that are frequently returned	Medium   Low   Medium 	High   Medium   Low	Foundational   Foundational   Quick Win Size/Fit	•	Improve size consistency based on returns size analysis	High	Medium	Foundational Digital Experience	•	Cross sell/up-sell “no return” items to reach free shipping limits •	Improve online product descriptions •	Invest in customer reviews, particularly around size/fit •	Build customer profiles to enable personalization	Medium   High  High  High 	Medium   Medium  High  High	Quick Win   Quick Win  Foundational  Foundational In-Store Experience	•	Rethink fitting rooms (staff for size/fit service and/or offer preferred service for high value/loyal customers)	Medium	Medium	Foundational

Figure 08

Publicis Sapient Return Modernization Impact Model

Unfortunately for retailers, there is no single “magic bullet” for returns optimization. Instead, success will be a cumulative effort as organizations refine internal processes to enable greater levels of efficiency and influence customer behavior through a range of incentives, perks and policies. To build momentum within the organization and demonstrate results, retailers should consider beginning return optimization efforts with one of the “quick wins”—those areas that offer high impact with relatively low complexity.  Over time, when put together, the efficiency gains will compound, enabling greater speed and cost reduction, while also serving as a valuable example for the business.


For many retailers, the shift to digital channels is happening far faster than they expected. In less than a decade, e-commerce sales are expected to quadruple, accounting for 40 percent of total revenue. This means that retailers must find ways to make this channel more profitable now, in order to remain viable in the future.

The biggest takeaway is advent of e-commerce has fundamentally changed the retail supply chain. But, at the same time, few retailers have made substantial changes to the business to support these new needs.  Further, the supply chain is no longer just a cost play, it also focuses on the customer experience. To reinvent the supply chain, businesses will need to be forward-looking, relying on new technology and operating models as opposed to past experience and legacy solutions.  

In today’s digital world, supply chains must move beyond a cost center or back office function. Instead, they should be viewed as a growth engine that can enhance the customer experience through improved choice, speed, convenience and quality.  


  1. D3 - The 2018 Retail Supply Chain Report
  2. Publicis Sapient research
  3. D3 - The 2018 Retail Supply Chain Report
  4. D3 - The 2018 Retail Supply Chain Report
  5. Invesp
  6. Invesp
  7. Supply Chain Management Review
  8. Supply Chain Management Review


Sudip Mazumder
Sudip Mazumder
Industry Lead, Retail North America
Guy Elliott
Guy Elliott
Executive Vice President

Related Articles

  • 8 Trends Accelerating the Future of E-commerce in 2024

    Which trends really matter? Our industry experts highlight the shifts that every retailer needs to know.


  • Five Opportunities and Challenges for Retail Healthcare in 2024

    As retail leadership teams craft their health and wellness strategies, they need to find ways to bridge physical and digital customer experiences.


  • Beyond Product Recommendations: How Retailers Can Increase Profits with Generative AI

    Our experts summarize the state of generative artificial intelligence in retail in 2024, including use cases across customer experience, supply chain and back-end e-commerce.