Amazon Supply Chain and Logistics

Dynamic pricing for logistics service providers to maximize profitability

Revenue management is an important business capability in the transportation and logistics industry. Key components of revenue management involve contract pricing, spot or dynamic pricing, and yield management to maximize profitability and optimize transport allocation. This post focuses on dynamic pricing of freight and shipments for spot buying.

The global freight market that combines trucking, shipping, and freight forwarding was worth approximately $3.3 trillion in 2021 according to combined data points from Armstrong & Associates, Transport Intelligence, and Drewry. Depending on the mode of transport and seasonality, the spot market represents up to 50 percent of the overall freight market.

Pricing capability is particularly important for logistics service providers (LSPs) as they both buy and sell transportation services. Dynamic pricing has a direct impact on profitability for LSPs because it helps them manage the spread between what shippers are willing to pay to move a shipment and the price at which they can buy the capacity.

Average gross margin (AGM), defined as the difference between sell and buy sides, has historically fluctuated between 15 to 35 percent, depending on factors such as the mode of transport and how competitive the shipment lane is. Given the large range in realized AGM, LSPs have an opportunity to maximize their profitability by improving their ability to respond to changing market conditions. By adopting dynamic pricing solutions, LSPs have the potential to increase AGM by up to 10 percent, especially if they are at the lower end of the range.

Cloud technologies are especially useful for dynamic pricing, e.g., for enriching historical internal data with external data, using machine learning (ML) models to drive granular insights, and creating market views to improve confidence in decisions driving revenue growth and profitability. In this post, we will demonstrate how AWS cloud solutions enable dynamic pricing and can help improve the LSP bottom line.

Why LSPs are losing opportunities on spot market

Spot market transactions rely on processes that are largely manual and still require multiple steps of offline interactions between shippers, LSPs, and carriers. A shipper requests a quote for one or multiple shipments, providing attributes such as equipment type, lanes that represent origin-destination pairs, and requested pickup and delivery dates.

LSPs evaluate the request and respond with an offered sell price. If the shipper accepts the sell price, then the shipment is awarded to the selected LSP for operational execution. The LSP now has to buy the capacity from carriers so that they can execute the shipment. There are various mechanisms that LSPs can use to buy capacity. For example, they post a buy price on internal or external load boards and have carriers respond, or they directly send a proposed buy price request to a set of core carriers in the LSP’s business network.

The diagram shows three components of the logistics market, i.e. Logistics Service Providers interacting with Shippers and Carriers.

Currently, LSPs lack automation to quickly and dynamically decide on sell and buy prices. Additionally, market data about attributes such as load-to-truck ratios – that is, the number of loads requested versus actual availability representing the balance of demand and supply – are usually limited or not available in a timely manner, leaving LSPs to rely on internal and historical data to make pricing decisions.

Minimal insights into market pricing by attributes such as length of haul (LOH), equipment type, and seasonality limit the ability of LSPs to drive profitable growth through optimal pricing based on the actual capacity versus demand. Limited data from carriers and LSPs constrains personalized strategies for carrier management.

Finally, given the fragmented nature of the transportation and logistics industry, even the largest LSPs have visibility only into a small percentage of the market, defined as total demand and capacity for lanes. This opaqueness leads to uncertainty on pricing both on buy and sell sides; this hinders an LSP from maximizing the opportunities for increased profits and business growth.

AWS solution for dynamic pricing

For LSPs to be on the higher end of the AGM scale, they need granular and timely insights based on attributes such as mode of transport, market conditions, and the competitiveness of the environment. The ability to optimize AGM comes from analytical capabilities that can model critical factors such as complexity of the request (e.g., dozens of lanes), timeliness (e.g., real-time quotes), service level (e.g., transit time or buy price driven by carrier selection), and seasonality (e.g., higher rates during peak season).

We recommend three key areas of focus for modernization of pricing capabilities: (1) enhancing analytics to determine optimal pricing; (2) creating behavioral models to personalize transactions; (3) creating a market view to better understand demand and capacity and adapt pricing dynamically.

Enhancing analytics for the buy and sell sides of the transaction

Pricing reflects strategy, and LSPs adopt different approaches that depend on market conditions, business model, internal processes, and human and technological capabilities. Irrespective of the approach, internal and external data and probabilistic models for pricing enhance both sides of the transaction, i.e., the sell and the buy sides.

For example, with one of our customers, the AWS team focused on the sell side to forecast demand and understand price elasticity to maximize profitability on a transaction basis. The team used internal parameters, such as the number of orders, win rates, and total brokers, in addition to external parameters, such as Statista container freight rate, in the analysis. The team built an ML model to improve forecasting accuracy by 13 percent. The team then determined the sell price using improved demand accuracy and an understanding of price elasticity.

On the buy side, another AWS customer implemented a carrier management strategy based on a set of recommended carriers by shipment profile and visibility of the respective capacity. The team used AWS SageMaker to create a probabilistic model for the sell price and articulate a range of prices in which the transaction could be completed. The team also ingested data from external sources to understand market conditions such as load-to-truck ratios and support the pricing decision. The following figure shows a dashboard implemented in Amazon Quicksight to support the buy side of the transaction.

Amazon Quicksight dashboard shows: (1) a list of loads in the pipeline with customer quotes; (2) Machine Learning deep diver into one of the loads with recommended carriers, load to truck ration, load prices and contract recommendations; (3) lane origin and destination on a map.

Consider the case of the highlighted load where the sell price is $1,053 from the load selection in section 1. The load drill down in section 2 shows the load buy price ranges from $566 to $741, and the load-to-truck ratio is 4.5, which implies that there is more demand than capacity in the market. Brokers use this information, as well as the information on recommended carriers and contracts, to determine the gross margin at which they could successfully complete the transaction. Leveraging such data and analysis over its book of business, the LSP, which had served around 500 lanes, increased overall AGM by 4 percent by improving pricing on low-density lanes as well as servicing new lanes. Section 3 helps brokers quickly get their bearings for the same filtered load in terms of lane origin and destination.

Behavioral models

Given the fragmented nature of the logistics market, it is critical to understand the various stakeholders involved in the business transaction as well as the context of the transaction. Entities such as carriers and drivers could have different needs at different times. For example, a driver might prioritize getting home for an important event over the price of the load. Entity behavioral models enable LSPs to better understand the needs of each stakeholder so that the sell and buy approaches are tailored to serve the shippers and carriers while maximizing profitability for the LSP or broker.

This idea is similar to how the retail industry creates behavioral models of consumers and develops relevant market development strategies. The information is also leveraged to create opportunities for LSPs – proactively soliciting shipments from shippers as they have increased visibility to where and when time capacity would be available.

Entity-based behavioral models that allow LSPs to optimize transactions for shippers, carriers, and LSPs represent a significant development in the space. This concept is still evolving, with most players still focused on collecting relevant data attributes. For example, one of our LSP customers wanted to increase the number of core carriers (defined by the weekly loads completed) by lane and consequently increase volume and improve SLA performance. The customer collected unstructured data on demand and capacity signals from emails and contact center calls to build out profiles for participants by lane. The customer used Amazon Connect and Amazon Comprehend to modernize and automate the process together with Amazon Personalize to create segments of shippers and carriers using data such as their key characteristics, transactional data, and behavioral data. Based on the carrier segments created, the goal was to increase the percentage of “core carriers” to 15 percent on high-density lanes and increase transactions profitably completed by 10 percent.

Market view

Shippers, LSPs, and carriers have a limited view of the market, even on the lanes that they serve. Stakeholders are looking for a more granular view of the market, including updated data on demand and capacity, so that more intelligent pricing decisions can be made.

An AWS customer, an LSP, created a market view for the reefer market that provided a forecast of demand and capacity for the following 48 hours and categorization of the market as hot or cold depending on market conditions. Market view allowed brokers to have a higher confidence in the state of the market, and they used the information to guide commercial and operational discussions with shippers and carriers. Leveraging a forward-looking forecast, brokers could also anticipate ahead of the competition when market trends were about to shift and adjust their pricing strategy accordingly. Using the market view, the LSP expanded the lanes served by 10 percent.

The ability to aggregate internal and external data to create a market view is a key ask from LSPs. However, LSPs need to take a measured approach based on aspects such as current lanes served, increasing visibility on low-density lanes, and extending into adjacent lanes. By ingesting data from various sources, shippers, LSPs, and carriers are able to get a broader perspective of market conditions and greater context to drive better procurement buy and pricing sell decisions. Today’s availability of external data sources such as DAT and SONAR (trucking), Xeneta (ocean and air freight), AAR (railroad), and AIS (ports) is providing information about market conditions and extending the market view to a level that was not even imagined until a few years ago to support operational decisions.

AWS Data Exchange also continues to add relevant data sources to augment customer visibility on external factors, such as macroeconomic indicators, weather, and traffic conditions. AWS technology supports easy integration of third-party data into data mesh. Amazon S3 and purpose-built databases such as Amazon Dynamo DB are used for data ingestion and transformation, and Amazon SageMaker ML capabilities expedite the creation of models for pricing and entity behaviors.

The ultimate LSP vision would be to implement all three of these components so that AGM is closer to the higher end of the 10 to 35 percent range. It is, however, possible to implement these components in a phased manner and improve AGM incrementally.

An example of a phased approach for dynamic pricing

The illustration shows phased approach for dynamic pricing implementation: assessment phase, load matching as phase 1, behavioural models as phase 2, and MarketWatch as phase 3.

We recommend a phased approach to modernizing pricing capabilities. AWS has worked with customers to define phases that would be achievable for each particular business case based on the mode and status of customer readiness. An example implementation approach encompasses three phases: (1) create a “load” matching minimum viable product (MVP); (2) enhance the buy and sell side of the transaction by building behavioral models; (3) extend to a broader set of lanes to create a market view.

AWS typically partners with customers in a two-pizza team model to deliver dynamic pricing capability. The two-pizza team needs expertise in the logistics industry, supporting data engineering (data ingestion and data transformation) and data science (ML engineers), and assuming that foundational aspects of security and development such as continuous integration and continuous deployment have already been addressed.

In these engagements, AWS partners with customers for the following:

  • Discovering customer’s goals, desired outcome, and time to value
  • Understanding current relevant business models, challenges, and desired end state
  • Establishing metrics of success, such as gross margin, win rates, and productivity
  • Defining architecture and roadmap
  • Creating MVP


In this post, we discussed the importance of the spot market for the transportation and logistics industry and why it is so important to drive profitability of LSPs. Currently, spot market transactions are quite manual and constrained by data limitations. LSPs acknowledge the potential for incremental margin and are looking to modernize pricing capabilities by using cloud-based technologies. LSPs continue to seek real-time data about market conditions, develop deeper insights into attributes that drive pricing, and personalize transactions to improve key financial metrics such as AGM. Availability of industry-specific data sources, technologies to ingest and transform the data, and ML analytics can help increase AGM of LSPs by up to 10 percent, depending on performance baselines, modes of transport, and business models.

In this post, we discussed various strategies LSPs are experimenting with to modernize pricing capabilities and improve financial performance. We recommend a phased approach for modernization, and the AWS Supply Chain, Transportation, and Logistics team partners with LSPs to develop a tailored solution to deliver business objectives. If you wish to explore how AWS could support your modernization journey, please reach out to your account manager to set up a discovery workshop with our business unit.

Dnyanesh Patkar

Dnyanesh Patkar

Dnyanesh Patkar is the Head of Transportation and Logistics in AWS Industries. In his role, he works with customers to envision, develop, and execute transformational business and operating model strategies in this space. He is a seasoned executive, and his background includes leading business transformation programs, serving in general management roles with P&L responsibility, and creating high-performance teams. He has over 25 years of experience at companies such as Schneider National, DiamondCluster, and National Semiconductor. Dnyanesh has his MBA from the Wharton School of Business and a Master of Engineering from Cornell University. Dnyanesh is based out of the AWS Atlanta office.

Luca Romanski

Luca Romanski

Luca Romanski is Principal Transportation and Logistics Americas at Amazon Web Services. He joined AWS in 2022 and works with customers to help transform their businesses through digital innovation and end-to-end solutions that leverage the power of AWS Cloud. Prior to joining AWS, Luca held executive positions at Kuehne+Nagel, Ceva Logistics, and Ariston Group, spending 15+ years in supply chain operations both in Europe and in the US. Luca holds an MS in Mechanical Engineering and an MBA in Business Administration. Luca is based out of the AWS New York City office.