Like all industries, the transportation and logistics industry is in the midst of extraordinary transformation. Digital technology is impacting both the industry and companies that operate in this industry in unprecedented ways. There are many factors driving this transformation, including global macro-trends such as e-commerce. Companies are being forced to rethink the ways in which they compete, acquire and service customers, and generate profit. Existing industry incumbents that are adopting technology in new and different ways are starting to find new competitive advantages. At the same time, new competitors are emerging who threaten to find innovative ways to engage customers, reduce friction, and create an existential threat to incumbents. The industry landscape is expected to dramatically change over the course of the next decade, and there will be clear winners and losers.

Executive Summary

One of the key technologies that is expected to drive this transformation of the industry and companies is artificial intelligence (AI). Within AI, there are three specific technologies that will drive most of this impact: machine learning, natural language processing, and robotics. For a trucking company to continue to create both differentiation and cost advantage, they will need to fully embrace these AI technologies over the next two to three years in their strategic visions and corresponding operating plans. Without doing this, trucking companies will certainly start to see market share and profitability erosion as both existing and new competitors start to use these technologies to gain competitive advantage.

The three technologies of machine learning, natural language processing, and robotics can be applied to specific processes within a trucking company to create differentiation. Prime process candidates include using machine learning to predict customer churn and equipment maintenance schedules, customer service functions like email processing using natural language processing, and automation in warehouses or terminals using robotics.


Current State

Many asset-heavy trucking companies that operate with a legacy of physical assets such as trucks, trailers, warehouses, and terminals are faced with the proverbial problem of being caught between a rock and a hard place. On the one hand, they have to continuously invest in upgrading their equipment, such as trailers, trucks, and terminals, to ensure they can transport goods efficiently and without damage. On the other hand, they must also invest in technology to find, retain, and service customers better, remove operational costs, and produce more profit for your shareholders. These companies also face other issues including a serious driver shortage, the threat from intermediaries such as 3PLs and digital brokers who are increasingly taking away revenue and profit share.

The best competitive strategy for dealing with these issues varies depending on the customer or market segment being serviced. In most cases, companies want to operate at the lowest operating cost while creating differentiation with a large national coverage network. In some market segments, such as transporting hazardous materials, providing last-mile residential delivery, or servicing the exhibit industry, the focus is on providing niche products that are highly customized to the customers’ individual needs.

Use of technology and application of AI:

Today, companies depend on decades-old legacy applications to keep their business running. These applications are often in dire need of replacement, yet the capital investment required is often redirected to other needs, such as upgrading equipment. Companies often know they could take more advantage of technologies such as mobile, cloud, and analytics, but are usually hampered by a lack of funds. AI can be a game-changing technology that could be deployed in many possible applications. Other technologies that organizations should be starting to deploy include telematics, as required by the government electronic logging device mandate, blockchain, and physical and digital robotics. How deftly businesses navigate the investment and value derived from the deployment of these technologies will determine their very existence over the next decade.

There are two key issues that companies will face in the deployment of these technologies. First, they will need to find ways to fund these investments in terms of both capital expenditures as well as an increase in short-term operating expenses. Second, businesses without the skill sets necessary to deploy these technologies will require upgrading some skills for existing employees, as well as augmenting these workers with new skills brought in from the outside through consultants and new employees. This will take creating teams with skills in process automation, customer journey mapping, statistics, and data science which do not exist today in many organizations today.

Proposed Roadmap

Within each company, there are several processes that could benefit from the implementation of machine learning, natural language processing, and robotics.


  1. Predicting customer churn: Customer churn occurs when a customer stops shipping because they have switched to a different carrier. Some customer churn is good — customers that are unprofitable or have limited growth potential are customers that businesses would rather not have. On the other hand, there are many customers that companies don’t want to lose to competition. Managing customer churn using machine learning involves using data sets such as customer complaints, damaged goods claims, competitive pricing data, and others. This will allow the company to reduce cost because it is significantly more expensive to sell to a new customer than to an existing customer.
  2. Prediction of equipment maintenance schedules: Based on the large amount of data generated by telematics devices that can be installed in equipment, machine learning can predict the maintenance schedule for equipment before they breakdown. This will both reduce cost as well as create a differentiated customer experience.
  3. Semi-automated customer service for routine queries via email: Currently, emails are typically processed by human customer service representatives. Using machine learning, emails can be analyzed for intent, and then either automatically responded to in the case of simple requests or routed to humans for more complex requests. This will reduce cost and create differentiation of the customer experience.

Each of the machine learning initiatives outlined above will require getting the right teams in place in order for them to be successful.


  1. Predicting customer churn requires a cross-functional team comprised of sales, marketing, customer service, and IT. This initiative should be co-sponsored by the SVP of sales and marketing. The CIO should also be a sponsor, as this project fits in nicely with an overall business and IT aligned strategy. The team should develop a project charter, scope, and expected outcomes as a first step. The team should then build the business case, including the various data sets they will need in order to be successful. These include customer complaints, damaged goods claims, competitor pricing, and other information. The team will work with a machine learning solution architect to determine the machine learning models, hardware, and software that will be used in this project.
  2. With a larger number of equipment now starting to deploy telematics devices which are capturing large sets of data on engine performance and driver behavior such as acceleration, braking, and rests taken, companies have the data necessary to predict when and how the equipment needs to be maintained. This initiative will require the involvement of teams from the pick-up and delivery operations, as well as customer service to determine project scope, timelines and expected outcomes. Equipment breakdown is a costly problem as it creates customer dissatisfaction, churn and reputational damage.
  3. With a definitive channel-shift for customer interactions, emails continue to increase in volume as customers find it easy to correspond via this asynchronous channel. However, the ability to interpret email intent and to respond via the most appropriate action happens today in a 100% manual human-centered approach. Using machine learning, project teams comprised of customer service and IT can create the scope for a pilot program to determine intent and to create a completely end-to-end automated solution for some easy emails and to route more complex emails to human customer service reps for action and resolution.


There are several processes in trucking companies that can benefit from the use of NLP.

  1. Customer service: Particularly in terms of routine inquiries such as order status and shipment status in a contact center environment where customer queries come in via both email and chat sessions, many queries can be handled using NLP. Today, hundreds of agents in customer care or call centers answer routine customer inquiries like order or shipment status that come in via phone, email or chat. These communications could easily be handled via an automated fashion using NLP. This will allow the company to redeploy these agents to handle more difficult issues, such as managing damaged shipment inquiries and refunds. This redeployment of agents will create cost leadership as well as a differentiated service offering as routine tasks will be handled much more quickly and effectively.
  2. Management reporting: One area where NLP can be applied is to create a voice-enabled assistant for routine management metrics reporting. For example, daily sales numbers can be inquired via an Alexa- or Siri-like interface. Most management reports today are either produced in spreadsheet format or using some kind of visualization tool such as Tableau or Cliq. Yet, often times, the person reading the report is looking for only a particular piece of information. A future state in management reporting could utilize NLP to ask for the specific management information using an Alexa- or Siri-like interface. This would create a differentiated way of managing information.
  3. Reputation and customer churn analysis: Call center recordings, as well as online customer complaints on social media, can be analyzed using NLP for both reputation management and predicting customer churn. By examining this data for issues related to damaged shipments, missed pick-ups and deliveries, and other information, a company can predict customer churn and enable sales and customer service teams to take corrective action before churn occurs.

Fundamental to the success of any initiative that involves the use of sophisticated and cutting-edge technology is gaining buy-in from business executives that will be using the technology. In the case of customer-service, the SVPs of sales and marketing in each business unit should be heavily involved in driving these projects, as they will require serious change management efforts in redeploying and potentially reducing the number of agents in call centers. Additionally, leadership from the IT side will have to be assigned to all the three use cases mentioned earlier. Trucking CEOs should consider involving outside third-party technology and change management experts who have been successful at deploying NLP projects. This will ensure a higher percentage of success for the project.

The three proposed process areas where NLP can be potentially deployed fit in nicely into the overall business and IT strategy of creating both differentiation as well as cost leadership in the marketplace. Customers are getting more and more used to having information available on demand, so an NLP-enabled automated inquiry system for routine queries will create an enhanced customer experience. This is essential to both retain and grow customer wallet share.

There will be some serious technical challenges in deploying these changes that should not be trivialized. Primary among the technical issues will be the ability to create the appropriate interfaces to extract the necessary information from the legacy applications that exist in the company. Second, the ability to have high confidence in queries answered via NLP will require the training of the machine learning models so they become more accurate over time. This will require time, effort, and money, and should not be overlooked as a critical success factor.


There are many applications where robots can be used to augment human intelligence in the current operations.

  1. Semi-automated picking and packing warehouse items: Today, most warehouses are operated manually. Machines such as forklifts are operated completely by humans. In the pick and pack process, the deployment of collaborative robots, or cobots, can create a cost leadership by optimizing the pick process by using cobots to travel distances within the warehouse more quickly than humans. This can increase the velocity of the pick process, creating more revenue and profit for the company.
  2. Moving items within a warehouse or terminal: Using robots to move items within a warehouse or terminal can significantly improve a company’s cost leadership. Robots can get to locations much more quickly, and can be deployed 24 hours a day, unlike humans. Second, safety can be increased by using robots instead of humans for dangerous activities or handling hazardous goods.
  3. Improving truck safety in long-haul operations: In long-haul truck operations, safety can be vastly improved with the use of shared control robots in the management of truck fleets. This can create a significant differentiation relative to competition.

The rollout of robots in the context of the processes outlined will require overcoming several challenges. First, companies will have to consider the investment required in the procurement, training and deployment of these robots. While in the long run, the deployment of the robots will result in a company requiring less people, there will be parallel run costs that will increase costs in the short run. Second, there will be cultural shifts that will need to be made in training the workforce in how to operate in the new world where there will be combinations of people and robots performing tasks that are currently only done by humans. The key personnel that will need to be involved in the deployment of robots in warehouses and terminals will be the supervisors and managers who run terminal operations, in addition to staff from technology teams to help deploy these robots. Finally, companies will need heavy involvement of the human resources teams to help with the change management process necessary for this initiative to be successful.

The deployment of robots in both warehouse as well as in the long-haul operations fits nicely with an overall strategy of making both these operations more cost effective and safe. In the long-haul operations, trucking companies should see a reduction in the probability of a catastrophic accident caused by human error with the deployment of a shared control robotic truck where human error can be corrected by the robot. This can create a huge impact on profitability given the fact that most trucking companies are self-insured for a catastrophic event involving an error by one of their drivers.

Some of the other technical considerations required for implementation in both the warehouse and long-haul use cases will be the swapping out of certain older equipment with newer robotic enabled equipment, or perhaps retro-fitting older equipment with newer robotic capability.

Plan of Action and Criteria for Success

The use of AI in a trucking company will clearly have an impact on the way that work is done in the future. Several categories of jobs will be impacted from the use of AI in the previously described use cases. The fundamental thing to realize is that while the entire role done by a human today in these job categories will not be completely eliminated, there will be many tasks that are done by humans today that have the potential for high levels of automation in the next 5-10 years using AI. large portion of the jobs that will be impacted fall into the category of clerical and customer service roles. There are several aspects of current clerical workers’ jobs in a company, such as financial analysts, purchasing managers, billing clerks, credit analysts, and others who will have large percentages of their tasks being done in the future by AI. That will create some potential for job elimination or job shifting to categories of work requiring reskilling or upskilling. Routine customer service tasks, such as billing inquiries and providing shipment statuses, will also become highly automated, therefore changing the kind of work performed by service representatives in the future.

Take the example of the hundreds of customer service representatives that are currently employed. For routine tasks like billing queries or shipment tracking, machines will be easily able to handle such tasks at high levels of speed and accuracy. However, there will be tasks within the customer service function, such as handling a cargo claim for an irate yet important customer, which will require the skills and experience of a seasoned customer service rep to ensure the task is handled in the most judicious manner. In order to set the boundary lines clearly between humans and machines, companies will need to deploy a team of AI experts who can determine which tasks are best handled by AI and which ones are best left to humans. This team will need skills that do not exist in the company today and will therefore require hiring experts from the outside who are well versed in the implementation of AI.

There are a few ethical concerns that we expect to come up from the use of AI in the use cases presented earlier.

One of the biggest ones involves bias introduced into the AI decision making process as a result of bad data. Companies are increasingly aware of garbage in, garbage out issues caused by bad data in management reporting and decision making.

In the case of AI, having bad data can have some serious consequences as the human may not be able to catch the error caused by the machine before it is too late. In order to handle these issues, companies will need to invest in resources that are adept at managing data assets in a very careful manner to ensure that the data used in the AI is secure, clean, current and without bias. Another ethical consideration will be the issue related to the use of autonomous or semi-autonomous trucks in future long-haul operations. Currently, in the case of a human driver, while it is not easy to determine fault in a catastrophic accident that causes fatalities, at least the liable parties are all humans. In the future, with an autonomous truck being driven by AI, determining fault will become much more difficult and will therefore involve new skills in managing the regulatory forces in play.

In conclusion, trucking CEOs can use the above to carefully craft a roadmap on the use of AI in their businesses. Not doing so is not an option.

Written by: EXL Service

Contact US