I recently attended the School Transportation News annual convention in Reno, NV. I was surprised to hear the buzz around telematics and ‘Predictive Analytics’ – promising millions of dollars in savings to customers.
For those of you who have not heard the term, here is the simple definition of Predictive Analytics. “Predictive analytics is the practice of extracting information from existing datasets in order to determine patterns and predict future outcomes and trends” (Source)
But my intent in this article is not to make you an expert on Predictive Analytics (I will defer that for my future articles).
I am going to write about basic pre-requisites for making Predictive Analytics a magic wand.
Approx. 15.5 million trucks operate in the US alone and, as per 2006 statistics, these trucks logged in 432 billion miles using 53.9 billion gallons of fuel. It’s a very thin margin business. For every 1-dollar of revenue, the costs are 95.2 cents.
As per USDOT there are approx. 500,000 truck accidents every year and the total costs of motor vehicle accidents are projected to exceed 450 billion dollars in year 2025. Tire and Jump-starts related issues cause 60% of emergency assistance road calls. (Source)
So even a small reduction in accidents, fuel consumption, or on-road breakdowns can have a huge impact.
Here are a few examples of how effective predictive analytics can be applied in the transport industry.
Which driver is at higher risk to have an accident?
Which route is more prone to an accident?
Which location, month, time period is more risky for employee injuries?
Which routes are likely to miss on-time arrival?
Which locations will have driver shortages?
Which routes will be loss making for the company?
Which buses are likely to have breakdown?
What’s the optimal time for oil change / preventive maintenance?
Which parts have higher reliability?
Which mechanics are more productive and do a quality job?
Which customers are more likely to use service again?
Which contracts are likely to be lost or retained?
Which marketing strategy is most effective?
I can go on and predict 100 more outcomes, which are possible with the help of Predictive Analytics.
You may now be asking: what’s the catch? Why is it not a Magic Wand?
Some basics need to be in place before you can accurately predict these outcomes and take the required action towards saving the promised millions of dollars.
Here are the 5 prerequisites,
Consistent, accurate, and timely data:
You must have consistent and reliable data. I am sure you have heard the term “Garbage In…Garbage Out”. So what good is your predictive analytics if you have no confidence in your existing data set? With technology advancement in tablets and telematics, you no longer need to manually input the data in your back office. Instead, these devices can be used to capture and store information automatically at the source.
The only way you can measure the effectiveness of action or process is when you have a well-documented process to perform the action and everybody is trained to execute that action in the same way. Without standardization, it is impossible to identify the root cause of anything based on predictive analytics. You don’t have to start with mature methodologies like ‘lean’ or ‘six sigma’. Sometimes starting with basic process for small function is a big step in right direction. But make sure to have a system to measure and audit the effectiveness of process too; else it will never get grounded.
Reliance upon technology:
This is a pre-requisite to having consistent data or following standard processes. You need to have basic systems to enable your sales team, dispatchers, mechanics and back office staff to perform their daily functions with minimum friction. In my career, I have implemented many technology platforms. Most companies try to pick the best system, which has all the bells and whistles. But the best features to look for in a good system are ease of use, ease of access, and integration with other technology. Remember, your technology should work for you, not the other way around. Finally, adaption of the system has to be from the lowest to the highest level in the organization – which means proper training at all levels.
Strong line managers:
What is the use of predictive analytics when your line managers are not capable of deciphering and acting upon information? They have to know that just getting the cargo on time and safely from one place to another is not good enough. It needs to be done in most efficient and effective way. We have all heard examples of the extent UPS goes to maximize productivity in their maintenance shops. This can be achieved by setting clear goals for line managers, maintaining robust hiring practices, establishing common key performance indicators, and conducting regular reviews at all levels.
When it comes to change, most companies use the term continuous improvement and I am using the term continuous learning. Companies have to create a strong and easily accessible platform for their employees to learn any changes in process or technology. Just having leaders believe in change will have minimum effect unless every driver, every dispatcher, and every mechanic understands the change. In order to get everyone on board with change, you first must establish learning goals, which are linked to employee performance and goals, robust training material, and easy to access learning management.