Every day, I’m asked, “How do you perform analytics on all the data being collected to, ultimately, impact your customers’ material business outcomes?”
With data from hundreds of thousands of vehicles coming from on-board systems, our safety program and third-party data feeds, SmartDrive is in a unique position to provide intelligent data to help fleets succeed. We’ve developed the processes and procedures to provide the results fleets require to efficiently – and effectively – run their businesses.
In this three-part series, we examine the steps to build a useful analytics solution, some of the methods leveraged to achieve the desired goals and, as a result, new solutions that make fleets safer and more competitive. In this week’s post, I’ll use a real-world example (driver tenure) to show the power of the process.
Effective Steps Toward Building Analytic Solutions
Having analyzed billions of data points over more than 15 years, we know that these four steps – if performed in sequence with a deep understanding of the end result – will directly translate to business value when applied properly.
We know, for example, driver tenure and its impact on the safe operation of a truck is a very important topic to fleet operators. The premise is that more experienced drivers will have better safety performance than those with less driving hours, and the learning curve can be accelerated with effective coaching and training. So, how do we get there?
Step 1. Understand and articulate the problem you are trying to solve.
Focus on value for users rather than technology. Using technical terms like ‘Machine Learning’ and ‘Artificial Intelligence’ in the ideation phase simply doesn’t further the understanding of the business value. Just because technologies exist (and are the new “shining object”) doesn’t mean they should automatically be used. Focus on what you’re trying to solve and then determine the best solution to get to the answer.
Driver Tenure: What is the problem/question we are trying to answer?
Turns out we are looking for a couple of things …What is the difference between tenured vs. drivers with less driving experience? What risky behaviors are more prevalent in newer drivers? And, what is the cross-over point when a driver should be considered “tenured?
Step 2. Identify data sources and the type of data available to answer your business questions.
Contrary to most common practices, this phase should always come after identifying the business problem you are trying to solve. Similar to what I stated in step one, too often, someone analyzing a problem will focus on the data available first – clouding the issue and the process. It’s helpful to classify the type of data into the following four groups:
- What data do you already have that you need?
Oftentimes, traditional data-driven decision-making stops at this point. Unfortunately, when data requirements end at this stage, customers are often left feeling unfilled with the end solution.
- What data do you have that you don’t need to include in your solution?
Remember to only include data if it clearly helps answer a specific question. Once again, just because it’s available doesn’t mean you have to use it
- What data do you not have, but can get?
Quite often there is publicly available data, or a simple data-enrichment step, that can get you the additional data you need.
- What data do you need to answer the question, but do not have – and there is no possible way to get it?This is a showstopper but should not be ignored. Identifying this upstream significantly reduces downstream problems. If the answer is no, you may need to revisit the first question and rescope the analytic solution.
Driver Tenure: What data do we have?
The good news is SmartDrive captures several key data points that will help us answer our questions:
– Real-time vehicle location, speed and direction
– Driving hours
– Triggers (such as hard-braking, shock, hard-acceleration, etc.), which are leading indicators for risky situations
– Observations – events that have been identified, visually, as a safety risk, i.e., short following, failure to come to a complete stop, distracted driving, etc.
– And more!
Driver Tenure: What data are we missing?
Weather is an important component to determining risk. While we don’t have weather from the on-board system, we can get weather for a particular location (which we do have as a data point) using an external service. Traffic conditions also can contribute to the effectiveness of the safety assessment.
Driver Tenure: What data is unavailable?
The driver’s driving history; this would require extensive research and raises possible privacy concerns.
Step 3. Define the audience to whom you are communicating your results.
This will largely inform what the deliverable (or what the product) should do and how the data should be visualized. Typical approaches may include dashboards, automated reporting or alerts – but understanding your audience is critical. It’s important to understand their expectations prior to communicating the results.
Driver Tenure: Who is our audience?
Coaches who will utilize the information to build out a coaching program that focuses on risky behavior prevalent in newer drivers.
Step 4. What additional insights have we learned?
This step is a natural result of performing steps 1-3, possibly providing new-found knowledge that answers additional questions and business opportunities, while potentially uncovering additional use cases.
Driver Tenure: Additional knowledge gained
What is the cross-over point between less seasoned drivers and those who have shown a material reduction in risky behavior; i.e., how many days/weeks/months does it take to on-board a new driver and convert into a safe driver? Also, what are the most significant things to coach/train on that result in the most material change to safety?
In our next blog posts, we will examine several other situations that can be addressed with data and analytics. Using the same framework, we’ll focus on the unique characteristics of the use cases and the impact of the results on your business.