In my neighborhood, it happens on Wednesdays and Thursdays. It is a muffled sound in winter. But, in spring and summer with windows open, you can hear the heavy duty diesel engines as the trucks patrol the streets. Wednesday and Thursday are trash collection days.
Each day, my wife and I take our constitutional in our neighborhood. This is in part a holdover from walking our family dog. He died last year, but we stayed with the walk. Because we have circled our neighborhood one or two times a day for 10 years, we know the patterns. We can predict, with amazing accuracy, what will happen and at what time.
Trash days are easy to predict. The main variables are trash hauler, number of buckets, and recyclables or no recyclables. Several trash haulers work our neighborhood (the subject of many pricing discussions on the neighborhood social media site). We have identified the factors that tell us which hauler works which houses. Certain houses use two large buckets, but the majority use one. And, depending on the hauler, the pattern for collecting recyclables varies (e.g., one time a week or every other week). We rock at predicting our neighborhood’s trash patterns.
What I mean, of course, is that we are good at predicting trash. The training has been valuable—we are healthier for the walks. But, the actual information we generate has no utility. Who cares what the trash hauling pattern is for our neighborhood? No one, even the trash haulers, wants the information.
Welcome to predictive analytics in the legal industry. We have many companies pushing their data analytics skills, with some focusing on predictive analytics. One popular area is predicting legal spending. The analytics may focus on entire law department budgets or on individual matters. For many, the idea is to train the learner (computer speak for teach the software) to look for spending patterns by looking at invoices. The problem is that spending data in the legal industry is garbage. And, as we all know, garbage in, garbage out. So, studying past spending patterns to predict future spending patterns is similar to knowing the waste hauling patterns in my neighborhood.
Examining the Garbage
Legal spending data is garbage. That is a strong assertion so you may ask what I have to support it. Start by breaking the billable hour process into components. It begins with each timekeeper recording what he or she does in the timekeeping system. We know there is variability in the system. Timekeepers enter time at their leisure. Some do it contemporaneously, but most do it in batches. For example, some enter time at night, some in the morning for the prior day, and some at the end of the week. The greater the gap between the time worked and the entry, the greater the inaccuracy of the time entry.
Even if the timekeeper enters time at the end of the work (say, every hour), timekeepers vary on how they code time. Some enter basic text descriptions. A basic text description has insufficient information to do a granular analysis of the work. “Draft letter” gives me nothing to analyze or improve a process. Those who use the UTBMS (Uniform Task-Based Management System) codes force their time into boxes and the descriptions match the boxes. As with any taxonomy, the gross level of data keeping does nothing beyond time buckets. Surprise! Most time in litigation goes to discovery.
All this is to say that timekeeping varies and each timekeeping entry is unique. But that is the tip of the timekeeping iceberg. The real problem lies in the processes captured by the timekeeping. As my example shows, processes vary from timekeeper to timekeeper.
Consider this simple example. Two lawyers share an office. Both receive the same assignment: review this contract. The processes the two lawyers follow, the time each takes, what changes each recommends, will vary. Play the role of the client and decide which revised contract you use. Translate that work into timekeeping, and you have a mess.
Apply analytics to that mess. You will produce statistics. The mean time to review a contract, the median, the standard deviation. You can produce nice metrics. The greater the number of data points (the number of contracts reviewed), the greater the accuracy your metrics. But, those metrics measure chaos. Some lawyers spend lots of time on irrelevant aspects of the contracts. Some focus on key parts. Some use inefficient processes (lots of waste built in). Some use efficient processes. The list of variables grows and grows. The metrics accurately measure nothing. We can calculate the mean time for a person in the United States to commute to work. But, that tells us nothing given all the variables underlying that metric and the lack of standardization in the processes.
Predicting Waste Has No Value
A counter-argument says that the data we have may be garbage, but it is the data we can collect. It is better than nothing and running predictive analytics on the data does give us information. That information is that our current, waste-ridden, chaotic process for doing contract review takes, on average, a certain amount of time. We know the mean, the median, and the standard deviation. If the process to review contracts stays the same, those statistics will help us predict the cost of future contract reviews, as wasteful as it may be.
That argument has merit. But, if that is all we do—predict waste—we look like the colorful wheel that spins as you wait for the computer to process your command. We need to move past predicting waste and gain control of the processes. With that control, we can go beyond predicting waste, we can reduce the cost of matters by eliminating waste. We can save money rather than predict waste.
A common client complaint is that they have metrics, but fail to get the improvement they expected from using the metrics. This is the “half equation” story. It is the ability my wife and I have to predict trash collection in our neighborhood. That ability has nothing to do with the volume of trash generated each week, the important half of the equation. To help reduce trash collection costs in the neighborhood we need to reduce the volume of trash generated. By doing that, we could reduce the frequency of trash collection and that should reduce the cost of trash collection.
Clients want to do the same thing. They want to work with their legal services providers to reduce the trash. As a team, they want to capture the process for reviewing the contract, improve the process, standardize the process, and drive out waste. The client undoubtedly contributes to the waste in the process. Having it on the team will help reduce the waste it generates. The law firm starts with the product the client handed it. If the law firm spends time removing waste from the contract, the client should look for the causes of the waste. The client can eliminate waste so the law firm avoids having to remove it, reducing review time.
The client can work with the law firm to focus reviews. If the law firm does a great job reviewing the indemnification clause, but the client is okay with the clause, the law firm has inserted waste. The client can guide the law firm on the scope of the review. Working as a team, the client and law firm can integrate the process for contract review across the walls of the client and law firm. They can standardize the process and coordinate timekeeping entries to the process. The timekeeping entries will provide useful, behavioral data.
The Real Value of Predictive Analytics
I have focused on predictive analytics using timekeeping records. This is, of course, the sideshow. The main act is predictive analytics focused on behavioral data that we can use to reduce risk and avoid expensive mitigation strategies. In other words, find what behaviors to change to avoid the lawsuits rather than focusing on how to reduce spending per lawsuit.
Sixteen years ago, I became general counsel of a company of a spin off company. I started with a law department of one. I discovered that I had a docket of 100 lawsuits involving personal injuries alleged to have happened in or near our retail stores. That was the current docket, but plaintiffs kept re-filling the docket. I would settle one lawsuit and another lawsuit would come at us. Each lawsuit cost the company $50,000 to $75,000. We were on a treadmill.
I had been a retail lawyer for several years before I became general counsel and I had no ready explanation for the size of the docket, the staying power of the docket, and the total cost of the lawsuits. I smelled opportunity and charged. In under one year, we worked the docket and brought it to 10 cases. It stayed in that range. The cost per lawsuit dropped to $5,000 to $15,000. I wanted to keep improving, but it was a great start.
We changed the metrics by focusing on behaviors. What triggered the lawsuits? What drove the costs (legal fees and expenses plus settlement costs) higher? As we dug in and learned the facts, we saw the patterns that led us to change behaviors. Putting a wire stand with copies of the recent store flyer outside a store was a good one. The wind would blow, the flyers would scatter on the sidewalk, and slip-and-fall claims would increase. Solution: put the rack inside the store.
Small changes can have big impacts. Tracking and analyzing the right data shows the way. Crunching timekeeping data would have helped me reduce the cost of those 100 lawsuits, but eliminating 90 lawsuits and ensuring they stayed off the docket was the ultimate cost saving approach.
Lean relies on eliminating waste. Analyzing waste to manage waste strikes me as a diversion. Going to the root cause of the waste seems a better way to spend our time and money. Let’s take out the trash rather than focus on predicting how much it will cost.