When I talk about metrics with many people in the U.S. legal industry, the conversation frequently turns to the use of UTBMS (Uniform Task-Based Management System) codes. The UTBMS system has been around for a while (Jim Hassett wrote a nice piece about UTBMS code history and resurgence posted here) and I know many in-house lawyers are fans of the UTBMS system. I, however, have never been a fan of the system. As I have worked over the years to get lawyers interested in becoming more efficient, more productive, and to drive quality higher, and as the metrics discussions turn to UTBMS codes, I have become more frustrated. To me, UTBMS codes take us in the wrong direction. So, recognizing the argument that sometimes you make do with what you have, let me explain why I think UTBMS codes are not the way to go.
Let’s start with the basic premise of the UTBMS code system – self-reporting. Timekeepers are required to self-report what they do throughout the day, using the codes, typically in six-minute increments. Self-reported data, especially when it involves things like measuring time spent on tasks, is notoriously unreliable. We can explore that a bit more.
Most firms require timekeepers to submit their time daily. Some timekeepers are diligent and record time throughout the day; others record time at the end of the day or intermittently. There are still many instances when timekeepers don’t meet the daily requirement. As the length of time between the event and the recording of the event increases, accuracy decreases.
Even when time is entered contemporaneously, we have a fair amount of inaccuracy built into the system. Assume someone does a quick task on a file that takes four minutes. They probably will enter that task as a six-minute task with a UTBMS code. That means there is a 50% error rate in the time entry (2 minutes/4 minutes). Put another way, the timekeeper recorded 150% of his actual time. Of course, the error can go the other way. The timekeeper might spend eight minutes on the task, but only record six minutes, a 25% reduction.
It is possible that with enough time entries using the same UTBMS code, these variances will even out. But, it also is possible that a given timekeeper always rounds up (anything over six minutes becomes 12 minutes) or rounds down (anything less than 12 minutes becomes six minutes). You won’t know when you look at the reports from your ebilling system. In fact, it would take some investigative work to find out. You would have to do time and motion studies of each attorney, and compare your results to what the attorneys records (and at the same time avoid what is called “observer-expectancy effect” – having the lawyer’s behavior change because he is being watched).
To further complicate matters, we are assuming that the lawyer knows whether he spent four, eight or eleven minutes doing a task. People can be good at gauging time. If each lawyer recorded his or her time immediately after finishing the task, the entries might be reasonably accurate (again, we wouldn’t know for sure unless we did some studies). But, if the lawyer doesn’t enter his or her time until later, the accuracy rate will fall. What might have been a reasonable “I spent about four minutes doing this” becomes more of a guess or fill in the blank (“I worked on this starting around 10:15 and had a meeting at 10:30, so I probably spent 15 minutes on it).
In my early days of learning lean, I did a lot of time and motion studies. The time people thought it took them to do something varied considerably (that is, often off by more than 100%) from the time it actually took them to do something. We never used self-reported data because it simply wasn’t sufficiently accurate.
UTBMS codes also raise the problem of coding error. Some portions of the UTBMS code system are more robust than others (e.g. litigation versus corporate). We know that to get anything out of the system, we have to make sure each timekeeper is well-trained in how to apply the codes to work. The training process introduces another source of variance.
Multiple trainers often have different versions of the truth – which code to use in a given situation. This variance becomes more pronounced when the trainers are from different law firms. We also have variances that creep in over time. The timekeeper walking out of the room applies the codes one way, but over time her practices creep. She shifts how she applies the codes, either because a particular client views tasks differently, or because the partner she works for has his own way of doing things, or just because she adopts practices that are easiest for her. The more timekeepers there are, the more variance in approaches, the more the quality of the data decreases.
Other Variable Inaccuracy
Of course, another issue with UTBMS codes is that they ignore the many variables that impact performance. Let’s look at some examples.
Assume we have two single-plaintiff lawsuits where the issues and facts are relatively similar, and both are in the same jurisdiction so the law applying in each case is the same. Lawyer 1 is handling the first case and Lawyer 2 is handling the second case.
One of the premises of the UTBMS code system is that we can use the data to compare performance by lawyers. We assume we can look at the time spent by Lawyer 1, compare it to Lawyer 2, and draw a conclusion about which lawyer is more efficient.
But, there are far too many variables with values we don’t know to make such a judgment. The following list contains just a few examples of those variables as they could apply to one piece of the case-the plaintiff’s deposition:
- Was the plaintiff in one lawsuit very experienced with depositions, and the plaintiff in the other inexperienced?
- What impact did the plaintiff’s attorney have in each case (preparing the plaintiff, at the deposition, otherwise in the case)?
- What other factors affected the deposition (e.g. mood of each participant, logistical issues)?
We can extend this list, and then create similar lists for each aspect of the lawsuits. Because the process for each lawsuit is largely uncontrolled, the number of variables affecting the lawsuits and the ways in which each could affect each lawsuit or combine makes the comparison weak, at best. While it is possible that over the course of a large number of lawsuits, the variances may even out, in reality we seldom if ever have sufficient lawsuits to make the comparison. In any event, we don’t track and measure the other key variables affecting performance on each of the lawsuits so it would be difficult to draw any conclusions about the lawyers.
Measuring Effort Not Value
Yet another challenge of UTBMS codes is that they measure effort, not value. Even putting aside other issues with UTBMS codes, all they measure is effort. We only know how many minutes it took to accomplish the task. As hopefully everyone in the legal industry knows today, effort does not equal value. Whether a lawyer spends one hour or 10 hours on a task, we don’t know how that task ties to the value of the matter, such as the outcome of the lawsuit.
Let’s go back to our two lawsuits. In each case, the time required to prepare for and depose the plaintiff was 10 hours. The effort required in each case was the same. But, in the first case the plaintiff’s deposition was pivotal to the outcome. In the second case, the plaintiff’s deposition had no impact on the outcome. The effort was the same, but the value was quite different. Because we aren’t measuring value, we aren’t comparing like things. If we magnify this problem by considering that for each step in the lawsuit we only measure effort and not value, you can see how we get to a big disconnect between what we are measuring with UTBMS codes and valued outcomes.
Another way to look at this is through the lens of behavior change. When we consider those two depositions, each taking 10 hours, we come to different conclusions about what to do if we also consider value. For one deposition (little value), we would want to reduce the time (effort) spent on the deposition. For the other deposition (high value) we may want to increase the time (effort) spent on the deposition (for those who would say we don’t know in advance, the simple case could be taking more time in the deposition itself). Without evaluating value for each piece of the lawsuit puzzle, we don’t know whether to increase, decrease, or not modify the effort. And, if we do adjust the effort, there is not guaranty we will positively impact value.
A Tool Past Its Prime
Back in the mid-1990s, when UTBMS codes first came about, we really did not have great alternatives. We were doing almost nothing with alternative fees and we knew very little about applying lean to services. We were, however, generating a lot of billable hour information, we already were capturing task descriptions, and putting some standardization around what we already were doing made sense.
In the past 20 years, we have learned a few things. Today, we have better tools. We can use value stream maps, lean accounting, and business-focused metrics to track processes, measure value, and make changes tied to value.
Continuing to spend a lot of time and effort on UTBMS codes, in my opinion, does not add value to improving the efficient, productivity, or quality of legal services. I don’t want the least effort low value service. I want a high value service delivered efficiently. To accomplish that goal, I want to focus on what adds value, measure the value, and ruthlessly eliminate things that don’t add value or detract from the value. I want the entire improvement process to deliver what I need (meaning don’t provide more or less service than I need).
Send me an email or a tweet and let me know if you disagree, why you think there still is a place for UTBMS codes.