Peeling the Onion: Six steps to optimize the data mining process

With the legal industry buzzing about metrics and analytics, corporate legal professionals are eager to get their hands on more information. Rather than diving blindly into the data that is housed in their enterprise legal management solution, a better approach involves identifying, first and foremost, the questions that are most important to the legal department and digging for data to help answer those questions.

As corporate legal departments face increasing pressure to do more with less and prove their value to the business, they are looking to data to help them answer important business questions, such as:

• What have we spent this year?
• Why are we spending varying amounts on similar types of legal work?
• Where might we deploy alternative fee arrangements?

Most legal department questions fall into three main categories, ranging from simple to more predictive in nature:

• “What?” These questions can generally be answered in a single report and don’t typically require analysis. For example, “What have we spent?” By firm, by practice area, by matter, etc.
• “Why?” These questions typically require some data mining tools and skills to answer. Reports enable answering such questions, but one report is unlikely to provide an answer to a more complex “why?” question.
• “What’s Next?” Unlike the other two categories of questions, these are proactive questions and typically require several analytic stages. Answering these kinds of questions requires greater analytic maturity, but when done well, it can be instrumental in helping the legal department plan for the future.

The good news is that answering any of the three types of questions doesn’t require days of running reports and combing through data. But thoughtfully answering business questions requires analytic discipline. By practicing the following six steps, the legal department can optimize the data mining process in order to come to valuable insights.

1. Filter. To best answer the legal department’s business questions, first define the universe of data being analyzed. For example, if you’re trying to understand the drivers of an increase in spending for intellectual property work, filter out other matter types, narrow the data to only the years being assessed, and exclude any known outliers or misclassified matters.
2. Visualize. Remember, analysis tells a story. Determine the data presentation that is going to most clearly articulate that story. Use line charts to show trends over time, bar charts for comparative purposes, box and whisker charts to show data distribution, etc. Create the report with the optimal visualization. An analyst can also create grids of data for analytic purposes, but visualizations often point to meaningful answers more rapidly.
3. Drill. Next it’s time to peel the onion by drilling down to greater levels of granularity. If IP spending is growing most in patent prosecution, perhaps there are particular business units or geographies driving those increases. Alternatively, it may be that it’s not a drill-down that’s the most relevant next step but rather a new business question that becomes obvious. For example, if patent prosecution costs are increasing, a natural question to ask might be, “Is the number of patents or the cost per patent increasing?” At this time it would be appropriate to revisit step one and begin the analytic process with the next question.
4. Focus. Identify the elements that are the most material drivers of the story. For example, when examining the components of unit costs for patents, the data may show that in the most recent year the fee paid for filing a patent in Argentina was considerably higher than in other countries. But focusing on the countries with the highest fees may only represent a tiny portion of total legal spending. Instead, after drilling down, focus on the components (countries in this example) that are driving the highest overall spending. Don’t get distracted by the outliers.
5. Interpret. Once several drill-down paths have been followed and tangential business questions addressed, the analyst must distill all of the information to answer the original business question as succinctly as possible. For example, referring back to the original business question related to increased IP spending, the answer might be, “The IP area driving the majority of the increase is patent prosecution, which has seen a slight increase in the number of patent filings versus prior years, and higher fees from two specific law firms for European filings.”
6. Leverage. Ultimately, data mining exercises are only helpful to the organization if used to plan for the future. Perhaps the hypothetical company should consider working with other law firms with European expertise or reduce the volume of certain filings. Modeling what costs would be under a different scenario moves this legal department to the maturity level of the “What’s next?” questions, which engages both the lawyers and the business in planning for the future.