Key Takeaway: Quality data is the starting point for legal teams hoping to produce reliable outputs. For data to become ready for AI, teams must have a shared understanding of what “good” looks like and how to use data accordingly. The best possible dataset requires ongoing alignment, sustained attention, investment, and dedication to the follow-through that it requires to get there. As a result, IP teams gain defensible, scalable insights they can trust.

What happens when you assume your data is understood, but it isn’t? Clean, well-prepared data is the backbone for everything IP teams are trying to do with AI, analytics, and automation. Part one of this series, Data Is the New Bacon, explores the value of this foundation.

Organizations that aim to prioritize data quality must confront the challenge of continuous improvement that clean data requires.

Data may be complete and consistent, but that does not mean it can be used consistently without interpretation. There are essential questions left to answer before your data is ready: What does the data represent? How should it be interpreted? What are its limitations?

This foundational understanding not only accelerates but enables teams to scale AI and analytics with confidence.

The Confidence Trap in Legal Data

IP departments often think their data is usable simply because it produces outputs: reports, dashboards, and metrics that appear consistent. The surface-level perspective obscures inconsistencies and gaps in how the underlying data is understood.

That kind of reportability creates the illusion of proper alignment and interpretation of source information. As with anything made from questionable ingredients, the output may look fine until someone has to consume it. This is the distinction organizations are running into: availability is not the same as reliability, and reliability is not the same as readiness.

The Hidden Work Behind Reliable Data

Teams move work forward and make decisions based on seemingly reliable data. Behind the scenes, they are compensating for gaps. They reconcile inconsistent data, account for exceptions that don’t fit standard processes, and bridge different interpretations of what the data means. That involves manually validating outputs before they are shared, adjusting reports to account for inconsistencies, or relying on side analysis and institutional knowledge to confirm whether the data is trustworthy. This is the invisible work that is necessary to make outputs usable. Teams that capture and standardize the insight that stems from institutional knowledge create the foundation for the outputs they are seeking.

Functions outside of IP are recognizing this dynamic as well. Gartner has noted that many organizations lack confidence in whether their data is truly ready for AI. They’re seeing initiatives fall short when readiness is lacking, reinforcing how gaps in understanding and alignment undermine well-intentioned efforts.

The Reason Misalignment Persists

The evolving nature of IP data is largely responsible for this problem. New filings, status changes, and ongoing activity reshape the dataset in real time. Teams working with the data operate under the assumption that they understand their data well enough to rely on it. This gap becomes visible in practical ways. Patent status is a common example. Organizations often use custom values, and even when labels are standardized, the application of those statuses shifts over time or across teams. As a result, the same status doesn’t necessarily have the same meaning. 

Application relationships present a similar issue, as continuation, divisional, and provisional relationships have distinct legal implications that may not be captured consistently. These challenges compound when similar information exists across systems.

For example, a status in a docketing system may reflect a legal milestone, while in a reporting or billing system, it may represent workflow progress or financial activity. If those meanings are not aligned, a portfolio report may overstate active matters, understate pending work, or suggest that prosecution activity has slowed when the issue is a difference in how systems define and update status. Teams see alignment on the surface but interpret it differently in practice. This results in inconsistent, flawed conclusions and additional manual effort to explain or reconcile the difference. 

These issues reflect a gap in how the data is understood. Teams building systems and outputs have access to the data, but not the critical business or legal context behind it. Domain experts understand that context but don’t always have visibility into the structure or interpretation of the data once it moves into systems and workflows.

This disconnect is the heart of the issue. While the data could technically be correct, it isn’t interpreted the same by the people using it. This is exactly where reliability breaks down.

The complexity and evolving nature of IP data mean that teams expect this kind of misalignment. The good news is that this gap is manageable, but the first step is recognizing where interpretation breaks down before teams can design a sustainable fix. That means identifying where interpretations are inconsistent, aligning across teams on what data is meant to represent, and maintaining that alignment over time.

AI Raises the Stakes if You Skip the Follow-Through

Research from the RAND Corporation shows that AI initiatives often fail because teams misunderstand or miscommunicate the problem they are trying to solve. If you feed AI data you do not fully understand, you lose the ability to recognize when the output is wrong. Without that insight, teams cannot validate it.

AI builds on these gaps, producing faster outputs that carry the same underlying inconsistencies. 

None of this is unsolvable, but it does require a different kind of commitment. The challenge lies in acknowledging that making data usable requires ongoing alignment, shared understanding, and sustained effort. This is not a call for a one-time data cleanup exercise. It is a call to treat interpretation, alignment, and follow-through as part of the operating model for reliable IP data.

Start with clarity, ensuring that key data points are consistently defined and understood across teams. Create visibility so the people interpreting the data understand how it has been structured, transformed, and used across different processes and platforms. Lastly, build the discipline needed to sustain that alignment over time as the data and processes evolve, preventing definitions and assumptions from drifting. With this strong foundation, AI becomes a force multiplier, reducing manual validation and ensuring consistency.

Follow-through is a key consideration that teams often overlook. There must be a shared understanding embedded into day-to-day workflows so teams interpret data consistently in practice. Leadership is responsible for prioritizing that work and supporting it with the attention and investment that will sustain it over time. 

You Get the Data You Design

Teams facing this challenge are underestimating what it takes to make data reliable to support sound decision-making. Align definitions, resolve inconsistencies, and create a shared understanding of what the information represents. As portfolios, systems, and business needs change, teams must maintain this discipline.

With a shared understanding, teams can trust the results and unlock the full potential of their technology.

Learn more about Epiq IP Advisory and Implementation.


Jennifer Karr, Senior Manager, IP Operations, Epiq Advisory

Jennifer Karr is recognized as a leader in legal operations and IP management with over 20 years of experience integrating law, technology, and business strategy. She has shaped industry standards through thought leadership roles, including serving as co-chair of CLOC’s IP Proficiencies Committee and LegalOps.com Vendor Management Committee. 

With experience managing the entire IP lifecycle, she has a comprehensive understanding of how to strengthen alignment between people, processes, and technology.