Audit Log

Audit Log

Reconstruct what happened on a matter to quickly investigate when something goes wrong

Reconstruct what happened on a matter to quickly investigate when something goes wrong

Reconstruct what happened on a matter to quickly investigate when something goes wrong

TEAM MEMBERS

1 product manager, 3 engineers

TIMELINE

Design: 1.5 months; build: 7 months

Background

Background

EvenUp is a legal AI company that specializes in personal injury law. The company was expanding from AI-powered document generationt tool into an end-to-end case management platform for personal injury law firms.

The problem

The problem

When something goes wrong on a matter, law firm staff lack a clear way to reconstruct what changed, when, and by whom to identify the cause.

The solution

The solution

Audit Log provides a complete, scannable history of matter activity with the ability to drill in for further details so teams can quickly trace changes and troubleshoot issues.

Define the use cases

Define the use cases

We explored three potential use cases for Audit Log. Customer research reinforced "troubleshooting" as the core use case, while the “get up to speed” hypothesized use case from leadership wasn’t strongly validated. Research also surfaced "productivity" as a real need, but we determined it was better served through Reporting.

Define the mental model

I started by mapping Audit Log, Notifications, and the Matter Timeline as separate surfaces, but that model broke down once we examined individual activities. Working with the Notifications designer, we developed a simpler model: everything that happens on a matter is an activity. Audit Log captures the complete history, while other surfaces show subsets relevant to their purpose.

Balance completeness and scannability

Rather than defining grouping behavior for every possible combination of activities, I worked with engineering to establish a scalable rule: group the same activities within an hour, while preserving chronology and meaningful distinctions between different activity types. This reduced noise without obscuring the information users need to troubleshoot what happened.

Sweat the details

What I learned at Slack is that craft has no shortcut. What it takes to go from good enough to great is more exploration and more iteration. I developed a methodology where I break up all detailed decisions and iterate on each one to push myself to sweat the details. Below are some examples of different aspects of the details I iterated on.

Impact

Impact

This feature is one of the baseline features that ensure the successful System of Work launches for our 2 initial customers. We received positive feedback from both customer research and UAT. One of the firm owners Tony in one of our research calls explicitly said "this is so much easier to scan than what we had before. I love it."

Since launch, we are seeing healthy engagement with ~27% of the users landing on the page opened the accordions, used filter, saved list views or search. Sentry session recordings showed no major usability issues, while customer hypercare surfaced few negative reports related to Audit Log.