Locus is a logistics technology company that helps enterprises optimize and automate their supply chain operations, from order management and route planning to last-mile delivery. Its platform serves retail, e-commerce, 3PL, manufacturing, and distribution companies across 30+ countries. Planners, dispatchers, and operations teams use Locus daily to manage complex fulfillment workflows at scale.
Software
201-1k
NA
Locus was preparing to launch Mycroft, an AI copilot for logistics operations. But the team needed to answer a fundamental question: were they building for how users actually worked, or how they assumed users worked? The dashboard is dense, and without a clear picture of real behavior, Mycroft risked automating the wrong things.
Locus used Pendo's feature usage analytics, click paths, and funnels to map how planners actually navigated the product. That data helped the team separate "ASK" moments (times when users needed information) from "ACT" moments (when automation could save the most time). Those insights shaped how Mycroft was trained and which workflows it tackled first.
Workflows that previously took 8-12 clicks to complete were reduced to a single AI interaction. The workflow scoping cycle shrank from four-to-six weeks to under ten days. 69% of early testers rated Mycroft "extremely valuable," and common tasks were completed 40% faster in pilots. Pendo also became the shared language aligning product, AI, engineering, and support teams around the same evidence.
Locus's dashboard is a workhorse. Planners use it to manage routes, reassign drivers, investigate delays, and track shipments. But like many enterprise platforms, it's dense, and every user navigates it a little differently. When the team set out to build Mycroft, an AI copilot that could answer logistics questions in plain language and automate repetitive tasks, they quickly realized they had a problem. They didn't fully understand how their own users moved through the product.
"The main thing while we started this research was, are we building this around the actual users, or are we just assuming things about them?" says Tanvi Palkar, a UX Researcher at Locus who played a central role in shaping Mycroft's direction.
"Pendo gives an unbiased view of our users. It shows real behavior. It doesn't show opinions. That's the best part about it."
— Tanvi Palkar, UX Researcher, Locus
Because Mycroft's training data was grounded in actual usage patterns, the results were immediate. Workflows that previously took 8–12 clicks were compressed into a single conversational interaction. In pilot sessions, teams completed common operational tasks 40% faster. New users onboarded significantly quicker thanks to Mycroft's "Teach Me" guidance, which was built around friction points Pendo had identified.
Early validation was strong: 69% of internal testers rated Mycroft "extremely valuable," and 58% said they were very likely to adopt it. ACT automations hit roughly 70% success rates on day one because they were based on the highest-frequency, most consistent workflows the data had surfaced.
The project has since opened the door to several follow-on initiatives: expanding Mycroft across more product surfaces, using Pendo cohorts to trigger proactive AI guidance at common friction points, and tailoring Mycroft's responses based on whether a user is a novice or an expert.
Tanvi doesn't hesitate when asked what life would look like without the tool. "We could still get the data," she says, "but it would take significantly more time and effort to turn that data into something meaningful. I would have to spend weeks stitching together events, writing queries, debating what the data actually meant. Pendo removes that overhead. It translates behavior into clear stories you can act on."
Pendoがどのように組織に価値を提供するかをご覧ください
