If there's one thing Sage understands, it's that introducing AI isn't the hard part. Getting customers to actually trust it and use it is where the real work begins.
Sage is a global technology company that helps small and mid-size businesses manage finance, payroll, and HR. When they set out to embed AI into everyday workflows with Sage Copilot, the goal was straightforward: help customers move faster, make smarter decisions, and stay in control. But they weren't just launching a feature. They were building a system to drive adoption, earn trust, and deliver measurable outcomes from day one.
When Copilot launched, customers were interested but cautious. They had questions: What does this actually do? How does it fit into my workflow? Can I trust it with my financial data?
It's a fair concern. AI can feel opaque or intrusive in any context, but especially in accounting, where accuracy isn't optional. As Niall Geraghty from Sage put it: "We're working with accountancy customers. They're very busy and it’s very important to make sure that the insights and data we're providing are accurate and reliable."
For Sage, that meant making Copilot feel intuitive and transparent—helpful without being disruptive.
To do that, Sage turned to Pendo. The idea was to combine in-app guidance with behavioral analytics to create a continuous feedback loop: guide users in the moment, see how they engage, learn from what's working (and what isn't), and iterate quickly.
As Geraghty described it, "There were two elements: the in-product awareness piece and the analytics side which showed us how people were using the agent and where they encountered friction or drop-off."
Early on, they relied on traditional in-app messaging, but ran into a problem most product teams know well: banner blindness. Users were closing guides before reading them, not because they didn't care, but because they were in the middle of something. "Even if it is something that's really beneficial, when you’re in the middle of a task, it’s just not convenient and they would close the guide,” explained Geraghty.
The fix was using embedded guides to provide contextual guidance placed directly inside workflows, showing users the "how" and the "why" without pulling them out of what they were doing. "Being able to use embedded content and influence the UI to get someone's attention when they're receptive was a real benefit to us," said Geraghty. Today, nearly half of Sage's guides use this format.
Beyond onboarding, Sage leaned on Pendo Analytics, Funnels, and Paths to understand exactly how customers were moving through Copilot. They could see where users dropped off in setup flows, which features drove engagement, and how end-to-end workflows were performing.
When something wasn't working, they didn't have to guess. Session Replay paired with AI-generated issue summaries let anyone on the team watch a real user hit a friction point and immediately understand why. "Anyone was able to look at a funnel and then see a replay of someone dropping off. It puts us in the shoes of our customers."
On the feedback side, they centralized NPS and qualitative responses through Pendo Listen. With thousands of submissions coming in, AI-generated theme summaries made it possible to surface what customers were really saying—fast. "Rather than having to troll through thousands of responses, we started getting automatic theme generation and feedback summaries."
The themes that emerged around control, clarity, and transparency led directly to product changes:
Small changes, but meaningful ones. They didn't just improve usability; they helped close the trust gap.
Sage also wove product-led growth directly into the experience. Rather than pushing upgrades through external campaigns, they surfaced premium features and advanced workflows in context while customers were already engaged and seeing value. The result was more organic upsell conversations, driven by the product itself.
The outcomes were tangible. Thousands of customers activated Copilot within weeks of launch. The VAT Assistant alone reached 5,700+ uses, with around 75% of users completing the workflow and 55% applying the suggested changes. "Click-only" usage, a sign for shallow engagement, dropped from 18.5% to 12%. Customers reported saving up to 30 minutes a day on routine tasks, and sentiment improved significantly after an initially rocky NPS start, landing at a +56 CSAT.
Operationally, the team moved faster too: standardized dashboards sped up decision-making, session replay cut down investigation time, and they developed a repeatable model for onboarding and experimentation they could apply across products and regions.
As Geraghty put it: "Without a tool like Pendo, we would have spent a lot of time guessing and it would have been a lot slower and a lot riskier."
What started as a Copilot launch has become a blueprint. Sage is now expanding Copilot across more products and regions, testing roadmap ideas with real customers before committing, and moving toward AI-driven personalization using segmentation and behavioral data.
They didn't just ship AI—they built the infrastructure to make it stick. And that's what earned them the Pendo Customer AI Innovator Award.
