At Amplitude, we redesigned our homepage to launch a chat-first experience. Then, we added company-level personalization with a feed of recent AI Chat conversations happening across the user's organization.

The goal is to help Amplitude users learn from their colleagues who were already using Amplitude AI Chat on prompting techniques, and create a sense that AI usage was already happening around them.

It worked! With the AI Chat activity feed, the overall engagement improved 10% in relative percentage and the time to adoption shortened by 40%. Better yet, we were able to drive feature engagement improvement without hurting broader product engagement with non-AI features.

Before we launched, an internal concern was raised by multiple teams: surfacing other people's AI interactions might feel invasive. If users felt like their own AI conversations were being watched and displayed in front of the whole organization, they might prompt less or avoid using AI altogether. (We’ve all used AI as our therapist).

The team took the concern seriously. I started by asking Amplitude’s AI Chat about sensitive topics like relationship advice and fortunately Amplitude’s AI Chat is designed to answer data-related questions.

To control risk, we started with non-Enterprise tier customers who we hypothesized would have less concerns about permissions and access controls than big companies on higher-level pricing tiers. I added a link to explanatory text that said, “Why am I seeing this?” And when the user clicked on the link a tooltip said, “By seeing other peoples’ prompts, the whole organization can discover, learn, and iterate on data analysis together.”

This was an experiment to explain why the user activity feed exists, including what it's surfacing, why it's intentionally public, and what benefit that creates for the user. Did it work?

Among users who viewed the message explaining the user activity feed, 64.6% went on to send an AI Chat message, vs. 47.4% for those who didn't — a ~30% relative lift.

My first instinct was to wonder if we were just measuring curiosity: maybe the kind of person who reads a tooltip is also the kind of person who is more likely to embrace AI. So I dug into the data to remove that selection bias.

Even after controlling for it, the explanatory message was doing real causal work. And the effect was concentrated in exactly the right place: Lower activity users (< 3 AI messages per week) who saw the message converted at 52.8% vs. 34.6% for those who didn't — a 52.6% relative lift.

Meanwhile, power users found their way into AI regardless. In other words, the tooltip with the explanatory privacy message moved the needle on everyone, notably the reluctant user who needed a reason to engage rather than a reason to avoid AI Chat.

In conclusion, the message about organization-wide sharing of user activity was communicating a product philosophy: Amplitude AI Chat is public to everyone within the organization by design, because seeing how others use AI is itself valuable. When users understand the intent, they opt in.

The story validated something larger for us: In the age of AI, users are eager to learn and catch up with their peers who are supercharging their data analysis workflow with AI. As a growth team, we can and should be more opinionated about when to break the norm and be the expert in teaching our users about how we successfully operate as a team in our product with AI.

Founders deploying AI agents across sales, marketing, and customer service are closing more with smaller teams. 65% more sales leads. Less headcount.

Download the free Practical Guide to Agentic GTM for Startups and start building the stack your competitors dream of.

Get the Free Playbook