Introducing the Next Frontier of Analytics: Automated Insights

For years, PMs and analysts have relied on dashboards to understand what is driving growth. AI promises to change that, but most AI analytics tools today still only tell you what happened, not why. Without causal understanding, it’s incredibly rare that they can effectively recommend what to do next.

This is why we are so excited to announce the newest capability of Amplitude's AI analytics platform: automated insights. This makes Amplitude the first AI platform that is designed to think and act like an expert analyst.

Zeroing in on the hardest problem in analytics today

To truly understand what is happening in your product, you need to get beneath the surface of the data. Early signals can be misleading and often point you toward explanations that do not hold up. Getting to the real story usually takes days of digging.

We experienced this firsthand at Amplitude. After launching Amplitude Made Easy last year, we saw a spike in data setup rates and assumed the new setup experience was working exactly as intended. At the same time, though, our overall conversion rate dropped. More people were setting up data, but a smaller percentage of signups were making it through the setup flow.

It took hours of checking segments, talking to teams, and reviewing releases before we uncovered the actual cause. A marketing campaign in South America was driving a large wave of unqualified traffic. Signups surged, but very few of those users converted. All the clues were in our data, but it took too long to piece them together. By the time we reached the real answer, the campaign had already spent its budget.

That experience made something clear. The hardest problem in analytics today is not in accessing the data. It is making sense of it quickly so we can do something about it.

Why “aha” moments take so long

Most PMs know this feeling well. You spot an anomaly, then start the long process of trying to explain it. You check every segment you can think of. You compare Android and iOS trends, look at cohorts, and examine geography or device patterns. You review experiments, releases, and campaign activity to see if anything overlaps with the change. You build and rebuild charts or export data for deeper analysis.

All the while, the clock is ticking.

How bias impacts causal analysis

Investigating data is also limited by your own experience and assumptions. Humans naturally look at the explanations that come to mind first. You might check the impact of a recent launch, but completely overlook a campaign. You might look at device differences but never think to check specific user groups or regions.

If you are new to a problem or do not have full context, your investigation is shaped by guesswork. This is no one’s fault. It is simply the reality of manual analysis.

High-quality causal analysis takes time, imagination, and context. Most teams do not have enough of any of these.

Building AI that gets to “aha” faster

The good news is that this type of analysis is not random. After years of watching how thousands of PMs, analysts, and data teams work, we realized something important. When people investigate product changes, they tend to follow the same set of steps, in the same order, every time.

Once we recognized that pattern, we began building those workflows directly into Amplitude and automating the parts that require the most manual effort.

Today, Amplitude's automated insights can replicate an expert's standard analysis process in a fraction of the time. It executes various chained tool calls, thinking like an analyst would. It searches charts and reports to investigate experiments, campaigns, and product releases. It incorporates business context to connect dots across relevant sources into a clear, data-backed story.

Our automated insights capability can review all that context and present clear findings and recommendations. If you want to examine the AI’s logic, it cites sources (including existing content and any new materials it created along the way) so you can verify the analysis or continue in a new direction.

It works like an expert analyst, systematically analyzing data on your behalf.

The Cursor moment for analytics

AI has the power to revolutionize how we work. Just like how Cursor used AI to reimagine coding, boost coding productivity, and make it accessible to everyone. We believe our new AI Analytics Platform will do the same to create a new generation of analysts.

Today, teams can leverage Amplitude’s AI platform to unlock deeper insights. It can reliably tease out data patterns to understand exactly what happened and hypothesize why.

To find out more, watch our latest demo here: