Data & Analytics · Analyze & Recommend
See where users drop off and why.
Doe queries your database, builds conversion funnels from raw events, and compares what users who convert do differently from those who don't. The analysis runs as real code in a sandbox you can inspect and modify. Results are ranked by effect size so you know what to act on first.
Works acrossNeon
What you get.
Doe builds conversion funnels from raw event data in your Postgres or MySQL database, identifies which step has the steepest drop-off, and compares feature adoption between users who convert and those who don't. Connects to Neon, Supabase, and PlanetScale. The analysis runs as inspectable Python code in a sandbox, not a black-box dashboard. Results are ranked by effect size so you know what to act on first.
Activation is 34%. The product team wants to know why.
The head of product shows you the Amplitude chart: 34% of signups reach the activation milestone. "Why are we losing the other 66%?" Answering properly means querying every event between signup and activation for three months of users, building a step-by-step funnel, segmenting by channel, plan type, and device, then comparing the event sequences of users who activated versus those who didn't.
In Amplitude, you can build a basic funnel in 20 minutes. The moment you need to go deeper (compare specific feature adoption between converters and non-converters, or re-segment by a dimension Amplitude doesn't track), you're back in SQL and notebooks. By the time you deliver, the product team already shipped a change based on gut feel.
What changes.
- 01Depth of analysisBefore · Funnel chart shows where users drop, not whyWith Doe · Behavioral comparison showing which features and actions differentiate converters
- 02SegmentationBefore · One segment at a time, each requiring a new chartWith Doe · Every segment computed in one run: channel, plan, device, cohort
- 03AuditabilityBefore · A chart with no source code or query behind itWith Doe · Every query and every line of analysis code visible and editable
- 04ActionabilityBefore · "Activation is 34%." OK, but what do we do?With Doe · Specific features and actions ranked by how much they differentiate converters.
How Doe analyzes your funnel and user behavior
- 01Pulls signup, interaction, and activation eventsNeonEvents grouped by user ID for the analysis window. SQL visible in the execution log.
- 02Builds a step-by-step conversion funnelDoeDoe computes conversion rates at each stage, identifies the steepest drop-off, and segments by channel, plan type, and device. Code inspectable and editable.
- 03Ranks behaviors that separate converters from non-convertersDoeDoe computes adoption rates for each feature among both groups and ranks differences by effect size. The biggest gaps surface first.
- 04Posts top findings to your channelSlackDoe delivers which funnel step has the biggest drop, which behaviors most differentiate converters, and how that varies by segment. Full data and analysis code attached.
- 05RecurringMonthly, or on demand after product changesDoe reruns the funnel analysis monthly. After a product change, trigger an on-demand rerun to see how conversion changed. Each run's queries and code are saved for comparison. Analysis posted to #product-data in Slack.
Up and running in under ten minutes.
- 01Connect your toolsOne-click OAuth for each integration. No API keys, no engineering.
- 02Describe what you need“Analyze our signup-to-activation funnel: landed on pricing, started trial, created first project, invited a teammate. Show where users drop off and compare conversion rates by referral source.”
- 03It runs on scheduleUpdated analysis lands in your team channel every month, with on-demand reruns after product changes.
Before you delegate.
- 01How is this different from Amplitude or Mixpanel?Doe runs against your raw event data with your definitions, so you're not limited by what your analytics tool tracks or how it segments. The analysis runs as code you can inspect, edit, and extend. And the behavioral comparison (what do converters do differently?) goes beyond what most product analytics dashboards offer out of the box.
- 02What does the behavioral comparison compute?For each tracked feature or event, Doe computes the adoption rate among converters and non-converters, then ranks the differences by effect size. Features where converters have a much higher adoption rate surface first. This helps you identify which behaviors matter most for conversion.
- 03Can I see the analysis code?Yes. Every SQL query and every line of Python analysis code is visible. You can edit the funnel steps, adjust the comparison logic, or add segments and rerun. Execution logs include query timing and row counts.
- 04Can I run this after a specific product change?Yes. Trigger a rerun with a date filter: "compare funnel conversion for users who signed up before vs. after March 1." Doe computes both funnels and shows the delta at each step.