Trial-to-Paid Conversion with Amplitude

Why Trial-to-Paid Conversion Fails (And What Amplitude Shows You)

Most trial users don't convert because they never reach the moment where your product feels irreplaceable. They sign up, poke around, and leave before experiencing the core value. Amplitude's job is to show you exactly where that breakdown happens — and which behaviors predict the users who do convert.

This guide walks you through a concrete implementation using Amplitude's Behavioral Cohort Analysis and Journey Maps to identify, segment, and act on the signals that separate converters from churners.

---

Step 1: Define Your North Star Conversion Event

Before you build a single chart, you need a clean definition of what "converted" means in Amplitude.

Create a custom event called `subscription_started` or `plan_upgraded` — whatever maps to the moment a trial user becomes a paying customer. This should fire server-side so it's not blockable by browser extensions or client failures.

Then define your trial population:

• User property: `account_type = trial` or `plan = freemium`
• Entry event: `trial_started` with a timestamp
• Conversion window: Match your actual trial length (7 days, 14 days, 30 days)

This gives you the two populations you'll spend all your time comparing: users who converted and users who didn't.

---

Step 2: Build the Conversion Funnel in Amplitude Funnels

Use Amplitude Funnels to map the critical path from trial start to paid conversion.

Set Up the Core Funnel

1. Open Funnels and select your trial cohort as the starting filter
2. Add these events in sequence:

- `trial_started`

- `core_feature_used` (your product's primary value action)

- `upgrade_page_viewed`

- `subscription_started`

1. Set the conversion window to match your trial period
2. Segment by acquisition source, plan type, and device type

What You're Looking For

The funnel will expose your biggest drop-off point. In most SaaS products, the heaviest drop happens between `trial_started` and `core_feature_used`. Users sign up and never activate. That's a different fix than drop-off between `upgrade_page_viewed` and `subscription_started`, which is a pricing or friction problem.

Amplitude lets you click into any funnel step and pull the raw user list who dropped off. Export that list and you have your first re-engagement cohort.

---

Step 3: Run a Behavioral Cohort Analysis to Find the Activation Signal

This is where Amplitude earns its place in your stack. Behavioral Cohort Analysis lets you group users by what they actually did, then compare conversion rates across those groups.

Build Your Cohorts

In the Cohorts section, create two cohorts:

• Converted: Users who performed `subscription_started` within the trial window
• Not Converted: Users who completed the trial window without `subscription_started`

Now run an Event Segmentation chart comparing these two cohorts across every significant product action. You're looking for events where the converted cohort has a dramatically higher frequency or rate of completion.

The Activation Metric Framework

Common patterns you'll find:

• Converted users completed `report_exported` at 3x the rate of non-converters
• Converted users invited at least one team member within the first 48 hours
• Converted users who used a specific feature within day 1-3 converted at 60%+ vs. 12% for those who didn't

Once you find 2-3 behaviors that strongly correlate with conversion, you have your activation metrics. These are not vanity metrics — they're the specific actions you need to drive more trial users toward.

---

Step 4: Map the Journey Using Amplitude Journeys

Amplitude Journeys (formerly Pathfinder) shows you the actual sequence of events users take, not the sequence you designed for them.

Run Two Journey Maps

1. Converted users: Start from `trial_started`, end at `subscription_started`. Look at the top 5 paths.
2. Non-converted users: Start from `trial_started`, end at `trial_expired`. Look at the top 5 paths.

The contrast is usually stark. Converted users move toward your core feature quickly. Non-converters wander through settings pages, read help docs, and stall.

Use this data to redesign your onboarding sequence. If converted users hit `project_created` → `collaborator_invited` → `output_shared` within 72 hours, those three steps should become the backbone of your trial onboarding flow.

---

Step 5: Create Persistent Cohorts for Downstream Activation

Once you've identified your activation behaviors, build persistent behavioral cohorts that update automatically. These cohorts become the audience segments you push to your messaging tools.

Build These Four Cohorts

• Activated Trials: Users who completed your activation sequence but haven't converted yet — highest-intent segment
• Stalled Trials: Users who signed up 3+ days ago, haven't hit the activation event, and haven't converted
• At-Risk Trials: Users in the last 48 hours of their trial with no activation event
• Converted: Users who completed `subscription_started` — use this to study post-conversion behavior

Syncing Cohorts to Your Messaging Tools

Amplitude integrates directly with tools like Braze, Iterable, HubSpot, and Intercom. Use Amplitude's Cohort Sync to push these segments automatically. Your "Stalled Trials" cohort flows into Braze or Intercom, where you trigger a targeted onboarding sequence pointing directly at the activation action they haven't taken yet.

This closes the loop between your analytics and your activation campaigns without manual CSV exports.

---

Step 6: Measure Impact With Amplitude Experiments

If you're testing onboarding changes, paywall copy, or pricing page layouts, use Amplitude Experiment to measure conversion lift with statistical rigor. Set `subscription_started` as your primary metric and your activation event as a secondary metric. This tells you whether a test improved conversion directly or simply drove feature usage without revenue impact.

---

Limitations of Amplitude for This Use Case

Amplitude is a strong analytical layer, but it has real gaps you should plan around.

• No native messaging. Amplitude identifies the cohort; it does not send the email or the in-app message. You need a separate activation tool (Braze, Intercom, Customer.io) connected via Cohort Sync.
• Attribution complexity. If your trial users come from multiple acquisition channels with different trial lengths, keeping your cohorts clean requires disciplined event taxonomy upfront. Sloppy instrumentation produces misleading funnel data.
• Qualitative blind spots. Amplitude shows you what users did, not why they stopped. Pair your cohort analysis with session replay tools like FullStory or Hotjar to understand the friction behind the drop-off numbers.
• Retroactive analysis limits. If you haven't been tracking the right events, you can't backfill behavioral data. Instrumentation quality determines analysis quality. There's no shortcut.

---

Frequently Asked Questions

How many events do I need to be tracking before Amplitude analysis becomes useful?

You need at minimum: a trial start event, your core feature events (5-10 key actions), an upgrade page view event, and a conversion event. That's enough to build meaningful funnels and cohorts. The depth of insight grows as your event coverage grows, but you don't need 200 events to start finding activation signals.

What's the difference between using Amplitude Funnels and Journeys for this analysis?

Funnels measure completion rates along a predefined path you specify. Journeys show you the actual paths users took without you dictating the sequence. Use Funnels to quantify where users drop off on your intended conversion path. Use Journeys to discover what users are actually doing instead — which often reveals activation behaviors you hadn't considered.

How do I know if a behavioral correlation in my cohort analysis is meaningful or just noise?

Look for behaviors where the cohort difference is large (10+ percentage points) and where the event volume is sufficient (at least 100+ users in each cohort performing the action). Run the analysis across multiple time periods to confirm the pattern holds. If "exported a report" shows a 4x conversion lift across three separate monthly cohorts, that's a reliable activation signal. A one-time spike in a small sample is not.

Can Amplitude tell me why users aren't converting, or just that they aren't?

Amplitude tells you what happened behaviorally — which actions were taken, which weren't, and in what sequence. It does not capture user intent or friction perception. To understand why, you need to layer in qualitative methods: exit surveys, user interviews, or session recordings. Amplitude narrows where to look; those other methods explain what you find there.