Win-Back Campaigns with Amplitude

What Win-Back Campaigns Actually Require

Re-engaging churned users is not a messaging problem. It is a segmentation problem. Most win-back campaigns fail because they treat every lapsed user the same — same email, same offer, same timing. Amplitude gives you the data infrastructure to stop guessing and start targeting based on what users actually did before they left.

This guide walks through how to use Amplitude's behavioral analytics to identify, segment, and activate win-back campaigns with precision. You will need a downstream messaging tool (Braze, Iterable, or similar) to execute the campaigns themselves. Amplitude's role is to tell you who to target and why.

---

Understanding Where Amplitude Fits

Amplitude is a product analytics platform, not a campaign execution tool. That distinction matters for win-back work.

What Amplitude handles:

• Identifying which users churned and when
• Analyzing what behavior preceded churn
• Building behavioral cohorts for export
• Measuring whether re-engagement campaigns actually worked

What Amplitude does not handle:

• Sending emails, push notifications, or SMS
• Triggering real-time messages based on live behavior
• A/B testing campaign copy or creative

You build the intelligence in Amplitude. You execute the campaign in your messaging platform.

---

Step 1: Define Churn for Your Product

Before you can identify churned users, you need a working definition. "Churned" means different things across products.

In Amplitude, open Behavioral Cohorts and define your lapsed user segment using the Performed/Not Performed filter logic. A practical starting structure:

1. Users who performed a core action (your activation event — a purchase, a project created, a report run) at least once in the past 90 days
2. AND have not performed any session event in the last 30 days

Adjust the window based on your product's natural usage cadence. A daily-use tool like a task manager should use a 7-day inactivity threshold. A quarterly-use tool like a tax product might use 120 days.

Name this cohort clearly — something like `Lapsed - 30 Day No Session - Previously Activated`. Precision in naming saves confusion when you have multiple win-back cohorts running simultaneously.

---

Step 2: Segment by Pre-Churn Behavior

This is where Amplitude's analysis earns its place in your win-back workflow. Not all churned users are worth the same effort. And not all of them left for the same reason.

Use the User Paths Chart to Find Drop-Off Patterns

Open the User Paths chart and set the starting event as your core activation action. Look at what happened in the sessions immediately before users went silent. You will typically see two or three distinct patterns:

• Users who completed your core workflow and then stopped (satisfied exit or unmet need)
• Users who hit a friction point — a failed action, an error event, a dead-end path
• Users who came back once or twice after initial activation and then dropped completely

Each pattern points to a different re-engagement message and offer.

Build Tiered Cohorts Using Behavioral Cohorts

Create separate cohorts based on what you find in the path analysis:

• High-value lapsed: Users who completed 3+ core actions before churning. These users understood your product. They need a reason to return, not a tutorial.
• Friction-lapsed: Users who triggered an error event or support event before going silent. These users hit a wall. They need reassurance that the problem is fixed.
• Early lapsed: Users who activated but never returned after session one or two. These users never fully engaged. They may need onboarding-style content, not a win-back offer.

Use Amplitude's Behavioral Cohort builder to define each group with specific event conditions and frequency parameters. Each cohort becomes a separate export for your messaging platform.

---

Step 3: Analyze Churn Timing with the Retention Chart

The Retention Analysis chart in Amplitude shows you exactly where in the user lifecycle people stopped coming back. Run a retention analysis using your activation event as the starting point and any session event as the returning event.

Look for the steepest drop-off point. If 60% of users who activated never came back after day 3, that is your critical window. Win-back campaigns targeting users who lapsed after that point require a fundamentally different approach than campaigns targeting users who churned after 60 days of regular use.

This data also tells you how long to wait before triggering a win-back sequence. If your natural re-engagement window is 7 days, a win-back email on day 45 is noise. Time your outreach based on actual retention curve data, not intuition.

---

Step 4: Export Cohorts to Your Messaging Platform

Once your cohorts are defined, export them to your execution tool.

Amplitude has native integrations with:

• Braze — sync cohorts directly into Braze Segments, then use Canvas to build the win-back journey
• Iterable — push cohorts into Iterable Lists, then use Workflow Studio to trigger the campaign sequence
• Klaviyo, Customer.io, and others — available through Amplitude's Destination catalog

To configure a sync:

1. Navigate to the cohort in Amplitude
2. Click Sync and select your destination
3. Choose sync frequency — one-time export for a batch campaign, or daily sync if you want the cohort to update as new users lapse
4. Map the Amplitude user identifier to the matching identifier in your messaging tool

Daily sync is the right choice for ongoing win-back programs. One-time exports work for isolated campaigns.

---

Step 5: Measure Re-Engagement in Amplitude

Do not measure win-back success solely inside your email platform. Open rates and click rates tell you what your subject line did. Amplitude tells you whether those users actually came back and did something meaningful.

Create a Funnel Analysis in Amplitude:

1. Step 1: Re-engagement event (your first tracked touchpoint after the campaign send — a session start is sufficient)
2. Step 2: Core activation action
3. Step 3: Retention event at 7 days and 30 days post re-engagement

Segment this funnel by your original cohort groups. High-value lapsed users and early lapsed users will re-engage differently. That data shapes how you structure the next campaign.

---

Known Limitations

Amplitude is powerful for cohort analysis, but there are real gaps to account for:

• No real-time triggering. Amplitude cannot send a message the moment a user lapses. You are working with batch cohorts, not event-based triggers. For real-time win-back triggers, your CDP or messaging platform needs to own that logic.
• Identity resolution can create gaps. If your users are not consistently identified across web and mobile, cohort membership becomes unreliable. Resolve your identity strategy before building win-back cohorts at scale.
• Cohort sync latency. Daily syncs mean there is up to a 24-hour lag between a user meeting your lapse definition and them entering your campaign. For time-sensitive products, this matters.
• No built-in control group management. You will need to manage holdout groups inside your messaging platform, not in Amplitude.

---

Frequently Asked Questions

How do I identify the right churn threshold for my product?

Start with your median session gap. Pull a distribution of days between sessions for active users using the Event Segmentation chart. The 90th percentile of that distribution is a reasonable starting point for your inactivity threshold. A user who exceeds your typical active user's gap by that margin is a candidate for a lapsed cohort.

Can Amplitude trigger win-back messages automatically?

Not directly. Amplitude does not send messages. However, with a daily cohort sync to Braze or Iterable, you can build a near-automated workflow where users enter your win-back campaign within 24 hours of meeting the lapse criteria. True real-time triggering requires a tool like Braze's own behavioral triggers or a CDP with streaming capabilities.

What events should I make sure I am tracking before I build win-back cohorts?

At minimum: a session start event, your core activation event (the action that defines an engaged user), any error or failure events, and a clear subscription or purchase event if your product is transactional. Without these, your cohort definitions will be too blunt to be useful.

How do I know if my win-back campaign is actually working?

Compare retention curves between re-engaged users and a holdout group using the Retention Analysis chart. If re-engaged users show a steeper retention curve at 30 days than users who received no campaign, the campaign is working. Click rates inside your email tool are a vanity metric for win-back programs. Behavioral re-engagement in Amplitude is the signal that matters.