Unity Analytics and session replay: find where players leave and why

A missing completion event tells you to look closer. It doesn't explain why the player stopped or whether they'll return. A replay can show a blocked interaction, but you'll still need to test your explanation.

Of 800 players who started the attack step, 480 completed it: a 60% step completion rate. The other 320 represent a 40% drop-off between those two steps in this example. The overall tutorial completion rate is 45%; it answers a different question.

Before changing the tutorial, verify that the completion event fires reliably. Check event order, duplicate emissions, build versions, and whether players had enough time to finish. A reporting error can look like a design problem.

Send matching events to both SDKs

Create the custom event definitions in Unity Analytics first. Initialize both SDKs and get any required player consent. Then call these methods when the attack step starts and when the game confirms it is complete:

using Playtrace;
using Unity.Services.Analytics;

public static class TutorialEvents
{
    public static void AttackStarted()
    {
        AnalyticsService.Instance.RecordEvent("tutorial_attack_started");
        PlaytraceEvents.Emit("tutorial_attack_started");
    }

    public static void AttackCompleted()
    {
        AnalyticsService.Instance.RecordEvent("tutorial_attack_completed");
        PlaytraceEvents.Emit("tutorial_attack_completed");
    }
}

This example uses Unity Analytics SDK 5.1 or later. The other funnel steps follow the same pattern. Keep Playtrace event names exact and case-sensitive. There is no automatic import of Unity Analytics events into Playtrace, and matching names do not automatically link users or sessions between the services.

For setup details, follow Unity's event recording documentation and the Playtrace event guide.

Create the ordered funnel

In Unity Analytics, open Funnels and add the five events in the order shown above. Use a consistent date range and build version; separate platforms where their controls differ. Compare the attack step's completion rate and time to completion. See Unity's funnel tutorial for the dashboard workflow.

Adjust the one-minute window to the time players normally need. If it's too short, slower players may be counted as drop-offs before they finish. Start by reviewing ten replays for recurring problems; use analytics to measure how common those problems are.

Understand which recordings are preserved

• Completion within the window: the current SDK stops and discards the recording. The attempt does not qualify for this drop-off campaign and is not preserved as a processed replay.
• No completion by timeout: the recording qualifies for preservation. After successful upload and processing, it appears in the campaign's replay list.
• Backgrounding: enabling “count backgrounding as drop-off” makes the built-in game_paused event stop the recording early. It is preserved only if the expected completion has not happened. With the setting off, backgrounding is not that early-stop trigger. Neither setting proves permanent churn or guarantees footage after a forced app termination.

Activate the campaign and test both a successful and an incomplete attempt before recording real players. Make sure the campaign has loaded before sending its start event. The setup checklist covers platform matching, event names, and replay availability.

Example finding: the prompt doesn't explain the required action

Suppose several clips show players tapping an attack control. The character swings, but the tutorial does not advance. The prompt says “Attack the enemy,” while the completion condition requires holding the control for a charged attack.

Observation: the character attacks, but the tutorial does not advance.

Hypothesis: players do not understand the charged-attack requirement.

Check the game logic and event emission too: the same footage could reveal a progression bug or a missing analytics event.

Note the replay timestamp, what the player did, and your explanation. Look for recurring patterns, then check the analytics to see how common the problem is across players.

1. Define the outcome: attack-step completion, measured as completions divided by starts. Track completion time and overall tutorial completion alongside it.
2. Compare fairly: run an A/B test if possible. Otherwise, compare similar groups playing each build: use the same platforms, similar player acquisition sources, and the same amount of time to finish.
3. Review fresh clips: check whether the original interaction problem remains and whether the change introduced new friction.
4. Use both tools: check the completion rate in analytics, then watch new clips to see where players still struggle.

More tutorial completions don't necessarily mean more players return the next day. Measure retention separately. If you're comparing results before and after a change, remember that a different mix of players can also affect the numbers.