---
title: "Failures you can't see: confirm the coverage gap"
description: Compare critical product journeys with your tracking plan to find failures that analytics can't observe.
product: data
token_estimate: 1293
---
# Failures you can't see: confirm the coverage gap

> For AI agents: a documentation index is available at [/docs/llms.txt](/docs/llms.txt). Append `.md` to any page URL for markdown, or send `Accept: text/markdown`.

Amplitude should receive a failure event when a checkout, login, payment, or upload fails. Often, you track only the success event. A broken step then looks like a user who left. Conversion appears to drop, but no chart can identify the cause.

Missing failure signals can cause teams to spend weeks optimizing a funnel that measures a product bug. Data Assistant only sees the events your product sends, so it can't identify a missing failure event.

Run this audit on journeys where a failure costs revenue, access, activation, or trust. Use a Funnel Analysis chart for instrumented steps. Session Replay can provide supporting evidence of a visible failure, but a replay doesn't replace a structured failure event.

![Diagram of a two-step funnel with a large drop between checkout started and payment completed, and a callout noting that no failure event exists to explain the drop](https://amplitude.com/docs/images/diagrams/invisible-funnel-failure.svg) ![Diagram of a two-step funnel with a large drop between checkout started and payment completed, and a callout noting that no failure event exists to explain the drop](https://amplitude.com/docs/images/diagrams/invisible-funnel-failure.dark.svg)

## Audit a critical journey

1. List the journeys where a failure affects revenue, access, activation, or trust.
2. For each journey, name the successful outcome and every distinct failure outcome that requires a different response.
3. In Amplitude Data, go to _Data > Events_ and search for the matching events.
4. Review each event's description and properties. Confirm that the definition separates a completed action from a failed attempt.
5. Build a Funnel Analysis chart for the expected path and look for unexplained drop-off before the success event.
6. Reproduce a safe test failure and use [Event Explorer](https://amplitude.com/docs/analytics/charts/event-explorer) to check which events arrive.
7. Record the missing events and missing diagnostic properties in the tracking plan before implementation.

> **Warning:** Drop-off doesn't prove a technical failure
>
> Users abandon flows without encountering an error. Confirm a missing failure signal with product behavior, support evidence, application logs, or a controlled test before you request instrumentation.

## What Amplitude can and can't do here

Amplitude can't count a failure event that your product never sends, and no chart can reconstruct the event later. Amplitude can help you locate and characterize the instrumentation gap:

- Use [Session Replay](https://amplitude.com/docs/session-replay/overview) to watch real sessions and check whether users encounter a visible error or leave without one. Every plan includes Session Replay with plan-specific monthly limits. Growth and Enterprise customers can purchase additional capacity and capabilities. Refer to [Pricing](https://amplitude.com/docs/pricing) for current details.
- After the failure event exists, an alert on the event volume warns you when failures spike. Availability of anomaly detection and alerts varies by plan. Refer to [Pricing](https://amplitude.com/docs/pricing) or contact [Amplitude Sales](https://amplitude.com/sales-contact).

> **Tip:** Ask Global Agent
>
> [Global Agent](https://amplitude.com/docs/amplitude-ai/global-agent-overview) can build the funnel and identify the drop from plain-language instructions. Try this prompt:
>
> "Build a funnel from Checkout Started to Payment Completed over the last 8 weeks and show me the step with the biggest drop-off."
>
> Review the generated funnel and supporting data to confirm where conversion falls. Global Agent can't add a failure event that your product never sends.

## Send the instrumentation gap to the tracking owner

Adding a failure event requires a code change. Send your engineers the confirmed instrumentation gap. Copy and complete this template:

```text
Our [FLOW NAME] flow has no failure event, so we can't see why users drop between
[UPSTREAM EVENT] and [SUCCESS EVENT]. Add an event that fires when the
[FLOW NAME] step fails, with safe diagnostic properties (for example flow, step,
error_type, error_code). No stack traces, tokens, payment details, or emails.
Add it to the tracking plan first. This is hiding real drop-off in our reporting.
```

## If you have no engineering resources

You can't create a missing failure signal inside Amplitude. Contact the following resources in order:

- Your Amplitude account team or Customer Success Manager.
- A partner or agency that manages your website or app analytics.
- Amplitude Support at [gethelp.amplitude.com](https://gethelp.amplitude.com) for product questions.

Until the failure event exists, treat the affected funnel as incomplete. Use Session Replay to understand individual cases.

To instrument the missing signals, continue with the [engineering solution](https://amplitude.com/docs/data/data-assistant-user-solutions/error-event-coverage-engineering).