このページの内容

Opportunities: anatomy and lifecycle (Early Access)

Early Access

This feature is in Early Access. During this time, aspects of the functionality may still be developed, and this documentation may not always be up to date. If you have any questions, contact Amplitude Support.

An Opportunity is a single, evidence-backed unit of product work: a problem to fix or an idea to pursue within a Product Area. Amplitude writes each Opportunity so a person or a coding agent can pick it up and act on it without redoing the analysis.

This page covers what an Opportunity contains and the lifecycle it moves through.

Anatomy of an Opportunity

Every Opportunity is more than a title. It bundles the analysis, the evidence, and a draft plan into one reviewable object.

Problem and solution

  • Product context: A plain-language summary of the current state and why this matters.
  • Solution plan: The proposed change, written so a person or coding agent can act on it.

Evidence

Each Opportunity links to the specific data that supports it through citations. A citation names the source (for example chart:abc123 or dashboard:uzqxz98z), a short summary of the finding, and the method behind it. Evidence is what grounds an Opportunity: every quantitative claim traces back to the chart, dashboard, session replay, or feedback theme it came from.

The system corroborates findings across multiple independent signals before promoting them. Findings backed only by a small number of session replays stay directional rather than becoming high-priority items.

Priority score (RICE)

Amplitude prioritizes Opportunities with a RICE score so the highest-return work rises to the top:

  • Reach: How many users or sessions the change affects.
  • Impact: The expected magnitude of improvement.
  • Confidence: How strong the supporting evidence is.
  • Effort: The implementation cost.

The score is (Reach × Impact × Confidence) / Effort. A higher score means a better return on effort.

Expected impact

Each Opportunity estimates its expected impact on your Product Area's metrics, for example "+3.5 percentage points on checkout completion." Impact lines reference the target metric and state a magnitude and unit so you can sanity-check the bet.

Specification confidence

A specification confidence of high, medium, or low signals how ready the Opportunity is to hand off. High-confidence specs are ready for a coding agent; lower-confidence ones may need a person to refine the plan first.

Acceptance criteria

Optional acceptance criteria describe how a reviewer functionally verifies the outcome: the user-visible behavior or end-to-end scenario that proves the work is done. These keep PMs and coding agents aligned on "done."

Execution plan

When you connect a repository to the Product Area, an Opportunity can carry an execution plan that names the target repository and grounds the change in your actual code, so a coding agent can draft a pull request.

Tags

Amplitude tags Opportunities across several families so you can filter and route them:

  • Sizing: strategic, feature, bug, quick win, or wildcard.
  • Execution method: code, guide, or hybrid.
  • Location: frontend, backend, or mixed.
  • Human dev required, agent autonomous, and feedback needed: how much human involvement a change needs.

The Opportunity lifecycle

Each Opportunity moves through a status lifecycle, from discovery to measured outcome. You and your team move opportunities forward; agents can also update status as they work.

Opportunity status lifecycle from New through Measured, with Dismissed as an optional branch and feedback improving future discovery Opportunity status lifecycle from New through Measured, with Dismissed as an optional branch and feedback improving future discovery

Review and give feedback

After you generate your first batch, review each Opportunity against three simple questions:

  1. Is the analysis or signal correct?
  2. Is the Opportunity a reasonable problem?
  3. Is the suggested plan useful?

Use the thumbs up and thumbs down controls when reviewing. Every labeled input gives Amplitude context on your team's preferences, so the next batch can improve automatically. Direct comments, dismissals, edits, and status changes also feed back into the system.

If an Opportunity's analysis or recommendation isn't exactly right, change the plan directly or use Global Agent to run follow-up analysis before acting.

Opportunity feedback panel with positive, neutral, negative, and comment controls

Next steps

Was this helpful?