Skip to main content

Growth Engine

Understand how Growth Engine organizes marketing work around business goals, action controls, and ongoing learning.

K
Written by Kassandra Villa Arroyo

TL;DR

Growth Engine is Triple Whale's AI-powered marketing system for turning business goals into supported marketing work. AI agents analyze connected data and business context, prepare recommendations and assets, and carry out permitted changes. Campaign modes, action permissions and guardrails determine what can run automatically and what needs review.

Overview


Growth Engine helps brands and agencies coordinate recurring marketing work across media buying, creative, conversion and retention. It is designed for teams that need to connect daily decisions with business goals and understand why changes are proposed, what ran and what happened afterward.

A goal provides direction, jobs organize the work, and tickets retain the reasoning and discussion behind recommendations. The CMO is the coordinating AI agent and the main conversational interface. Customers provide strategy, context, permissions and feedback without needing to orchestrate individual specialist agents.

Growth Engine organizes ongoing work around goals and schedules. Moby chat supports open-ended questions and requests; a separate Moby chat conversation is not a prerequisite. Set up your first Growth Engine job puts this workflow into practice.

Watch the walkthrough


Key terms


Term

Meaning in Growth Engine

Goal

The business outcome the workspace works toward, including its target, timeframe, and relevant constraints.

Job

An area of work the agents handle, such as testing ads or analyzing creative. A job can involve several supported actions.

Ticket

A work item that brings together an opportunity or problem, a hypothesis, investigation findings, proposed actions, and follow-up.

Action

A specific supported operation in a connected platform, such as changing a budget or an ad's status.

Operating mode

The Control, Copilot, or Autopilot setting that governs how the selected scope participates in automated work and approval.

Guardrail

A configured limit on supported changes, such as a budget-change cap or a cooldown between actions.

How it works


Goal and context → Investigation → Proposed actions → Review or permitted execution → Follow-up and learnings

A goal gives the work direction

A useful brief describes the overall marketing outcome, timeframe, priorities and constraints, then explains each channel's role. Prospecting and retargeting may have different success criteria, as may branded and non-branded search. These differences help agents evaluate channel work in the context of the business goal.

The saved goal interpretation needs to match that intent. Where available, How your goal is measured explains the saved interpretation. Measurement depends on the target, dates, included channels, metric definition and attribution settings.

Brand guidance, approved assets, promotions, margins and inventory constraints help agents interpret opportunities. Goals and context need updating when strategy changes or an offer ends. Instructions guide the agents but do not replace configured software limits.

Jobs organize recurring work

Jobs organize work such as testing ads, managing campaigns, developing creative, improving conversion and bringing customers back. The Growth Engine jobs reference explains the available work areas and their inputs. Availability depends on the connected tools and account.

Schedules determine when enabled work runs. Scheduled work does not imply continuous monitoring or instant execution. Example schedules are not universal defaults or guarantees of when a first recommendation will appear.

Tickets connect recommendations with their reasoning

The Board groups tickets by work stage. Each ticket can contain a hypothesis, findings, and several proposed actions. Its estimated value describes an expectation to evaluate, not revenue already earned.

Ticket conversations and rejection feedback provide context for revisions. Useful feedback explains the business reason, such as an expired offer or work already completed manually. Feedback is not a promise of perfect learning or a claim about foundation-model training.

Modes determine how supported changes proceed

Mode

How it works

What remains the team's responsibility

Control

The selected scope is outside that automated action workflow, although it may still appear in monitoring or comparisons.

Managing that scope separately. Control does not disable unrelated integrations or automations.

Copilot

Growth Engine prepares supported changes and waits for the required approval.

Reviewing the exact targets, proposed values, assets, and budget impact before approval.

Autopilot

Supported, enabled changes can run without individual review, within configured permissions and guardrails.

Setting appropriate scope and limits, then monitoring execution and results.

Campaign modes govern where Growth Engine can act. Action permissions govern which changes may run automatically. A campaign in Autopilot can still require review for an action, such as creating an ad. Shared action settings can affect more than one job. Platform permissions and supported actions also apply.

Warning: Enabling a job is not necessarily a review-only preview. Store promotions always require approval, including when an action catalog label shows Autopilot. A budget cap should not be assumed to exist, and some changes cannot be fully reversed.

Generation permissions for images, videos and landing-page drafts are separate from permission to publish or launch. A generated asset is not a live ad or page. The Growth Engine modes and guardrails reference explains these controls.

Execution and follow-up are separate outcomes

Approval, successful execution and measured results are different events. A ticket can have mixed action results, and a preview does not establish publication. The guide to reviewing proposed work explains the checks before and after approval.

Follow-up evaluates what happened after execution. The Learnings tab stores observations for later work, not a guaranteed per-action incremental revenue or profit calculation. The ticket's actual follow-up information determines the applicable timing.

When to use Growth Engine and trade-offs


When it is a good fit

Growth Engine is useful when recurring marketing work needs shared direction, documented reasoning and follow-up. Copilot provides a review point while a team evaluates recommendation quality. Autopilot requires appropriate scope, configured limits and ongoing monitoring.

Limits to account for

  • Support varies by workflow. A connected platform does not imply support for every campaign type, ad format, or publishing operation. The Moby Actions library describes general capabilities, not guaranteed availability in every Growth Engine job.

  • Publication and reversal have limits. Draft-only behavior and edit/unpublish support should not be assumed. Spend already incurred, content already seen and some deleted items cannot simply be undone.

  • Performance needs context. Pacing in Performance is meaningful only against the intended goal and measurement settings. Return on ad spend (ROAS) is attributed revenue divided by ad spend, not profit. A higher ROAS for one operating-mode group does not prove that the mode caused it; campaign mix and time in each mode can affect the comparison.

  • Access is account-specific. A visible navigation item does not establish activated access. The account team can confirm availability, trial dates, usage allowances, and continuation terms for the intended store.

How it relates to other Moby features


Moby Actions are supported operations in connected platforms. Moby Automations are scheduled workflows for recurring work. Growth Engine is the goal-led marketing workspace that organizes jobs, tickets, and follow-up.

If you use Growth Engine alongside other workflows that can change the same account, review their targets and permissions to avoid overlapping or conflicting changes.

Did this answer your question?