Campaigns

A campaign belongs to one Brand Space and carries more than a name and a budget. The parts you will meet in the interface are:

• Status — draft, active, paused, completed, or archived. Status drives what the detail page lets you do.
• Metadata — objectives, target audience, total budget and currency, start and end dates, and a priority of low, medium, high, or urgent.
• A flow — nodes and edges. A campaign is a graph, not a flat form, which is what makes multi-step campaigns and the workflow view possible.
• Workflow steps — ordered steps typed as content creation, approval, deployment, optimization, analysis, audience targeting, or budget allocation, each optionally bound to a platform such as Meta, Google, TikTok, LinkedIn, X, YouTube, Instagram, or Shopify.
• An approval workflow — whether approval is required, which stages exist, the current stage, and the full history of who approved what and when.
• A performance summary — impressions, clicks, conversions, spend, ROI, and engagement rate, kept on the campaign itself so the list view can rank without a separate report.
• Schedules, budgets, deployments, and creative assets — each stored against the campaign rather than being free-floating.

1. Go to Campaigns and click Create. The form asks for a campaign name, an objective (brand awareness, conversions and sales, customer engagement, or retention), a description, start and end dates, a target audience segment, and a total budget with a daily limit. The campaign is saved as a draft.
2. Open the campaign to reach its detail page. It has three tabs — Overview, Funnel, and Workflow — plus Edit, Pause or Activate, and a menu with Refresh, Archive, and Delete.
3. Build the mechanics on the Workflow tab, where the ordered steps live. This is also where deployment steps bind the campaign to the platforms you have connected.
4. If your brand requires sign-off, submit the campaign for approval. It then appears in the Approval Queue, where a reviewer approves it or rejects it with a written reason; the decision is recorded in the approval history.
5. Activate the campaign. Execution can run in test, preview, or production mode, and each run reports a status, a progress percentage, per-platform deployment states, and a log.
6. Watch it on Analytics and Performance under the campaign, and in the platform-wide Analytics and Attribution screens. Pause or archive from the detail page header when it is done.

The two are deliberately different things. A campaign is the marketing initiative: objective, budget, schedule, approvals, and the platforms it deploys to. A funnel is the customer journey that initiative uses — stages and steps a person moves through. A funnel points at the campaign it belongs to, which is why the campaign detail page carries a Funnel tab and links straight into the funnel view.

There is a third object worth naming so it does not surprise you: a campaign space, the asset library created when you import historical campaign data from an ad platform. Campaign spaces hold imported creative and cached platform metrics; they are a source of material for funnels, not a replacement for campaigns.

Optimization analyses a running campaign against goals you set and returns recommendations in four categories — budget allocation, audience targeting, creative optimization, and schedule adjustment. Each recommendation carries a description, an expected impact, a confidence score, an implementation effort, and a flag for whether it can be applied automatically; you can also ask for recommendations to be auto-applied. The standalone Optimize screen currently previews these categories rather than running against live campaign data, so treat it as an outline of the model, not a live report.

Recreate is a six-step wizard for rebuilding a competitor ad as your own campaign: Input (a competitor URL or screenshot), Analysis (the ad is broken down), Audience (choose target profiles), Customize (adapt the copy culturally), Preview, and Activate.

A/B testing lives on its own screen and works on experiments. An experiment has a name, a target metric, a traffic percentage, and two or more weighted variants; you start it, stop it, and select a winning variant. Two kinds run side by side: public landing-page experiments, which are tracked entirely through Google Analytics 4 events and read in GA4 rather than in En Garde, and in-product experiments on authenticated pages, whose metrics are stored and read back inside the platform. Only draft experiments can be edited.

There are two budget layers. At campaign level you set a total budget, a daily limit, and a currency when you create the campaign, and budgets can also be attached as records to an existing campaign.

Above that, agencies have budget allocation: funds move from an agency wallet to a client, and from a client to a specific brand, with an available balance you can check and spend recorded against the allocation it came from. That layer answers who the money belongs to; the campaign budget answers what it is being spent on.

Campaigns reach the outside world through deployments. A deployment step targets a platform you have already authorized in Integrations, and each deployment tracks its own state and the identifier the platform gave it, so a sync problem is visible per platform rather than as one opaque failure. Creative assets can be uploaded against the campaign and imported historical assets are available through campaign spaces.

Agent workflows sit alongside this rather than inside it. A workflow can create a campaign as one of its actions, and campaign workflow steps typed as approval or analysis are the natural hand-off points between automation and a human decision. Anything an agent flags instead of executing shows up in the agent inbox and, for agencies, in the approvals queue.

A single-brand owner has a segment of lapsed customers. They create a campaign with a retention objective, a six-week window, and a total budget with a daily cap, then set the target audience to that segment.

On the Workflow tab they add a content creation step, a deployment step bound to Meta, and an analysis step. Content Studio produces the creative; because the brand runs approvals, the campaign is submitted and approved before anything goes out. They activate it in test mode first, check the execution log and the platform deployment status, then run it for real.

Two weeks in, the campaign performance summary shows a healthy click rate but weak conversion. They set up an experiment with two creative variants at a 50/50 weight against a conversion target metric, let it run, and select the winner. The final numbers are read in Attribution rather than in the platform dashboards alone, because most of these customers touch more than one channel before they come back.

• Campaigns are scoped to a Brand Space. There is no cross-brand campaign; agencies compare brands from the agency dashboard instead.
• Deployment only reaches platforms you have connected in Integrations. A deployment step against a platform with no authorization will not run.
• Duplicating a campaign is explicit about what comes with it — schedules, budgets, approvals, creative assets, and audience segments are each opt-in, and performance data can be reset so the copy starts clean.
• Approvals are per campaign, not per asset. If nothing appears in the Approval Queue, the campaign has not been submitted or the brand does not require approval.
• Landing-page A/B test results are not visible inside En Garde. They are read in GA4.
• The performance summary on a campaign is a cached rollup. For the underlying journey and multi-touch credit, use Attribution.