campaign

Campaign — Outbound Engine by BuildYourSales.tech

You launch the prospecting pipeline and supervise execution.

Prerequisites

• client.yaml + at least one campaign.yaml with status: draft
• Leads available via one of 3 sources (see below)

Flow

1. Get leads

Ask the user how they want to provide leads. Support 3 sources:

Option A — Source from Lemlist (recommended) Use the Lemlist MCP tool lemleads_search to find leads matching the campaign's ICP criteria (persona, geo, company size, industry from campaign.yaml and ciblage.md).

• Present the search results to the user for validation before proceeding
• Once validated, save as leads.json in the campaign folder for the pipeline

Option B — Import CSV/Excel If the user provides a CSV or Excel file:

• Read the file and auto-detect column mapping (first name, last name, company, email, LinkedIn URL)
• Show the mapping to the user for confirmation
• Convert to leads.json format and save in the campaign folder

Option C — Manual JSON If leads.json already exists in the campaign folder, use it directly.

2. Verify leads

Read leads.json and check:

• Required fields: firstName, lastName, companyName
• Number of leads
• Flag leads without email (Lemlist will attempt findEmail)
• Flag leads without linkedinUrl (LinkedIn channel will be disabled)

2. Propose launch

Options:

• Single campaign: python -m pipeline run --campaign {path}
• Batch: python -m pipeline run --client {path} --campaigns all
• Selection: python -m pipeline run --client {path} --campaigns C01,C04
• Dry run (no injection): add --dry-run

3. Launch and monitor

Run the command via Bash. The pipeline displays progress on stderr. At the end, read report.json (or report_batch.json) and display a summary:

• Leads enriched / written / injected
• Any errors
• Estimated cost

4. Error handling

If a campaign fails:

• Read campaign.yaml → the status indicates where it stopped
• Read report.json → error details
• Propose a fix (re-run, fix leads, etc.)
• The run command automatically resumes at the right status

5. Lemlist injection

If the pipeline is in --dry-run mode or without a Lemlist ID:

1. Create the Lemlist campaign via MCP (create_campaign_with_sequence)
2. Retrieve the campaign ID
3. Run the injection: python -m pipeline inject --campaign {path} --lemlist-id {id}

Rules

• Always confirm before launching (show estimated cost + number of leads)
• NEVER launch without the user validating the leads
• Show estimated cost before every batch

Next step

Once campaigns are live, direct the user to: "Run /outbound-engine:monitor to track performance."