--- title: PaperTube ABM Campaign Automation type: article created: '2026-04-05' updated: '2026-04-05' source_docs: - raw/2026-02-06-weekly-call-w-120526026.md tags: - abm - papertube - salesforce - account-engagement - pardot - ai-agents - automation - api - mcp layer: 2 client_source: null industry_context: null transferable: true --- # PaperTube ABM Campaign Automation ## Overview PaperTube is executing an Account-Based Marketing (ABM) strategy targeting ~256 companies sourced from ZoomInfo and loaded into Salesforce. The core challenge: ABM requires personalized, company-specific outreach — not bulk email blasts — making manual campaign creation at this scale impractical. The solution is a two-stage AI automation pipeline: (1) generate deep account research reports per company, then (2) use those reports to produce personalized campaign sequences, with campaigns created programmatically via the Salesforce API. This work is tracked under [[clients/papertube/index]]. --- ## Stage 1: AI-Generated Account Research ### What It Does An AI agent (running locally with ~100 MCP servers providing web scraping and document tooling) processes a deduplicated list of target companies and produces a structured research report for each one. **Input:** A Google Sheet of ~350 rows (with duplicates) → deduplicated to **256 unique companies**, each with company name and website URL. **Output:** One Markdown (`.md`) file per company, saved to a local `PaperTube Research/` folder. ### Report Contents Each report covers: - Company overview (HQ, website, industry) - Recent product launches, rebrands, or product changes - Sustainability commitments and packaging strategy - Distribution strategy and growth signals - Decision-making unit: key contacts, roles, and recommended engagement order ### Why Markdown Reports are generated as `.md` files because: - AI models (Claude, etc.) can read and process Markdown faster than `.docx` or `.pdf` - Clean import into Google Docs via **Edit → Paste from Markdown** produces fully formatted output (headers, tables, checkmarks) with no manual cleanup - Files are portable and reusable as inputs to downstream automation steps ### Runtime Each report takes approximately 2 minutes to generate. The agent runs sequentially through all 256 companies, posting progress updates as it goes. Total runtime is several hours for the full batch. --- ## Stage 2: Personalized Campaign Sequence Generation Once a research report exists for a company, it is fed to Claude with a prompt instructing it to generate a full ABM campaign document for that target. The output includes: - **Strategy summary:** primary and secondary contacts to target, who to avoid initially, and rationale - **Sequence overview:** 4–5 touches across email and LinkedIn, with specific send-day cadence - **Guardrails:** e.g., never send two emails in the same week to the same person; pause and go conversational if a contact engages - **Individual message drafts:** subject lines and full body copy for each email; connection note text for LinkedIn touches Emails are plain-text (no HTML template), which simplifies campaign creation and improves deliverability for cold outreach. --- ## Stage 3: Automated Campaign Creation via Salesforce API ### The Problem Creating 256 individual campaigns manually in Salesforce Account Engagement (Pardot) is prohibitively slow. Each campaign requires: creating the campaign record, creating a folder, and adding the associated emails/list sends. ### The Solution Use the **Salesforce REST API** to programmatically create campaigns. When Connected Campaigns is enabled (the standard Salesforce configuration), campaigns are created through the Salesforce API and sync automatically to Account Engagement. ### Connected App Setup (Completed) A Connected App named **"Asymmetric API Integration"** was created in Salesforce Setup: | Field | Value | |---|---| | App Name | Asymmetric API Integration | | Contact Email | mark.hope@asymmetric.pro | | Callback URL | https://auto.asymmetric.pro | | OAuth Scopes | `managed_pardot_services`, `refresh_token` | Credentials retrieved: - **Consumer Key** (active) - **Consumer Secret** (active) - **Staged Consumer Key / Secret** (generated and applied) - **Business Unit ID** (from Marketing Setup → Pardot Account Setup) - **Security Token** (generated from user profile settings) ### Current Blocker Mark's Salesforce user account requires **SSO enablement for Account Engagement** before the OAuth connection can be completed. This is a Salesforce admin action. **Owner:** Karly to request SSO enablement from the Salesforce admin. --- ## Data Enrichment Requirements For the automation to generate high-quality personalized sequences, contact records in Salesforce need LinkedIn profile URLs. ZoomInfo data was imported with up to 3 contacts per company, but LinkedIn URLs may not have been included in the initial import due to a missing field. **Action items:** - Add a LinkedIn Profile field to Salesforce Contacts (may already exist as a standard field) - Re-import ZoomInfo data with LinkedIn URLs mapped to that field --- ## Related Work - [[clients/papertube/index]] — client overview - [[clients/papertube/domain-email-setup]] — `papertube.pro` DNS and email sending configuration - [[knowledge/outbound-sales/salesforce-api-connected-app-setup]] — reusable notes on Salesforce Connected App and OAuth credential retrieval - [[knowledge/tracking/automated-website-tracking-audits]] — parallel automation work discussed in the same session --- ## Open Action Items - [ ] Karly: Request Salesforce admin enable SSO for Account Engagement on Mark's user account - [ ] Mark: Complete Salesforce OAuth connection once SSO is enabled - [ ] Mark: Build and test campaign creation automation via Salesforce API - [ ] Karly: Add LinkedIn Profile field to Salesforce Contacts; import ZoomInfo LinkedIn URLs - [ ] Mark: Schedule call with Parag to discuss website/SEO services expansion