HG Insights|Phoenix

ABM Campaign Builder

From ICP criteria to personalized outreach in one conversation — build, validate, segment, and draft with HG data at every step.

ABM Campaign · SAP in Germany · Enterprise $1B+

3 segments, 6 contacts, 6 personalized emails — from one ICP prompt

50
Accounts queried from HG
3
Pain-point segments
6
Personalized emails
Step Output Highlight
1. Build list 50 accounts via hg_data_query SAP + Azure + $1B+ rev
2. Segment 3 pain-point segments Prospect-centric names
3. Contacts 6 decision-makers enriched Per-segment personas
4. Emails 6 personalized drafts Humanizer-checked
Sample email opening
“Peter — Siemens unwound $1B+/year in Atos IT services and the HCLTech replacement covers a fraction of that scope. With the managed cloud contract at 50% renewal odds, you’re building the next-gen vendor stack right now.”

Overview

End-to-end ABM campaign workflow: build targeted account lists from HG data, validate and segment by pain points, find decision-maker contacts, and draft personalized outreach emails — all in one conversation with human review at each step.

Use cases

  • Full-funnel ABM in one conversation

    Start with 'companies using SAP in Germany with $1B+ revenue' and end with personalized emails to named decision-makers — with human review at every step. The workflow chains HG's data warehouse queries, pain-point segmentation, contact enrichment, and email drafting into a single guided conversation.

  • Pain-point segments your AEs can actually use

    The workflow infers prospect-centric segments from data patterns — 'The 5+ monitoring stack with no correlated alerting' instead of 'Cross-sell opportunity.' Each segment suggests who feels the pain most, so the contact search and email drafting are already persona-targeted.

View workflow prompt
# ABM Campaign Builder

## Parameters

- `{{criteria}}` *(required)* — Describe your ideal customer profile. Example: `Enterprise companies using SAP in Germany with $1B+ revenue`
- `{{pain_points}}` *(optional)* — Optional pain points for segmentation — if omitted, inferred from data patterns. Example: `Legacy ERP migration, cloud cost overruns, security tool sprawl`
- `{{target_persona}}` *(optional)* — Target contact persona — if omitted, inferred from pain-point segments. Example: `VP of IT`
- `{{value_prop}}` *(optional)* — Your value proposition for email personalization. Example: `We reduce cloud operations cost by 40% with AI-augmented managed services`

## Purpose
Guide the user through a 4-step ABM campaign: build a target account list from {{criteria}}, validate and segment by pain points, find decision-maker contacts, and draft personalized outreach emails. Pause after each step for human review.

## Process
1. **Build list** — Translate {{criteria}} into ClickHouse SQL via `hg_data_query`. Call `hg_catalog` first, then `list_product_categories`/`get_vendor_information` to verify names. Safe patterns: CTE + NOT IN, GROUP BY + HAVING (order by aggregate e.g. `max(intensity) DESC`). Always `LIMIT 100`. Join `install_global` to `company_locations` on `url_id` with `is_ghq = true`. For momentum: `install_intensity_momentum_global`, order by `intensity_momentum`. Present table, wait for approval.
2. **Validate + prioritize + segment** — Present list. Ask user: approve, modify, or replace. Iterate until approved. Then:
   - **Prioritize:** Score each account (intensity weight 4, IT spend 3, tech fit 3, revenue 2, industry 2). Rank A/B/C tier. Present ranked list for approval.
   - **Segment:** If `{{pain_points}}` provided, classify accounts against them. If not, infer: run `company_technographic` + `company_spend` on top 10–15 accounts, detect repeating patterns (sprawl, capability gaps, declining momentum, spend anomalies). Name each segment by the prospect's situation (never seller's motion). Assign severity (severe/moderate/mild) and recommend a target persona per segment. Present segments with tier distribution for approval.
3. **Find contacts** — Persona sourcing (in priority order): (1) `{{target_persona}}` if the user provided one, (2) the per-segment persona recommendations from Step 2 — each pain-point segment suggests who feels the pain most (e.g. "Observability sprawl" targets VP IT Ops), so different segments may target different personas, (3) if neither, ask the user. User picks top-priority accounts. Process 2–3 per batch: one `contact_search` per account (all title variations combined), then `contact_enrich` for best matches. Present each batch for approval.
4. **Draft emails** — Per contact: use Steps 1–2 research + segment + severity context. Ask for `{{value_prop}}`/tone/CTA if not provided. Draft: subject + body (80–120 words) + personalization hooks. **Apply the humanizer skill to every draft** — use contractions, start mid-thought, one idea per paragraph. No AI vocabulary (leverage, synergy, holistic, robust). No filler openings. No em dash cascades. No rule-of-three lists. Active voice only.

## Output Format
- **Step 1:** CSV table — Company, Domain, Industry, Country, Revenue, Employees, IT Spend, Key Tech Detected, Intensity/Momentum Signal
- **Step 2:** CSV table with ICP Tier (A/B/C), Segment, Severity columns + segment descriptions with recommended personas
- **Step 3:** CSV table — Company, Domain, Contact Name, Title, Email, LinkedIn, Seniority, Segment
- **Step 4:** JSON-style cards per contact — `{ company, contact, subject, body, personalizationHooks }`