A copy-paste Claude prompt that takes your Account Intent Scoring output, segments it into a 3-tier × 3-role grid, and produces 9 LinkedIn matched audience CSVs — each pre-formatted for upload, with audience sizing, frequency caps, budget allocation, and creative direction tuned to where each segment sits in the buyer journey. The production half of Track 03's diagnose-and-build loop.
A B2B SaaS team uploads 500 ABM target accounts to LinkedIn as a single matched audience and runs a single creative pointing at the demo CTA. Nothing converts. The reason: a Tier-1 enterprise CFO who's been in deep evaluation for 6 months and a Tier-3 mid-market user who's never heard of you both see the same ad. The CFO finds the demo CTA tone-deaf. The user finds the value prop overclaimed. Both bounce. The audience structure was wrong before the creative ever loaded.
The deeper problem is that a single matched audience erases all the diagnostic work. Track 03's Account Intent Scoring Engine identified which accounts are at what intent level. The Role-by-Role Engagement Map identified which roles within those accounts are engaged vs dark. Both pieces of intelligence collapse to zero the moment you upload one big audience. The intent tier and the role segment determine what creative, frequency, and budget allocation actually work — and using only the underlying account list throws all of that away.
This workflow rebuilds the audience structure to match the diagnostic work. 3 intent tiers × 3 role segments = 9 distinct audiences, each with its own LinkedIn targeting parameters, frequency cap, budget allocation, and creative direction. Output is 9 CSVs ready to upload to LinkedIn Campaign Manager plus a campaign architecture document specifying which audiences run which campaigns at which budget weights. Run monthly to keep the audiences fresh as accounts move between tiers.
The gold variables — your brand, monthly LinkedIn budget, ICP role definitions, and the upstream intent scoring output — are the parts you edit. Save the prompt as a markdown file you reuse monthly.
hubspot and linkedin_ads via the free Growthspree MCP. Run Account Intent Scoring Engine first to produce the upstream account list — without it, this workflow has nothing to segment.Sample output for a hypothetical mid-market B2B SaaS account with $30K/month LinkedIn budget, 410 prioritized target accounts (50 T1, 160 T2, 200 T3), and a 90-day sales cycle. Tier-1 cells get the largest budget allocation; T3 cells get smallest but still funded for awareness build.
Building the 3×3 audience matrix from your Intent Scoring output now.
Segmented 410 accounts into 9 audience cells with 6,847 total contacts. All 9 cells exceed the 1,000-member auction efficiency floor — no consolidation needed. The T1 Decision Makers cell (200 members across 50 enterprise accounts) is the highest-leverage segment and gets 35% of monthly budget. Total monthly budget allocation: $30,000.
Account intent shifts monthly. Run Account Intent Scoring Engine on the 1st, immediately run ABM Audience Builder, refresh the 9 LinkedIn audiences via re-upload.
This workflow assumes a current intent scoring output exists. If you don't have one, run Account Intent Scoring Engine first to score accounts and identify the prioritized list. The two workflows together form Track 03's diagnose-and-build loop. Both authorized via the same Growthspree MCP — no extra setup needed.
Run Intent Scoring →Edit the gold variables — your brand, average ACV, monthly LinkedIn budget, sales cycle, and the three role definitions (Decision Makers / Influencers / Users). The role definitions are the most important configuration — they translate directly into LinkedIn job-title targeting parameters. Be specific: "VP Engineering, CTO, Head of Engineering" not "executives." Save the configured prompt as a markdown file you reuse monthly.
Claude pulls the account list from HubSpot, queries contacts by role pattern, segments into 9 cells, validates audience sizes against the 1,000-member floor, and outputs CSVs in LinkedIn Matched Audience upload format. Audience size validation is the most important automated check — it prevents you from shipping cells that will burn budget on auction premiums.
Upload the 9 CSVs to LinkedIn Campaign Manager → Matched Audiences → Upload list. Build the 4 recommended campaigns mapping the 9 audiences. Critical: layer company-level frequency caps via QLA or equivalent — LinkedIn's native capping is per-member only, which causes 80% of budget to concentrate on 20% of accounts. Without company-level caps, the segmentation work is half-undone.
Why company-level caps matter →Same intent-tier × role-segment foundation, different scope. Pick the one that matches your team size and budget.
9 cells require enough budget to keep each cell above the auction efficiency threshold. For mid-market budgets, the compressed version uses 2 tiers (Active vs Awareness) × 3 roles = 6 cells, concentrating budget where signal is strongest.
For accounts where T1 cells fall below the 1,000-member floor. Builds lookalike audiences from T1 contacts to expand reach to similar profiles at non-target accounts. Useful when target account list is small (< 30 accounts) but ACV is high.
For teams running ABM with coordinated SDR outreach. The same intent score + role segmentation that drives ad audiences should drive SDR sequence selection. Produces 9 SDR sequence templates parallel to the 9 ad audiences — same buyer state, two channels.
Run Account Intent Scoring Engine on the 1st, run ABM Audience Builder right after, upload the 9 CSVs, layer company-level frequency caps, ship. The architecture is the work — once it's set up, monthly refreshes take 30 minutes. Or have senior GrowthSpree operators build and maintain the 9-cell architecture, coordinate creative per cell, and run QLA-enforced company-level frequency capping — the same operating motion run across 300+ B2B SaaS accounts.