Multi-brand automotive group

The Problem

I joined a multi-brand automotive dealership group (10 brands, including prestige and EV marques) on the NSW Central Coast. Marketing ran on dealer-principal instinct and OEM co-op spend. Each brand bought its own digital, no one owned the funnel, and the CRM held duplicate leads with no source attribution. Sales blamed marketing, marketing blamed Facebook, and nobody could say which channel paid for which car. The starting cost per qualified lead, averaged across brands, sat at roughly four times the rate I would settle it at.

The System

I rebuilt the funnel as one system across all 10 brands. Google Ads and Meta moved to a shared account structure with brand-level conversion tracking in GA4. I designed a four-tier lead-scoring schema (test-drive booked > finance enquiry > brochure download > newsletter) and wired it into the dealership CRM so sales saw a ranked queue, not a flat list. I rebuilt UTMs end-to-end and modelled data-driven multi-touch attribution (last-non-direct click as fallback) against booked appointments — not against clicks. I set a weekly budget-shift ritual: every Monday the team reallocates spend by attributed CPL, not by which brand shouted loudest. Claude drafts the first-pass campaign copy and the segment briefs; about 60% gets rejected against a human-written control, the 40% that ships outperforms baseline on open-rate often enough to keep the loop. n8n syncs the qualified-lead score back to the dealer-management system so the ranked queue is the queue sales actually works.

The Outcome

Across six years the system shifted the economics. Cost per qualified lead dropped 83% versus the pre-2019 baseline ($123 → $21). Lead-to-sale conversion lifted 163% (0.8% → 2.1%). The CRM now ingests 400–500 qualified leads per month, attributed to source and ranked for sales. Marketing is accountable for $5M in attributed annual sales across the group, measured against booked deliveries, not clicks. The $920K annual marketing budget is reallocated weekly on attributed-CPL data, and the weekly meeting takes 30 minutes because the numbers do the arguing. I run a team of two plus an external creative agency.

The bet I lost: the first attribution model was last-click. Six months in, the data was telling us to over-invest in branded search and starve top-of-funnel — exactly the wrong move for a 10-brand portfolio with low organic recognition outside the metro. I rebuilt to data-driven multi-touch the following quarter and recalibrated the budget. Worth saying because it's the kind of mistake the system was designed to catch fast.

What Transferred

The component I expect to translate to the next role is the lead-scoring schema and the Monday budget-shift ritual — they convert a marketing team from a creative function into an operating discipline, regardless of industry. The AI loop — Claude for first-pass research and creative, n8n for the clerical wiring — is the same loop in any vertical. The 60% rejection rate is the discipline; the 40% kept is the speed.