Nexus Capital Group
Multi-country lead generation · CRM-to-Ads feedback loop
Nexus Capital Group runs financial services lead generation across 10+ international markets via Google Ads. Lead quality varies significantly by geography — client LTV and close rates differ by 3–5× between high-value and low-value markets.
At audit, all markets ran from a consolidated campaign structure with blended bidding targets. The Conversion API was not configured — meaning CRM close data (which leads became clients) never informed the bidding model. Google Ads was optimizing toward cheap leads, not profitable clients. Four genuinely high-LTV markets were effectively subsidizing seven unprofitable ones.
Countries were reclassified into three tiers based on CRM close rate, average LTV, and historical CPL — then given separate campaign structures, budgets, and tCPA targets.
Aggressive tCPA · max budget allocation · dedicated campaigns
Conservative tCPA · volume caps · test and scale
Awareness budget only · no conversion optimization
The core structural fix: closing the loop between CRM close events and Google Ads bidding signal.
- →Offline conversion uploadGoogle Ads API
CRM close events (client signed, deposit paid) uploaded to Google Ads as offline conversions — tagged with the original GCLID from the lead form submission
- →Conversion API configuredServer-side events
Conversion API installed alongside browser events — recovery of leads where the browser-side tag fired but the click ID was not captured
- →GA4 → BigQuery exportRaw event pipeline
All GA4 events exported to BigQuery for multi-touch attribution modeling — enabling custom attribution windows per country tier
- →Weekly country-priority reviewOptimization cadence
Automated report scoring each country on CPL, close rate, CRM-confirmed LTV, and trend — budget reallocation decisions made weekly, not monthly
An account running across 10+ countries can show a healthy blended ROAS while several individual geos operate at a loss. The profitable countries subsidize the unprofitable ones — and without geo-segmented reporting, there is no signal to act on. The fix is not better creative. It is structural: separate the countries so each can be evaluated independently.
CPL is a cost metric, not a quality metric. A campaign producing cheap leads from low-LTV countries looks efficient. A campaign producing expensive leads from high-closing, high-LTV clients looks inefficient. Without feeding CRM close data back to Google Ads as offline conversions, the algorithm cannot distinguish between the two — and optimizes toward the wrong signal.
10+ countries cannot each be managed independently at scale. The solution was a tiered framework: tier 1 (high LTV, high close rate) gets aggressive tCPA and maximum budget priority; tier 2 gets conservative tCPA with volume caps; tier 3 is monitored but not actively scaled. This created leverage without requiring constant manual adjustment per market.
Standard GA4 reporting does not handle multi-country, multi-campaign attribution at this level of granularity — especially once offline conversion data enters the picture. Exporting raw events to BigQuery and building attribution queries directly gave the control to define attribution windows and touchpoint weighting per country group, rather than accepting GA4's default last-click model.
10+ countries restructured into a three-tier geo framework. The CRM-to-Ads feedback loop was closed for the first time — offline close events feeding directly into Google Ads bidding via API. Four high-LTV markets identified and separated from the aggregate; Tier 1 campaigns now operate on CRM-confirmed conversion data rather than raw lead volume. Weekly country-priority scoring replaced monthly manual reviews.
Fix: 3-tier geo framework · offline conversion API · GA4 + BigQuery attribution
Result: CRM loop closed · 4 high-LTV markets isolated and prioritized
Client name is a pseudonym. Report ID is internal. Country groupings and CPL figures adjusted to preserve client confidentiality. Results vary by market, product, and CRM quality.