From “budget drain” to 40–50 leads per week: how Infinity Logistix reduced CPL to $13–20 by testing 70+ hypotheses over 14 months

Optimization stages: from “cheap leads” to stable contracts

• Stage 1. Realizing that CPL ≠ results
  At the start, campaigns produced “nice-looking” leads at $16–25, but the funnel data showed failure: people didn’t answer calls, didn’t match criteria, and didn’t convert into deals. It became clear that CPL is a superficial metric that does not reflect business outcomes. From that point, every hypothesis was evaluated based on lead movement toward qualification and contracts.
• Stage 2. Hypothesis testing and audience discovery
  We tested B2B, B2C, different language segments, and landing pages. B2B looked good on paper but didn’t generate contracts, while Typeform proved inefficient in cost per lead. In contrast, Russian-speaking and Central Asian audiences delivered higher CPL but significantly better lead quality.
• Stage 3. Working with look-alike and quality filtering
  We launched look-alike audiences and separate optimization events. Some hypotheses failed due to either high cost or lack of funnel progression. The best results came from LAL based on website visitors, where both lead volume and real movement toward deals appeared.
• Stage 4. Funnel optimization and abandoning “cheap leads”
  We shifted from “lead” optimization to deeper funnel events (funnel key events) and adjusted geography. CPL increased, but lead quality improved: more qualified calls and signed contracts. This reinforced a key principle — it’s better to pay more if it leads to real revenue.
• Stage 5. Repositioning and focus on the Car Hauler segment
  We restructured offers around real driver motivations and pain points. This revealed a strong segment — Car Hauler — which became a separate campaign with dedicated creatives and messaging. It resulted in scalable lead flow, gradually lower CPL, and consistent deal flow.
• Stage 6. Failed experiment with broad automation
  We tested a single broad campaign with automated budget allocation (Andromeda). It led to higher CPMs and CPCs without qualified leads. The experiment showed that in niche markets, relying solely on algorithms doesn’t work — precise segmentation and manual control are essential.