39% CPL reduction in 11 weeks — insurance lead gen rebuilt from the ground up.

The Challenge

Acko's Google Ads account had grown organically over 3 years without structural review. By the time we inherited it, it had 120+ ad groups, overlapping keyword themes, manual CPC bidding across all campaigns, and no separation between brand, category, and competitor intent. Monthly search term reports revealed ₹1.8 lakh in spend on irrelevant queries — competitor brand names, informational queries, and entirely unrelated searches. Generation CPL was ₹680 against a target of ₹450. The account had volume but no efficiency.

The Solution

Results

The combination of structural changes, AI-driven optimisation via Adtrafix OS, and weekly cadence produced measurable improvement starting in week 3, compounding through month 3:

Deep Dive: Why Search Term Waste Compounds So Fast

One of the most counterintuitive aspects of Google Ads management is how quickly search term waste compounds when left unaddressed. In Acko's account, a single broad match keyword like "insurance" was matching against thousands of irrelevant queries — competitor names, informational research queries, job listings, and unrelated financial products. Each of these queries generated a small amount of spend individually. Across 120 campaigns, those small amounts accumulated to ₹1.8 lakh per month.

The reason monthly audits miss this is simple: the waste distributes across so many individual terms that no single term looks alarming on its own. Only a systematic, threshold-based audit — flagging every term above ₹500 with zero conversions — reveals the full picture. This is what Adtrafix OS automated, running every week rather than every month.

The tCPA Migration Timeline

Week 1–2 were spent entirely on structural rebuild — no bidding changes. The cardinal rule of tCPA migration is that the algorithm can only learn on clean data. Migrating to Smart Bidding before fixing campaign structure means the algorithm optimises toward a flawed baseline and never reaches its potential.

Week 3–4: First campaigns migrated to tCPA — starting with the three highest-conversion campaigns (brand keywords, top city clusters) where learning period completion was fastest. Target CPA set at 20% above observed average to give the algorithm room to explore without overspending.

Week 5–8: Remaining campaigns migrated in batches. Each batch given a 2-week protected learning window — no budget changes, no target adjustments, no campaign restructures during this period. Adtrafix OS monitored learning status daily, flagging any "learning limited" signals for immediate review.

Week 9–11: Full account on tCPA, all learning periods complete. CPL began its consistent decline as the algorithm accumulated conversion signal on clean, intent-clustered campaigns. By week 11, Generation CPL had moved from ₹680 to ₹412 — a 39% reduction — with Triggered CPL declining proportionally and trigger rate improving from 19% to 24%.

What Sustained the Improvement

Many CPL improvements are one-time wins — a structural fix that delivers a step-change improvement and then plateaus. Acko's improvement continued compounding beyond the initial 11-week period because the weekly optimisation cadence via Adtrafix OS kept removing new waste and reallocating budget toward emerging high-intent signals. The structural fix created the foundation; the continuous operating cadence built on it week after week.

Key Takeaway

This case demonstrates what changes when AI is embedded into the operating model rather than used as an occasional tool. The results compound weekly — each optimisation feeds the next — rather than spiking once and stabilising. Adtrafix OS running behind every decision is what separates a one-month win from a structural improvement.