B2B SaaS Lowers Google Ads CPA by 35% with Better Measurement and Experimentation

Client Background

An anonymized B2B SaaS company selling a workflow tool for mid‑market teams was primarily relying on inbound leads from content and referrals.

They had been investing in Google Ads for several years with:

• Monthly Google Ads spend of $35K–$45K
• A mix of brand, competitor, and non‑brand search campaigns
• Self‑serve free trial and “Request a demo” as key CTAs

Despite healthy traffic, the marketing and sales teams were struggling with:

• Rising CPAs on non‑brand search
• Inconsistent demo quality across campaigns
• Lack of confidence in the numbers from GA4 vs Google Ads

Challenge

Before we started, the main issues were:

1. Measurement gaps – form submissions, trial sign‑ups, and demo requests were tracked inconsistently across GA4 and Google Ads
2. High CPA / low ROAS – especially on non‑brand search, where many leads did not progress to opportunities
3. Unclear campaign impact – the team could not confidently say which campaigns or keywords actually drove pipeline, not just form fills
4. No systematic testing – bid strategies, structures, and landing pages had been changed ad‑hoc without clear hypotheses or evaluation

Approach

Phase 1: Measurement & Data Audit (Week 1)

We began by aligning GA4 and Google Ads so that everyone was looking at the same story.

• Audited GA4 events and Ads conversion actions
• Discovered:
• Duplicate conversion actions in Ads still marked as primary
• GA4 events for `demo_request` and `trial_signup` firing on page load instead of successful submissions
• No distinction between all leads and sales‑qualified leads (SQLs)
• Redesigned the event and conversion setup to:
• Use a single, clean `lead_submit` event with parameters for lead type and source
• Create a secondary GA4 event for `sql` synced from the CRM to tie ads to down‑funnel quality
• Import the right conversions into Google Ads (lead + SQL) and disable legacy actions
• Set up custom GA4 reports showing leads and SQLs by campaign/keyword

Deliverables:

• Measurement map showing click → session → lead → SQL
• Issues & fixes log documenting each tracking change

Phase 2: Account Diagnosis & Quick Wins (Week 1–2)

With reliable data, we audited the account structure and performance.

• Reviewed campaign and ad group setup:
• Dozens of small campaigns with overlapping themes and budgets
• Broad match keywords mixed with exact/phrase in the same ad groups
• Little separation between high‑intent and research queries
• From the last 90 days of data we found:
• ~22% of spend on keywords that had zero SQLs despite occasional form fills
• Brand campaigns optimized to all leads, masking the true cost of qualified demos
• Geo/device segments with consistently poor performance

We implemented several quick wins:

• Paused and excluded obviously irrelevant and non‑converting queries
• Split high‑intent keywords (e.g. “{category} software”, “{category} platform”) into focused campaigns
• Tightened budgets on under‑performing regions and devices
• Changed optimization for brand campaigns to focus on SQLs instead of all leads

Deliverables:

• Account audit summary (observations → issues → recommendations)
• Prioritized quick wins list with estimated impact and risk

Phase 3: Experiment Design & Implementation (Week 3–4)

Next we designed a small set of experiments to test bigger strategic questions.

Key hypotheses:

1. Switching from manual CPC to tCPA on core non‑brand campaigns would lower CPA without hurting volume.
2. Separating “solution aware” keywords (e.g. “{category} software”) from “problem aware” keywords (e.g. “how to manage {problem}”) would improve both conversion rate and SQL rate.
3. A more focused demo‑first landing page would improve demo request rate from high‑intent traffic.

Implementation:

• Used Google Ads Experiments to A/B test manual CPC vs tCPA on selected campaigns
• Created new campaigns for high‑intent vs problem‑aware groups with tailored ad copy and budgets
• Set up an A/B test between the existing landing page and a new version emphasizing social proof and friction‑reduced form

Success metrics:

• Primary: cost per SQL and SQL volume
• Secondary: lead conversion rate, overall CPA, and demo‑to‑opportunity rate from CRM

Deliverables:

• Experiment design docs for each test
• Experiment log summarizing configuration and expected lift

Phase 4: Results Analysis & Account Restructuring (Week 4+)

After 4–6 weeks of testing, the results were clear:

• tCPA on core non‑brand campaigns reduced cost per SQL by ~18% while maintaining volume
• High‑intent campaigns showed significantly higher SQL rate and more stable CPA compared to mixed‑intent setups
• The new landing page improved demo request rate by 23% from high‑intent traffic

Based on these findings we:

• Rolled out tCPA bidding to more non‑brand campaigns with guardrails
• Consolidated overlapping campaigns into a smaller set of clearly‑defined structures (brand / competitor / high‑intent / problem‑aware)
• Shifted budget towards high‑intent campaigns and paused consistently weak segments
• Made the new landing page the default for core demo traffic

Deliverables:

• Before/after charts for CPA, lead volume, and SQL volume
• Updated account map showing the simplified structure

Phase 5: Ongoing Optimization & Reporting

With the foundation in place, we moved to a lighter ongoing cadence.

• Weekly:
• Monitor for anomalies and bid/budget adjustments
• Review search terms and add negatives where needed
• Monthly:
• Update performance summary by campaign/keyword and by lead stage
• Plan 1–2 new experiments (e.g. new messaging angles, new geos)
• Share a concise report with key insights and agreed actions

Deliverables:

• Example monthly report and dashboard views
• Rolling experiment roadmap snapshot

Results

Six months after the engagement started:

• Overall Google Ads CPA decreased by 35% (from \$420 to \$275)
• Qualified leads (SQLs) increased by 1.5×, with a higher demo‑to‑opportunity rate
• Non‑brand campaigns, previously barely breakeven, became a reliable source of pipeline
• The client gained a much clearer view of which campaigns and keywords truly drove revenue

Key Lessons

Tie the outcome back to your principles:

1. Measurement first – why fixing tracking changed the conversation
2. Hypothesis‑driven testing – how structured experiments reduced guesswork
3. Continuous optimization – how a simple cadence prevented performance drift

Call to Action