Case Study: How an E-commerce Brand Cut Wasted Ad Spend by 38%

Industry: Fashion E-commerce
Monthly Ad Spend: Rp 150-200 million (~$10,000 USD)
Primary Channels: Meta Ads (Instagram, Facebook)
Problem: High click volume but consistently low conversion rates

38%

Bot Traffic Detected

2.4x

ROAS Improvement

Rp 57M

Monthly Savings

The Problem: Clicks Without Customers

When the marketing team at this Jakarta-based fashion brand came to us, they had a familiar story. Their Meta Ads campaigns were generating thousands of clicks per day, CTR looked healthy at 2.1%, and CPC was reasonable for their category.

But something wasn't right. Their conversion rate had been declining for months—from 1.8% to 0.9%—and nothing they tried seemed to fix it.

They had already:

A/B tested 12 different landing page variations
Worked with a copywriting agency on new ad creatives
Adjusted targeting multiple times
Increased their product photography budget

None of it moved the needle. The clicks kept coming, but the sales didn't follow.

The Investigation: Finding the Real Problem

The first step was to audit their traffic quality. We installed tracking on their landing pages and began collecting behavioral data alongside the standard analytics.

Week 1: Data Collection

Within the first week, patterns emerged that weren't visible in Google Analytics or Meta's dashboard:

Session duration gap: 67% of paid traffic sessions lasted under 3 seconds. Organic sessions averaged 2 minutes 15 seconds.
Scroll depth: 41% of paid visitors never scrolled at all—the page loaded, then they "left."
Mouse movement: 38% of sessions showed no mouse movement or showed patterns inconsistent with human behavior (perfectly linear paths, no hesitation).
JavaScript execution: 12% of "visitors" weren't executing JavaScript properly, suggesting automated browsers.

Key insight: The gap between Meta's reported clicks and actual engaged sessions was 38%. More than a third of their ad spend was going to traffic that would never convert—because it wasn't human.

Week 2: Source Analysis

We traced the bot traffic to understand where it was coming from:

Audience Network: 62% of detected bot traffic came from placements outside Facebook and Instagram proper
Specific publishers: Three mobile apps in the Audience Network accounted for 28% of all bot clicks
Time patterns: Bot activity spiked between 2-5 AM local time, when human traffic was lowest

The Solution: Multi-Layer Filtering

Week 3

Placement Exclusions

Removed Audience Network entirely and excluded the worst-performing placements. Immediate 15% reduction in wasted spend.

Week 4

Traffic Filtering Deployment

Deployed behavioral detection on landing pages. Bot traffic filtered before reaching checkout flow.

Week 5-6

Conversion Optimization

With cleaner traffic data, optimized for actual conversions. Meta's algorithm began learning from real customer behavior.

Week 8

Steady State

New baseline established. Bot rate below 5%. Conversion rate stabilized at 2.1%.

The Results

After eight weeks of implementation and optimization:

Bot traffic reduced: From 38% to under 5%
Conversion rate: Increased from 0.9% to 2.1% (133% improvement)
ROAS: Improved from 1.8x to 4.3x (2.4x improvement)
Monthly savings: Rp 57 million in previously wasted ad spend
Cost per acquisition: Decreased by 52%

"We spent months blaming our landing page and creative when the real problem was that a third of our traffic wasn't even human. Once we fixed the traffic quality issue, everything else started working again."

— Marketing Director

Why This Happens

Bot traffic in digital advertising isn't a bug—it's an economic reality. When publishers get paid per click, some will inflate their numbers. When ad platforms optimize for clicks (not conversions), they'll send traffic that clicks—regardless of whether it buys.

The Audience Network is particularly vulnerable because it includes third-party apps and websites that Meta doesn't directly control. Quality varies wildly between publishers, and fraudulent traffic finds its way into the system.

This isn't unique to Meta. We see similar patterns in TikTok, Google Display Network, and native advertising platforms. Any system that pays for clicks creates an incentive to generate fake ones.

Key Takeaways

Don't assume clicks are real. If your conversion rate is declining despite healthy CTR and engagement metrics, investigate traffic quality before changing your landing page.
Look beyond platform dashboards. Meta's reporting shows clicks, but it doesn't distinguish between human and automated clicks. You need additional tracking to see the full picture.
Audit placements regularly. Audience Network can be valuable for some advertisers, but it requires careful monitoring. Consider excluding it entirely if you're seeing quality issues.
Calculate your real conversion rate. If 38% of your traffic is bot, your 0.9% conversion rate is actually 1.45% among humans. That changes how you think about optimization.

Is Bot Traffic Hurting Your Campaigns?

Get a free traffic quality audit and find out how much of your ad spend is going to bots.

Request Free Audit

Methodology Note

Traffic classification in this case study used behavioral signals including session timing, interaction patterns, browser characteristics, and network analysis. Bot/human classification used a confidence threshold of 85%—traffic below this threshold was classified as "uncertain" and excluded from both categories in reporting.

Results are specific to this advertiser's situation. Actual bot rates and improvement potential vary by industry, geography, ad platform, and targeting parameters.