AdvertisingIndustry ContextWednesday, August 5, 20265 min read

The Averages Illusion – Why Your Account Average Lies to You

PPC Hero8h agoamazonwalmart
The Averages Illusion – Why Your Account Average Lies to You
Executive Summary

A healthy-looking $60K/month account was feeding half its budget to junk conversions. How blended averages lie to you, and the segment that shows the truth. The post The Averages Illusion – Why Your Account Average Lies to You first appeared on PPC Hero.

Source Lens

Industry Context

Useful background context, but lower-priority than direct platform, community, or operator intelligence.

Impact Level

medium

Use this briefing to decide whether your team needs an immediate workflow, policy, or reporting change.

Key Stat / Trigger

No single quantitative trigger surfaced in this report.

Focus on the operational implication, not just the headline.

Relevant For
Brand SellersAgencies

Full Coverage

By Igor Ivitskiy - Wednesday August 5, 2026 Share (Twitter) WhatsApp Summarize ChatGPT Perplexity Grok Google AI On average, Elon Musk and I are worth about half a trillion dollars each (he’s worth a trillion, I’m slightly less). True, and completely useless. A blended ROAS does the same thing: a couple of winners hide a pile of money-losing segments.

Digging through other people’s ad accounts is what I do all day, and almost every account greets me the same way: one healthy average on top, and underneath it, segments that would never survive on their own. Which is which? No way to tell until you pull the number apart.

One recent audit showed the whole mechanism so cleanly that I’ll just walk you through it. The One Question That Unravels a Healthy Account A store, roughly $60,000 a month in ad spend. Strong ROAS, a CPA the owner was happy with, solid numbers for that niche. His words: the ads work great, nothing to fix here.

All he wanted was to know which parts were pulling the weight and which were coasting. My first question, always the same one: what exactly counts as a conversion here? He shows me two conversion actions, both imported from Google Analytics, both primary. One named Purchase. One named Leadgen. You read names like that and relax.

Purchase means somebody bought, Leadgen means somebody filled in a form. That’s what I figured. He’d believed the same thing all along. Then, halfway into the call, it came out: Leadgen was a bucket. Inside Analytics, that one event collected a real consultation form, clicks on the phone number, clicks on the address, and visits to the contact page.

Actions worth real money and actions worth almost nothing, all counted as one thing. Why would anyone build it that way? Turns out the developer who set up the tracking had decided that many conversion actions would be messy, so he made a single action and stuffed every event into it. Two mistakes, stacked. Unequal events blended into one conversion action.

Then that action got promoted to primary and treated as an equal of actual sales. Purchases were running about five times behind the bucket. So who was steering the bids? Not buyers. Whatever tripped the counter most often, and the counter mostly counted phone-number clicks. We kept digging.

His YouTube Demand Gen campaign, which he rated as highly as Search, was mostly feeding the bucket and bringing few sales. His Dynamic Search campaigns, which he believed were sweeping up long-tail queries nobody had time to build out, were mostly catching his own brand name.

Out of $56,000 spent in the 30 days I reviewed, roughly half went to campaigns tuned to the wrong signal. And the blended average looked great the entire time. The owner went quiet for a while. Then he wrote a single word in his notes: fix. Smart Bidding Chases Whatever You Feed It Ten years ago, a setup like this cost you clean reporting, and that was that.

Today it costs the budget itself, because Smart Bidding reads your primary conversions and goes shopping for more of whatever fills them. With narrow exceptions like custom goals, primary actions are what bidding optimizes toward in the campaigns that use those goals; Google’s documentation spells this out.

A junk bucket sitting in the primary column becomes a budgeting instruction, and the machine follows it faithfully: find me more people who click phone numbers. The Fix, and What It Costs You Split the bucket into separate conversion actions. Small stuff goes to secondary, where you can watch it without letting it steer bids.

Purchases stay primary, plus the one form that actually predicts revenue. Values get assigned by what each action is worth to the business, not one-for-one. And before you call anything junk, check margin and lead quality first. A phone call can be worth more than an order; only the client’s sales data can settle that. One warning.

The reported numbers will get worse before the business gets better. Conversion counts drop. Bidding re-enters a learning period, so volume will wobble for a few weeks. Let it. What you’re watching is the account switching from a comfortable number to an honest one.

How to Run This Diagnostic Yourself Image: Segmenting the campaign table by conversion action in Google Ads. Open the campaign table. Click Segment, pick Conversions, then Conversion action. The single conversion column opens up into its composition, and the picture changes completely. One campaign: 150 conversions, 100 of them purchases, 50 phone clicks.

Its neighbor: also 150 conversions. Zero purchases. The unsegmented column shows you twins. The segmented one shows a money-maker sitting next to a click-collector. Carry the same segment down into ad groups, keywords, audiences. Image: Example of a conversions column broken down by conversion action. The Bottom Line The savings here are a small prize. Th

Original Source

This briefing is based on reporting from PPC Hero. Use the original post for full primary-source context.

View original
LinkedIn Post Generator

Style

Audience