Is it right to blame all our woes on Google?
Not necessarily. I don’t think it’s Google’s fault I burned myself on a pan this morning. But I do think it’s Google’s fault that I spent years being deeply confused about CPC vs CPM, Smart Bidding, conversion optimization, bid caps, automated delivery strategies, and whatever horrifying acronym LinkedIn invented this quarter.
I was confused.
Then I launched a few ad campaigns and started fielding way too many internal questions about why CPCs suddenly doubled, why lead quality collapsed, why “maximize clicks” somehow produced worse results, and why every ad platform seemed to behave differently despite using almost identical terminology.
So this article is my attempt at a maximally useful breakdown of how internet advertising systems actually work. Because once you understand the architecture underneath the platforms, almost all of the confusing behavior suddenly becomes much easier to reason about.
The First Thing I Learned: These Systems Have Layers
The biggest misconception in digital advertising is that CPC, CPM, bidding, optimization, and delivery are all basically the same thing.
They aren’t.
Modern advertising systems are really several different systems stacked on top of each other. The auction system determines which ad wins placement. The billing model determines when you’re charged. The bidding strategy controls how aggressively the platform participates in auctions, while the optimization goal tells the algorithm what outcome it should prioritize.
These layers constantly interact with each other, which is exactly why ad platforms feel so confusing when you first start using them. A campaign can bill on CPC while optimizing for conversions. Another can bill on impressions while still attempting to maximize leads. Two campaigns may technically use the same billing model while behaving completely differently because the optimization layers underneath them are different.
Once I understood this separation, a lot of previously irrational platform behavior suddenly started making sense.
CPC and CPM Are Billing Models, Not Strategies
For years I thought Google Ads simply meant:
“You pay per click.”
That’s not completely wrong. Google Search was originally built around PPC advertising, and Search campaigns today are still primarily CPC-driven. But the broader advertising ecosystem evolved far beyond that model years ago.
CPC simply describes when you’re charged.
If your ad receives:
- 10,000 impressions
- and 50 clicks
you only pay for the 50 clicks.
CPM works differently. With CPM billing, you pay based on impressions regardless of whether users engage with the ad at all. If your CPM is $10 and your ad receives 10,000 impressions, you spend roughly $100 whether anyone clicks or not.
At first glance this makes CPM feel riskier, but the important realization is that billing and optimization are not the same thing.
That distinction is the thing most marketers accidentally collapse together.
Billing Is Not Optimization
This was the mental shift that finally made the whole ecosystem click for me.
A platform can:
- bill you on CPC,
while: - optimizing for conversions.
Or it can:
- bill you on impressions,
while: - optimizing for engagement, leads, or purchases.
Those systems are related, but they solve different problems.
Modern ad platforms are no longer simple traffic marketplaces. They are prediction engines attached to auction systems. The platform is constantly estimating:
- who is likely to click,
- who is likely to convert,
- what a user is statistically worth,
- and how aggressively it should bid to acquire them.
That prediction layer is what transformed advertising platforms from relatively straightforward marketplaces into deeply complex machine-learning systems.
And it’s also why lower CPCs frequently fail to improve campaign performance.
Why Lower CPCs Often Produce Worse Results
Most marketers instinctively assume that lower CPCs mean better campaigns. On the surface, this sounds perfectly reasonable. If traffic becomes cheaper, performance should improve.
The problem is that internet traffic is not uniformly valuable.
The cheapest clicks are often accidental clickers, curious browsers, low-intent researchers, or people casually engaging with content while doomscrolling through six other tabs. Cheap traffic is easy to acquire precisely because nobody else in the auction particularly wants it.
The users most likely to convert, on the other hand, are often significantly more expensive because every advertiser in the system is competing for them simultaneously.
This creates one of the strangest dynamics in digital advertising: campaigns sometimes improve because CPCs increase.
At first that feels backwards. But once you understand that modern platforms are optimizing toward predicted outcomes instead of raw traffic volume, it starts making much more sense. The algorithm is no longer trying to buy the cheapest clicks. It’s trying to buy the users most likely to generate value.
The Four Main Bidding Strategies
Once you understand the difference between billing and optimization, the various bidding strategies across Google, Meta, and LinkedIn become much easier to interpret.
Even though every platform names things slightly differently, most systems revolve around four core approaches.
Manual CPC
Manual bidding is the most straightforward system. You set the maximum amount you are willing to pay for a click, and the platform participates in auctions within those constraints.
This gives advertisers the highest degree of control, which is why many people still prefer it early in a campaign lifecycle. If CPCs begin getting out of control, manual bidding creates hard ceilings that prevent runaway costs.
The downside is that you are now responsible for estimating competitive bid levels yourself. If your bids are too low, impressions collapse, auctions are lost, and delivery slows dramatically. This becomes especially painful on platforms like LinkedIn where inventory is limited and clicks are naturally expensive.
Enhanced CPC (eCPC)
Enhanced CPC is essentially a hybrid system between manual control and automation.
You still set your bids manually, but the platform is allowed to raise or lower them slightly based on predicted conversion likelihood. Think of it as giving the algorithm some flexibility while still maintaining guardrails around aggressive bidding behavior.
This is often where advertisers feel most comfortable because it allows them to benefit from machine learning without completely surrendering control to Smart Bidding systems.
And honestly, this is where many campaigns probably belong until conversion tracking becomes reliable enough to support heavier automation.
Maximize Clicks
This strategy tells the platform:
“Get me as much traffic as possible within my budget.”
That sounds attractive in theory. More traffic feels productive. Dashboards look healthy. Click volume increases.
But maximizing clicks frequently produces low-quality visitors because the platform is incentivized to acquire the cheapest possible engagement. The algorithm is not optimizing for lead quality or purchasing intent. It is optimizing for click volume.
This is why “maximize clicks” campaigns can look incredible on surface-level metrics while producing terrible business outcomes underneath.
You bought traffic successfully.
You just may not have bought useful traffic.
Maximize Conversions
Maximize Conversions is where modern advertising platforms become heavily machine-learning driven.
Instead of pursuing cheap clicks, the system aggressively bids toward users it believes are statistically likely to complete the desired action. That action could be:
- a purchase,
- a demo request,
- a lead form,
- or a booked call.
This often causes CPCs to rise significantly.
And this is usually the moment advertisers panic.
The campaign suddenly becomes more expensive on a per-click basis, which feels alarming until you realize the platform is intentionally paying more for users it predicts are more valuable. Higher CPCs can actually be evidence that the algorithm is becoming more selective rather than less efficient.
That distinction is incredibly important.
Google Search Isn’t Really “Just PPC” Anymore
Historically, Google Search Ads were true PPC systems. You manually bid on keywords and paid when someone clicked your ad.
That system still exists.
But modern Google Ads now layers in:
- Smart Bidding,
- conversion prediction,
- audience signals,
- Target CPA,
- Maximize Conversions,
- and real-time auction adjustments.
Even when campaigns technically still bill on CPC, Google is evaluating dozens of signals before deciding how aggressively to bid on a particular auction. Device type, search intent, location, historical behavior, demographics, and predicted conversion probability all influence bidding behavior in real time.
Google no longer sees all clicks as equal.
Some users are statistically worth dramatically more money than others, and the system increasingly prices traffic accordingly.
Why Meta Feels Completely Different
Meta often feels fundamentally different from Google Search because the architecture underneath the platform is different.
Search advertising is intent-driven. Users explicitly tell Google what they want through keywords. Meta, meanwhile, is largely discovery-driven. The platform is attempting to predict what users may engage with before they actively search for it.
That changes the economics of the system dramatically.
Even when advertisers focus on CPCs or cost per lead metrics inside Meta, the platform is usually still operating through impression-based auctions underneath the hood. The algorithm then attempts to optimize delivery toward whichever outcome the advertiser selected:
- conversions,
- engagement,
- clicks,
- or watch time.
This is why Meta campaigns can sometimes produce incredibly cheap CPMs and low CPCs while simultaneously generating awful lead quality. The system succeeded at finding inexpensive attention. It just didn’t necessarily find buying intent.
The Dangerous Side of Automation
Smart Bidding is extremely powerful.
It is also capable of doing extremely stupid things if the underlying data is bad.
This is the uncomfortable reality many advertisers discover after blindly enabling automated bidding systems too early. Automation amplifies the quality of your inputs. If your conversion tracking is weak, your attribution is noisy, your budgets are too small, or your lead quality signals are unreliable, the algorithm can become aggressively wrong very quickly.
This is one of the reasons hybrid systems like Enhanced CPC are often valuable. They create constraints around automation while still allowing the platform to optimize intelligently within those boundaries.
The goal is not to eliminate machine learning. The goal is to prevent the system from confidently optimizing toward garbage data.
The Real Takeaway
Modern advertising platforms are simultaneously:
- auction systems,
- billing systems,
- prediction engines,
- and optimization machines.
Most marketers think these platforms are selling traffic.
In reality, they are selling prediction quality.
Once you understand the distinction between billing, bidding, optimization, and delivery, the platforms stop feeling random. The strange campaign behavior, the rising CPCs, the inconsistent lead quality, the differences between Google and Meta — all of it starts becoming much easier to reason about.
And honestly, that understanding is probably the real skill modern marketers are being paid for now.
Leave a Reply