Most conversations about Google Ads performance start from a quiet assumption: that Google is, broadly, on your side. The interface reinforces it. The recommendations tab, the optimization score, the helpful tooltips — all of it is framed as if Google is trying to help you succeed.
That framing makes sense as a user experience. It does not make sense as a description of what the system is actually doing.
Google Ads is not a tool for finding advertiser efficiency. It’s a marketplace, and the marketplace has its own objective function. Once you accept what that objective actually is, a lot of performance reports that look confusing start to read more clearly.
What Google is actually optimizing for
The cleanest way to describe Google’s objective is this: expected revenue per query, subject to a user experience constraint.
In plain language, Google is trying to maximize some combination of how likely someone is to click an ad, how much that click is worth, and how many ads it can show on the page — but only up to the point where the search experience gets bad enough to hurt future usage. User experience isn’t the goal. It’s the guardrail.
Your conversion rate, your cost per acquisition, your return on ad spend — none of these are in that function. They matter only indirectly, and only because advertisers who lose money eventually pull budget, which would reduce Google’s revenue over time. Advertiser success is a sustaining condition for the machine, not its purpose.
This is not a cynical take. It’s how two-sided marketplaces work. It’s how any auction-based platform with market power works. Pretending otherwise leads to bad diagnosis.
Why this reframe changes how you read reports
Here’s a pattern I’ve been looking at recently, and it’s a good example of how the standard story breaks down.
An account had a performance drop. Impression share was falling. Lost impression share due to rank was rising. The instinct — and it’s the instinct most teams go to first — was that the ads had gotten worse. Maybe ad copy had fatigued. Maybe the landing page experience had slipped. Maybe Quality Score was eroding and taking expected click-through rate with it.
That’s a reasonable first hypothesis. But when I actually looked at the data, it did not hold up.
Click-through rate had not structurally collapsed. Landing page experience was still strong. Some quality signals had softened — ad relevance, expected CTR — but not by enough to explain the size of the drop.
The more interesting pattern was this: the account was showing up in fewer auctions overall, but when it did show, it was showing near the top of the page. And the average price it was paying per click had actually come down.
That combination does not say “your ads are weak.” It says something quite different.
What “the ads got worse” stories miss
If the creative and the landing page were the real problem, you’d expect the account to be in lots of auctions but losing the good positions. You’d expect click-through rate to drop, quality scores to slide meaningfully, and costs to climb as Google penalized weaker ads.
That’s not what was happening. The account was entering fewer auctions, but competing just fine in the ones it did enter.
That’s a very different diagnosis. It says the issue is not “you are weak inside the auction.” It says “you are being filtered out of auctions earlier in the process.”
Several things can cause that, and most of them have nothing to do with your creative. Bidding may have become more conservative, either because the automation got cautious or because budgets tightened. Smart Bidding might be filtering harder because its prediction of value in certain auctions dropped. Competitors may have gotten more aggressive in the higher-value auctions, shifting the threshold you need to clear. Or Google’s own auction dynamics may have shifted — ad load, eligibility rules, or the definition of what counts as a “commercial” query.
Any one of those can produce the exact pattern described above. And none of them is a creative problem.
The reframe in one sentence
Once you stop assuming the system is pointing you toward efficiency, the question you ask about a performance drop changes.
The old question is: what got worse about our ads?
The better question is: what changed about the set of auctions we’re allowed to seriously compete in?
That second question is usually the right one, because the platform can unilaterally change the answer. Your ads didn’t decide to show less. Your creative didn’t decide to be filtered out of a subset of queries. The auction decided. And the auction is run by a system whose objective is not your objective.
What to do with this
I want to be careful here. The reframe is not “Google is adversarial.” It’s more practical than that. The reframe is that Google’s interests and your interests overlap in some places and diverge in others, and you should know which is which.
A few things follow from that.
Treat platform recommendations skeptically
Every auto-apply suggestion, every “opportunity,” every broad-match-by-default behavior points toward platform revenue first and your efficiency second. Sometimes the recommendation is genuinely useful. Sometimes it’s useful to Google in ways that don’t show up on your P&L for another quarter.
When performance drops, investigate auction entry first
If impression share is falling but the auctions you do win are still strong, the problem is probably not your ads. It’s probably where you’re being allowed to compete. Fix that before you rewrite the creative.
Ask uncomfortable questions about the automation
Smart Bidding works well a lot of the time. It also occasionally makes decisions that are optimal for the platform’s throughput and merely acceptable for your return. Those are not the same thing, and the difference adds up over a quarter.
The takeaway
Advertiser performance is not the product. You are not the customer in the way the interface makes it feel. Revenue per query is the product, and you are one of the inputs the system tunes to produce it.
That doesn’t mean Google Ads is a bad place to spend money. It’s often the best place. It does mean that when something looks off in the reports, the first question should not be “what did we do wrong.” It should be “what did the auction decide to do differently this month.”
Once you see it that way, the data stops feeling random.