Smart Bidding Optimizes Superbly. The Question Is What.
The account is switched to Smart Bidding — by choice, or by a recommendation banner that would not go away. Within weeks the dashboard confirms the decision: conversions up, cost per conversion down. The bank account, checked at the end of the month, declines to confirm anything.
Nearly everything written about Smart Bidding argues one question — manual or automated. It is the wrong question, and both sides of it miss what actually decides the outcome.
The right question is not whether the algorithm optimizes well. It optimizes superbly. The question is what it is pointed at.
What Smart Bidding Is
Smart Bidding is the family of automated bid strategies in Google Ads — Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value. At every auction it sets the bid using signals no human can process at that speed or scale: device, location, hour, browsing session, audience membership, dozens more.
At the mechanical task of bidding, no manual operator competes. Conceding this early matters, because the real issue is somewhere else entirely.
The Wrong Debate
Manual-versus-automated compares two ways of steering. It skips the destination. An algorithm, like any optimizer, does exactly one thing: it maximizes the number it is given. Judging Smart Bidding by how well it optimizes, without asking what it optimizes, is grading a courier on speed without asking where the parcels go.
So the debate worth having is about the number itself.
What It Is Pointed At
Every Smart Bidding strategy optimizes toward conversions — as defined, counted, and attributed by the platform. That count is not a record of payments. It includes counter firings on pages, modeled conversions estimated where tracking is incomplete, view-through and cross-device attributions assigned by the platform's own rules.
A payment arriving in a payment system is a fact the algorithm never sees. It steers, auction by auction, toward the platform's own tally of what advertising success looks like — and it reaches that target with real efficiency. Whether the tally corresponds to money is outside its job description.
Why the Dashboard Improves
Point a competent optimizer at a number and the number improves. This is close to a law. Cost per conversion falls, conversion volume rises, and the report page reads as vindication — regardless of what happened in the business.
The improvement is real, in the metric's own terms. Traffic shifts toward users likely to fire the counter: the easily-counted, the repeat visitors, the almost-bought-anyway. None of that requires a single additional sale to show up as progress on screen.
Whose Number It Is
One step deeper sits the fact that reframes the whole arrangement. The conversion count is kept by the platform — the party whose revenue is the advertiser's spend. The same count then denominates its recommendations: raise the budget, expand the targeting, switch the remaining campaigns to automation.
There is no conspiracy in this, only structure. The seller keeps the score, the score justifies the next purchase, and the optimizer — honestly and skillfully — maximizes the seller's scoreboard. Who to Trust to Run Paid Ads applies the same two-question test to every party in the field; the platform's automation answers both questions the same way: paid on spend, optimizing its own count.
The Test That Settles It
Whether Smart Bidding is working for a particular account is checkable, but not on the platform's screen. The check: payments — actual, payment-system payments — attributed to each ad, against what each ad spent, over the same period, in records the platform does not keep. When that comparison is visible, the algorithm's work has a verdict. When it is not, "it's working" is a quote from the seller.
Imported conversion data narrows the gap — payment events can be fed back into the bidding — but the counting and the attribution remain the platform's, and the platform still grades its own homework. The verdict has to live outside. Why the analytics screen next door cannot deliver it either is its own story: Google Analytics Is Useless.
Growity points optimization at the other number: every ad measured by actual payments attributed to it, budget moved to the ads that verifiably sell, the rest stopped — continuously and automatically.
Common Questions
Does Smart Bidding work?
At its own objective — reliably. It raises the platform-counted conversion number with real efficiency. Whether that number moves together with payments is a separate question, checkable only in records the platform does not keep.
Is manual bidding better than Smart Bidding?
At the mechanical task of setting bids, no. A human cannot process auction-time signals at that scale. The manual-versus-automated debate misses the deciding factor — the number the automation is pointed at.
Why do reported conversions rise while sales stay flat?
Optimizing a number improves that number. Traffic shifts toward users likely to fire the counter — the easily-counted, the almost-bought-anyway — which registers as progress without requiring a single additional sale.
What does Smart Bidding actually optimize toward?
Conversions as the platform defines and counts them: tag firings, modeled estimates where tracking is incomplete, view-through and cross-device attributions by the platform's rules. Not payments.
Can Smart Bidding optimize toward real revenue?
Payment events can be imported into the bidding, which narrows the gap. The counting and attribution remain with the platform, which is also the seller. The verdict on whether it works still has to come from outside — payments against spend, per ad.