There's a fantasy that lives rent-free in the head of every PPC marketer: flip on Smart Bidding, lean back, and watch the algorithm quietly print money while you work on other things.

Unfortunately, it is just that, a fantasy.

Before blaming Google's algorithm, take an honest look at the four foundations below. They won't make for a flashy agency presentation, but they're usually what separates an ad account that grows from one that slowly drains budget while everyone points fingers and nobody finds answers.

Susan Yen shares the foundations and exactly how to manage your ads for every location in this episode of PPC Real Talk:

The Four Foundations Smart Bidding Actually Needs

Smart Bidding is only as intelligent as the foundation you build beneath it. Most campaigns that struggle with automation aren't suffering from a platform problem.

They're experiencing a setup issue. Before you blame the algorithm, check whether you've given it what it actually needs to succeed. These four foundations are what separate campaigns that compound over time from ones that spin their wheels on autopilot.

Clean Conversion Signal

Smart Bidding is only as smart as what you feed it. The algorithm optimizes for whatever you tell it matters. If every form fill, including bots, job seekers, and competitors checking your pricing, gets logged as a conversion, the system learns to generate more of exactly that. 

The fix isn't complicated, but it requires clean signals to help it do it's job as efficiently as possible. Import your offline revenue data, weight conversion actions by real business value, and separate qualified leads from raw contact attempts.

If you Feed it spam leads, unqualified form fills, or missing offline revenue, and you've corrupted the signal before the system has a chance. If you teach it garbage leads are valuable, garbage leads is what you will get. 

Intentional Structure

A structural problem shows up in nearly every audit: brand keywords, competitor terms, and cold prospecting all sharing the same campaign, across multiple match types, with no funnel segmentation. These are three completely different conversations. Someone searching your brand name already knows you. Someone searching for a competitor is comparison shopping. Someone searching for a generic category term has never heard of you.

Lumping them together means the algorithm is learning from contradictory signals. It's not confused because it's limited. It's confusing because you handed it chaos and asked for clarity. Blended intent means you're asking the system to optimize for noise, only to wonder why the results feel random.

Enough Data Volume

Nobody wants to hear this one: if your campaign isn't generating enough conversions, Smart Bidding is guessing. Machine learning needs repetition. The learning phase will occur regardless, so ensuring it learns the right information is what actually matters. Without enough data points, the model can't find meaningful patterns. What you get instead is variance, inconsistency, and the creeping feeling that the algorithm is just doing whatever it wants.

If you're pulling fewer than 30 conversions a month, Target CPA isn't automation. It's an expensive coin flip dressed up in a dashboard. Sometimes the fix is expanding geography. Sometimes it's loosening match types temporarily to build signal density. Sometimes it's consolidating campaigns that are each too small to learn on their own.

Patience and Discipline

Smart Bidding has a learning phase and needs room to breathe. Changing targets mid-week or pulling the plug after a bad Tuesday is like benching your quarterback after one incomplete pass. 

What an Audit Actually Found

A new account came in with a clear team diagnosis: Smart Bidding is broken, CPA is rising, leads are garbage, and Google has failed us. A routine audit told a different story.

The audit found the usual suspects: every form submission counted as a conversion, brand and non-brand traffic lumped into the same campaigns, no offline revenue data, and volume too thin to support Target CPA. Smart Bidding was doing exactly what it was told; unfortunately, it started with unclear data and learned from it.

After rebuilding campaign structure, cleaning conversion tracking, and importing real revenue data, CPA dropped 28% in 30 days, and lead quality improved significantly. 

The Takeaway

Automation reveals your strategy more than it replaces it. The accounts where Smart Bidding actually works aren't the ones where someone handed the algorithm a vague objective and waited for results. They're the ones where someone did the unglamorous work first and built something worth automating: clean signal, clear structure, real data. The algorithm performed well because someone ensured it had solid data to work with.