Smart Bidding is Google's machine learning-powered bidding system, and it genuinely outperforms manual bidding once the algorithm has enough data to work with. But done wrong (or chosen too hastily), it can burn through your budget faster than any manual mistake you could possibly make.
The two heavyweights in the Smart Bidding arena are Target CPA and Target ROAS. They solve entirely different problems, and relying on the wrong one is a significant, yet highly avoidable, mistake.
The Core Difference Explained
Target CPA (Cost Per Acquisition) tells Google: "Get me as many conversions as humanly possible, just keep my average cost per conversion at or below $X."
Target ROAS (Return On Ad Spend) tells Google: "Get me as much revenue as possible, but make sure I get a return of $X for every $1 I spend."
This distinction is massive because:
- Target CPA treats every single conversion as equal in value.
- Target ROAS cares deeply about the revenue each specific conversion generates.
If you sell a single product at a fixed price (like a SaaS subscription or a flat-fee service), either strategy can work perfectly fine. But if you have an ecommerce catalog with products ranging from $10 to $1,000, Target ROAS is almost always the right choice. Why? Because it actively prioritizes high-value transactions over cheap, low-margin ones.
When You Should Use Target CPA
Target CPA shines in a few specific scenarios:
Your conversions have a highly consistent value. Lead generation is the classic example here. Whether a user fills out a form for a roofing quote or signs up for a free software trial, the immediate value isn't captured in the transaction itself. Asking Google to optimize for revenue doesn't make sense when every lead looks identical on paper.
You are early in your Smart Bidding journey. Target CPA generally has a slightly shorter learning phase and stabilizes a bit faster than Target ROAS when dealing with limited data.
Your conversion value data is a mess. If your tracking setup is broken and it isn't passing accurate revenue values back to Google consistently, Target ROAS is going to make terrible decisions based on garbage inputs. Fix your tracking first; rely on Target CPA in the interim.
When You Should Use Target ROAS
Target ROAS is the ultimate goal, but you have to earn the right to use it:
You sell products at wildly different price points. If your Shopify store processes orders ranging from $20 to $500, you absolutely need Target ROAS. Otherwise, the algorithm is equally thrilled to spend $15 acquiring a $20 sale as it is acquiring a $500 sale.
Your conversion value data is rock solid. This means your tracking is flawlessly passing dynamic value and currency parameters on every single conversion.
The Conversion Data Reality Check
According to official Google Ads documentation, the technical minimum to use Target ROAS is 15 conversions in the past 30 days.
However, as an expert managing these campaigns every day, I can confidently tell you that 15 conversions is a recipe for erratic behavior. To actually get the algorithm working beautifully for you, you really need 30 to 50 conversions in the last 30 days at the campaign level. Give the machine more data, and it will reward you with much better stability.
Setting Your ROAS Target (Don't Be Greedy)
A fatal mistake I see constantly is setting an aspirational ROAS target rather than a realistic one based on historical data.
Pull your actual ROAS from the last 60 to 90 days. If your historical average is 300%, set an initial target slightly below that, around 250% to 280%. This gives Google the breathing room it needs to spend while it learns. Once performance stabilizes, you can start inching the target up in 10% to 15% increments.
If you set a target of 500% when you've only ever averaged 300%, you are telling Google that the current volume of conversions isn't good enough. The algorithm will immediately choke your spend trying to find ultra-profitable transactions that simply don't exist in enough volume.
When You Should Use Neither
Despite what Google's automated recommendations might tell you, Smart Bidding isn't always the right move. Stick to Manual CPC or Maximize Clicks when:
- Your campaign is barely scraping by with fewer than 15 conversions a month (there simply isn't enough signal).
- You've just launched a brand new campaign and are entirely in the data-gathering phase.
- You just overhauled your conversion tracking, meaning all your historical data is now mismatched.
In these situations, a human touch will outsmart the algorithm. Wait until the volume is there before handing over the keys.
Surviving the Learning Period
Every time you switch your Smart Bidding strategy, or drastically change your ROAS/CPA targets, the campaign plunges into a "learning period" that typically lasts 1 to 2 weeks. Performance will usually dip during this window. Don't panic.
What to avoid during the learning phase:
- Changing the bid strategy again (this immediately resets the clock).
- Adjusting the target by more than 20% in a single day.
- Pausing your campaigns.
- Making massive changes to your ad copy or budget.
A Practical Decision Framework
Before you make a switch, ask yourself:
- Do I have 30 to 50 conversions per month? → Yes: consider ROAS. No: stick to CPA or manual.
- Are my conversion values flawless? → Yes: ROAS is viable. No: fix tracking first, use CPA for now.
- Are all my conversions worth the exact same amount? → Yes: CPA is fine. No: ROAS will capture significantly more value.
- Is my account brand new? → Start manual, collect data, then transition.
Getting your bidding strategy right is completely dependent on getting your conversion tracking right. If the data feeding into Smart Bidding is flawed, the output will be flawed too, and Google's algorithm will confidently optimize your account in the completely wrong direction.
If you are unsure whether your tracking is feeding Google the right signals, book a free tracking audit with us. We'll pop the hood and tell you exactly what Smart Bidding is actually seeing.