Smart Bidding in Google Ads. How It Works and When to Use It

Smart Bidding is Google’s suite of automated bid strategies that use machine learning to optimize for conversions or conversion value at auction time. Instead of you setting a static bid that applies to every search, Google adjusts your bid for each individual auction based on hundreds of contextual signals — device, location, time of day, browser, operating system, remarketing list membership, and more. The system processes all of that in milliseconds, every time someone triggers your ad. When it works well, Smart Bidding finds conversion opportunities a human couldn’t spot and adjusts faster than any manual process.

At Gorilla Marketing, we run Smart Bidding across campaigns for US businesses at every spend level. This guide explains how each strategy works, which one fits which goal, what the algorithm needs to perform, and the common mistakes that burn through budget without delivering results.

What Makes Smart Bidding Different From Other Automated Bidding?

All Smart Bidding strategies are automated, but not all automated bidding is Smart Bidding. Google Ads offers several automated strategies — Maximize Clicks and Target Impression Share, for example — that optimize for traffic or visibility rather than conversions. Those are automated bidding strategies, not Smart Bidding.

Smart Bidding is specifically the subset that optimizes for conversions or conversion value. Four strategies fall into this category:

Maximize Conversions — get the highest number of conversions within your budget

Maximize Conversion Value — get the highest total conversion value within your budget

Target CPA (cost per acquisition) — get as many conversions as possible at a specific cost target

Target ROAS (return on ad spend) — get as much conversion value as possible at a specific return target

The defining feature is auction-time bidding. Older automated strategies adjusted bids periodically throughout the day. Smart Bidding evaluates every single auction individually, factoring in the full context of that specific search. Two people searching the same keyword at the same time can trigger different bids because the algorithm assessed their likelihood to convert differently based on signals like location, device, past browsing behavior, and list membership.

How Does Auction-Time Bidding Actually Work?

Every time someone searches and your ad is eligible to enter the auction, Smart Bidding runs a prediction. It considers hundreds of signals simultaneously to estimate the probability that this particular user, in this particular context, will convert or generate value.

The signals Google has confirmed Smart Bidding uses include:

Device — mobile, desktop, or tablet

Physical location — down to city level, even if your location targeting is broader

Time of day and day of week

Remarketing list — which list the user belongs to and how recently they were added

Browser and operating system

Language settings

Search query — the actual terms typed, not just the keyword matched

Ad creative — which ad in the ad group is being considered

Based on that prediction, the algorithm sets a bid it calculates will meet your campaign goal. If the prediction says this user is highly likely to convert, the bid goes up. If the signals suggest low conversion probability, the bid drops. This happens for every auction, in real time.

The practical impact: Smart Bidding can push aggressively on high-value opportunities and pull back on low-value ones far more granularly than any manual or rules-based approach. But it only works that way when the system has enough data to make accurate predictions.

Maximize Conversions: When Volume Is the Goal

Maximize Conversions tells Google to get you as many conversions as possible within your daily budget. There’s no cost target. The algorithm will spend your full budget every day, bidding whatever it takes to capture each conversion opportunity it identifies.

When to use it

This strategy fits when you care about volume and your budget is the primary constraint. Lead generation campaigns with a fixed monthly spend, for example. Or accounts that need more conversion data before moving to a target-based strategy.

It’s also a solid starting point for new campaigns that don’t have enough historical data for Target CPA. Google’s own product team has pushed back on the conventional wisdom of “start with manual CPC and switch to Smart Bidding later.” Their position is that you can start with the strategy you want to optimize toward, and the system will learn as conversions come in.

What to watch for

Maximize Conversions will spend your entire budget. Always. If you set a $200 daily budget, it will find a way to spend $200. That means it may bid aggressively on lower-quality conversions toward the end of the day just to use remaining budget. Monitor your cost per conversion closely. If CPA starts climbing and conversion quality drops, it might be time to add a target.

Target CPA: Setting a Cost Ceiling

Target CPA is Maximize Conversions with a cost constraint. You tell Google your target cost per acquisition, and the algorithm tries to get as many conversions as possible at or near that number. Some conversions will cost more than your target, some will cost less. The goal is averaging out at your CPA target over time.

When to use it

Target CPA works when you know what a conversion is worth to your business and you need to control costs. If your average customer value supports a $75 CPA but not a $120 CPA, setting that target gives the algorithm a boundary to work within.

The strategy requires enough historical data to be effective. Google suggests a minimum of roughly 30 conversions in the last 30 days for the algorithm to optimize reliably. Fewer than that, and the predictions get noisy.

Setting realistic targets

This is where most advertisers get it wrong. They look at their best-performing month, grab that CPA number, and set it as their target. The algorithm then struggles because the target is based on an outlier, not a sustainable average.

Pull your average CPA over the last 60-90 days. Set your Target CPA at or slightly above that number. Once the algorithm stabilizes and the learning phase is complete, you can gradually tighten the target in small increments — 10-15% at a time. Dropping your target by 40% overnight is a fast way to choke your campaign’s reach. The algorithm will simply stop bidding on auctions it can’t win at that price.

Maximize Conversion Value: When Not All Conversions Are Equal

Maximize Conversion Value tells Google to optimize for the highest total conversion value within your budget. This strategy needs conversion values assigned to your conversions — transaction revenue for e-commerce, or assigned values for different lead types.

When to use it

E-commerce is the obvious fit. If you’re selling products at different price points, you want the algorithm to prioritize the $500 sale over the $15 sale, even if both count as one conversion. Maximize Conversion Value does that automatically.

It also works for lead generation businesses that use value-based bidding. If a demo request is worth $200 to your pipeline and a newsletter signup is worth $5, assigning those values lets the algorithm chase the high-value actions. This requires your CRM data to flow back into Google Ads through offline conversion imports or a platform integration.

What to watch for

Like Maximize Conversions, this strategy will spend your full daily budget. And it doesn’t care about return on spend unless you add a ROAS target. It’s entirely possible for the algorithm to chase a few high-value conversions at extreme CPCs, technically maximizing value while destroying your margins.

Target ROAS: Tying Bids to Return

Target ROAS is Maximize Conversion Value with a return constraint. You set a target return on ad spend — say, 400% — and the algorithm aims to generate $4 in conversion value for every $1 spent.

When to use it

Target ROAS makes sense for e-commerce accounts and any campaign where conversion values are tracked accurately. It’s the most ROI-focused Smart Bidding strategy, directly tying your bids to financial return.

The data bar is higher here. Google generally recommends around 50 conversions in the last 30 days for Target ROAS to work effectively. The algorithm needs enough conversion events with associated values to identify patterns in which auctions produce high-value outcomes and which don’t.

Setting the target

Same principle as Target CPA: start with your historical average. If your account has been running at a 350% ROAS, don’t set a target of 600% and expect the algorithm to figure it out. Start at 350%, let the system stabilize, then nudge the target up in modest increments.

Unrealistic ROAS targets produce the same result as unrealistic CPA targets — the algorithm restricts bidding, impressions drop, and you end up with less revenue than you started with. A target that’s 10-15% more ambitious than your current average is a reasonable first move.

Portfolio Bid Strategies: Sharing Intelligence Across Campaigns

Most advertisers apply Smart Bidding at the campaign level. But Google also supports portfolio bid strategies, which apply a single bidding strategy across multiple campaigns. The algorithm pools data from all included campaigns and optimizes toward a shared target.

Portfolio strategies make sense when individual campaigns don’t generate enough conversion volume on their own. A campaign with 12 conversions a month might not give the algorithm enough data, but four similar campaigns pooled together at 50+ combined conversions create a much stronger signal.

You can set a single Target CPA or Target ROAS across the portfolio, and the algorithm will distribute bids to hit that target across all included campaigns collectively. Some campaigns may run above target and others below, as long as the portfolio average hits the mark.

The trade-off: you lose visibility into per-campaign efficiency. If one campaign in the portfolio is consistently underperforming, the portfolio average might mask it. Review individual campaign performance regularly, even when using portfolio strategies.

The Learning Phase: Why Patience Matters

Every time you make a significant change to a Smart Bidding strategy — switching strategies, adjusting a target, changing the budget substantially, or modifying conversion actions — the algorithm enters a learning phase. During this period, performance will fluctuate. CPAs may spike. Volume may drop. The system is recalibrating its predictions.

The learning phase typically lasts one to three weeks, depending on your conversion volume. Campaigns with higher conversion volume exit the learning phase faster because the algorithm has more data points to recalibrate from.

The single most common mistake during this phase: panicking and making another change. Every change resets the learning period. If you switch to Target CPA, see CPAs spike in week one, panic and adjust the target, then pause a few ad groups for good measure, you’ve essentially told the algorithm to start learning from scratch three times.

Set your strategy. Set a reasonable target. Wait two to three weeks. Then evaluate. If you can’t stomach the volatility during the learning phase, Smart Bidding may require a mindset shift before it requires a strategy change.

Seasonality Adjustments and Data Exclusions

Smart Bidding learns from historical data, which means it can be slow to react to sudden shifts. Google provides two tools to help with this.

Seasonality adjustments let you tell the algorithm to expect a temporary change in conversion rates. Running a Black Friday sale and expecting conversion rates to double? Set a seasonality adjustment with the expected conversion rate lift and the date range. The algorithm will bid more aggressively during that window and revert afterward.

Data exclusions let you tell the algorithm to ignore a specific date range. If your conversion tracking broke for three days and reported zero conversions, the algorithm would normally read that as “nothing converts anymore” and drastically reduce bids. A data exclusion tells the system to discard that period and bid based on the data from before and after.

Both tools are available in the Shared Library under Bid Strategies. They’re underused, and they matter. A few days of bad data can take weeks to recover from if you don’t exclude it.

Common Mistakes That Waste Budget

Smart Bidding isn’t set-and-forget. Most performance issues trace back to a handful of recurring mistakes.

Broken or inaccurate conversion tracking

The algorithm optimizes toward whatever you’re tracking. If your conversion tag fires on page load instead of form submission, or counts every page view as a conversion, the system will optimize toward junk. Accurate conversion tracking is the foundation. Without it, Smart Bidding is optimizing toward the wrong signal.

Unrealistic targets

Setting a Target CPA that’s 50% below your historical average, or a Target ROAS that’s double what you’ve ever achieved, doesn’t push the algorithm to try harder. It pushes the algorithm to stop bidding. The system won’t enter auctions it calculates it can’t win at your target, and your campaign goes quiet.

Constant adjustments during the learning phase

Changing targets, budgets, or strategy every few days gives the algorithm no time to learn. Each change resets the clock. Commit to a setup, give it two to three weeks, then evaluate with enough data to make a real judgment.

Too many conversion actions

If your campaign is optimizing toward five different conversion actions — form fills, phone calls, newsletter signups, PDF downloads, and purchases — the algorithm has no clear signal. It’ll chase whichever conversion is easiest to get, which is usually the lowest-value one. Streamline your primary conversion actions to the ones that actually matter to revenue.

Ignoring search term reports

Smart Bidding controls bids, not targeting. If your keywords are pulling irrelevant search queries, the algorithm will bid on them and try to squeeze conversions from bad traffic. Negative keyword management is still your job. Review search terms regularly and exclude what doesn’t belong.

Should You Use Smart Bidding?

For most US advertisers running Google Ads with reasonable conversion volume, Smart Bidding outperforms manual bidding over time. The system processes more signals, reacts faster, and scales in ways that manual bid management can’t match. That’s not opinion; it’s the structural advantage of auction-time machine learning over a human refreshing a spreadsheet.

But “over time” is doing real work in that sentence. Smart Bidding needs accurate data, realistic targets, and patience through the learning phase. Accounts with fewer than 15-20 conversions per month, broken tracking, or wildly fluctuating conversion values will struggle with any automated strategy. Fix the fundamentals first.

The best approach for most accounts: start with Maximize Conversions to build data and validate that your tracking is accurate. Once you have a reliable baseline CPA over 30-60 days, add a Target CPA constraint. For e-commerce or value-based accounts, follow the same path with Maximize Conversion Value into Target ROAS. Each step gives the algorithm more signal and gives you more control.

At Gorilla Marketing, our PPC team doesn’t default to Smart Bidding because Google recommends it. We match bidding strategies to what the data supports, set targets based on historical performance rather than wishful thinking, and monitor the learning phase instead of leaving it on autopilot. If your Google Ads campaigns aren’t delivering the return you need, that’s worth a conversation.